Spaces:
Runtime error
Runtime error
File size: 226,090 Bytes
db5855f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 |
{
"cells": [
{
"cell_type": "markdown",
"id": "66dd36b3-e847-4a51-8d5b-db21505645d3",
"metadata": {},
"source": [
"# SPEECH EMOTION DETECTION\n",
"Speech Emotion Detection (SED) is the task of automatically recognizing and classifying emotions conveyed through speech signals. It involves the use of techniques from signal processing, machine learning, and natural language processing (NLP) to analyze vocal cues and identify the emotional state of a speaker.\r\n",
"\r\n",
"### Key Aspects of Speech Emotion Detection:\r\n",
"\r\n",
"- **Signal Processing**: Techniques such as feature extraction from audio signals (e.g., MFCC - Mel Frequency Cepstral Coefficients) to capture relevant characteristics related to pitch, intensity, and spectral content that correlate with emotions.\r\n",
"\r\n",
"- **Feature Engineering**: Selection and transformation of acoustic features that are relevant for distinguishing different emotional states.\r\n",
"\r\n",
"- **Machine Learning and Deep Learning**: Utilization of various models and algorithms, including traditional machine learning methods (e.g., Support Vector Machines, Random Forests) and deep learning architectures (e.g., Convolutional Neural Networks, Recurrent Neural Networks like LSTM), to classify emotions based on extracted features.\r\n",
"\r\n",
"- **Emotion Classification**: Assigning one or multiple predefined emotion labels (e.g., happiness, sadness, anger, neutrality) to audio segments based on the analysis of extracted features and model pr\n",
"future directions.\r\n",
"edictions.\r\n"
]
},
{
"cell_type": "markdown",
"id": "86a9b89c-5976-4e93-96e6-5258ef696328",
"metadata": {},
"source": [
"### RAVDESS Dataset Description\r\n",
"\r\n",
"The RAVDESS dataset contains 1440 audio files, structured as follows: 60 trials per actor multiplied by 24 actors, resulting in 1440 files. The dataset comprises recordings from 24 professional actors (12 female and 12 male), each vocalizing two lexically-matched statements in a neutral North American accent. The speech emotions included are calm, happy, sad, angry, fearful, surprise, and disgust, with each emotion expressed at two levels of emotional intensity (normal and strong), along with an additional neutral expression.\r\n",
"\r\n",
"#### File Naming Convention\r\n",
"\r\n",
"Each file in the dataset has a unique filename format: `03-01-06-01-02-01-12.wav`. Here's the breakdown of the filename identifiers:\r\n",
"\r\n",
"- **Modality**: `01` for full-AV (audiovisual), `02` for video-only, `03` for audio-only.\r\n",
"- **Vocal Channel**: `01` for speech, `02` for song.\r\n",
"- **Emotion**: \r\n",
" - `01` = neutral\r\n",
" - `02` = calm\r\n",
" - `03` = happy\r\n",
" - `04` = sad\r\n",
" - `05` = angry\r\n",
" - `06` = fearful\r\n",
" - `07` = disgust\r\n",
" - `08` = surprised\r\n",
"- **Emotional Intensity**: `01` for normal, `02` for strong (Note: 'neutral' emotion has no strong intensity).\r\n",
"- **Statement**: \r\n",
" - `01` = \"Kids are talking by the door\"\r\n",
" - `02` = \"Dogs are sitting by the door\"\r\n",
"- **Repetition**: `01` for 1st repetition, `02` for 2nd repetition.\r\n",
"- **Actor**: `01` to `24`. Odd-numbered actors are male, and even-numbered actors are female.\r\n",
"\r\n",
"#### Filename Example: `03-01-06-01-02-01-12.wav`\r\n",
"\r\n",
"- **Audio-only (03)**\r\n",
"- **Speech (01)**\r\n",
"- **Fearful (06)**\r\n",
"- **Normal intensity (01)**\r\n",
"- **Statement \"dogs\" (02)**\r\n",
"- **1st Repetition (01)**\r\n",
"- **12th Actor (12)** (Female, as the actor ID number is even)\r\n"
]
},
{
"cell_type": "code",
"execution_count": 92,
"id": "6ce0fa5b-a908-4486-80c7-c07e46d98354",
"metadata": {},
"outputs": [],
"source": [
"# Define the path to your RAVDESS dataset\n",
"DATASET_PATH = r'C:\\games\\audio'\n"
]
},
{
"cell_type": "markdown",
"id": "4a0f0509-02a2-44f9-8250-51518278af97",
"metadata": {},
"source": [
"## DATA PREPROCESSING\r\n",
"ing.\r\n",
"\r\n",
"### Function to Load an Audio File\r\n",
"\r\n",
"The `load_audio` function loads an audio file from the specified `file_path` and returns the audio waveform and its sample rate (`sr`). It utilizes the `librosa` library for audio loading and handles exceptions if the file loading fails.\r\n",
"\r\n",
"### Function to Extract MFCC Features\r\n",
"\r\n",
"The `extract_mfcc` function extracts Mel-Frequency Cepstral Coefficients (MFCC) features from the audio data. MFCCs are commonly used in speech and audio processing to capture the spectral characteristics of sound. This function computes the MFCCs and returns their mean values across time (`axis=0`).\r\n",
"\r\n",
"### Dummy Function to Extract Label from Filename\r\n",
"\r\n",
"The `extract_label_from_filename` function extracts the emotional label from a given filename based on predefined mappings (`emotion_dict`). It parses the filename and retrieves the emotion label using the third part of the filename (which corresponds to the emotion identifier).\r\n",
"\r\n",
"#### Emotion Dictionary:\r\n",
"- `'01': 'neutral'`\r\n",
"- `'02': 'calm'`\r\n",
"- `'03': 'happy'`\r\n",
"- `'04': 'sad'`\r\n",
"- `'05': 'angry'`\r\n",
"- `'06': 'fearful'`\r\n",
"- `'07': 'disgust'`\r\n",
"- `'08': 'surprised'`\r\n",
"\r\n",
"This function is used to map numeric identifiers from filenames to human-readable emotional labels, facilitating the labeling of audio data during the processing pipeline.\r\n",
"\r\n",
"g the processing pipeline.\r\n",
"\r\n",
"coder`.\r\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "6810c70c-3f17-4f3a-8789-8bd848ba0cb4",
"metadata": {},
"outputs": [],
"source": [
"# Define the path to your RAVDESS dataset\n",
"\n",
"\n",
"# Function to load an audio file\n",
"def load_audio(file_path, sr=22050):\n",
" try:\n",
" audio, sample_rate = librosa.load(file_path, sr=sr)\n",
" return audio, sample_rate\n",
" except Exception as e:\n",
" print(f\"Error loading {file_path}: {e}\")\n",
" return None, None\n",
"\n",
"# Function to extract MFCC features\n",
"def extract_mfcc(audio, sr, n_mfcc=40):\n",
" mfccs = np.mean(librosa.feature.mfcc(y=audio, sr=sr, n_mfcc=n_mfcc).T, axis=0)\n",
" return mfccs\n",
"\n",
"# Dummy function to extract label from filename\n",
"def extract_label_from_filename(file_name):\n",
" emotion_dict = {\n",
" '01': 'neutral', '02': 'calm', '03': 'happy', '04': 'sad',\n",
" '05': 'angry', '06': 'fearful', '07': 'disgust', '08': 'surprised'\n",
" }\n",
" parts = file_name.split('-')\n",
" emotion_label = emotion_dict[parts[2]]\n",
" return emotion_label\n"
]
},
{
"cell_type": "markdown",
"id": "46240c96-c06b-4169-841b-19b0769faa72",
"metadata": {},
"source": [
"## Loading and Processing Audio Dataset\n",
"### Iterating Through Audio Files\n",
"The code iterates through all `.wav` files located in the `DATASET_PATH` directory and its subdirectories using `os.walk`.\n",
"### Extracting Features and Labels\n",
"For each audio file:\n",
"- **Label Extraction**: The label is extracted from the filename (`filename[7:8] - 1`). This assumes the label information is encoded in a specific position within the filename. \n",
"- **Loading Audio**: The `load_audio` function is called to load the audio file and obtain the audio waveform (`audio`) and its sampling rate (`sr`).\n",
"- **Extracting MFCC Features**: If the audio loading is successful (`audio` is not `None`), the `extract_mfcc` function extracts Mel-Frequency Cepstral Coefficients (MFCC) features from the audio.\n",
"- **Appending Features and Labels**: Extracted MFCC features (`mfcc_features`) and their corresponding labels (`label`) are appended to lists `X` and `y`, respectively.\n",
"\n",
"### Handling Errors\n",
"\n",
"- **Exception Handling**: If there is an error during any step (e.g., loading audio, extracting features), an error message is printed, and the file is skipped (`except Exception as e`).\n",
"\n",
"### Final Dataset Preparation\n",
"\n",
"- **Conversion to Numpy Arrays**: After processing all audio files, lists `X` and `y` are converted into NumPy arrays (`np.array(X)` and `np.array(y)`).\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 37,
"id": "358a2d96-be6c-4cec-9140-a453d665180c",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Finish Loading the Dataset\n",
"X shape: (2880, 40)\n",
"y shape: (2880,)\n"
]
}
],
"source": [
"X=[]\n",
"y=[]\n",
"for dirname, _, filenames in os.walk(DATASET_PATH):\n",
" for filename in filenames:\n",
" if filename.endswith('.wav'):\n",
" try:\n",
" label = int(filename[7:8]) - 1 # Extract label from filename\n",
" wav_file_path = os.path.join(dirname, filename)\n",
" audio, sr = load_audio(wav_file_path) # Load audio file\n",
" if audio is not None:\n",
" mfcc_features = extract_mfcc(audio, sr) # Extract MFCC features\n",
" if mfcc_features is not None:\n",
" X.append(mfcc_features)\n",
" y.append(label)\n",
" except Exception as e:\n",
" print(f\"Error processing {filename}: {e}\")\n",
"\n",
"X = np.array(X)\n",
"y = np.array(y)\n",
"\n",
"print(\"Finish Loading the Dataset\")\n",
"print(\"X shape:\", X.shape)\n",
"print(\"y shape:\", y.shape)"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "3262162d-2945-4c0a-8c22-533ad367ccf0",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([0, 0, 0, ..., 7, 7, 7])"
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"y"
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "3730f074-a59d-476d-88ca-1ab2695e38b2",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[-6.9779260e+02, 5.4890041e+01, 6.6346520e-01, ...,\n",
" -1.7701062e+00, -3.5011320e+00, -1.5380874e+00],\n",
" [-6.9285577e+02, 5.5363899e+01, -1.5483192e+00, ...,\n",
" -3.0869722e+00, -3.2097483e+00, -2.3071594e+00],\n",
" [-6.9158789e+02, 5.8024662e+01, 1.5946463e-01, ...,\n",
" -2.7295372e+00, -3.5089064e+00, -2.2463746e+00],\n",
" ...,\n",
" [-5.4125165e+02, 2.9595709e+01, -1.8001362e+01, ...,\n",
" -3.8906682e-01, -9.7194743e-01, -3.9210208e-03],\n",
" [-4.9265253e+02, 2.3887981e+01, -6.0266590e+00, ...,\n",
" 9.1637713e-01, -1.4326899e+00, 1.4300117e+00],\n",
" [-5.1798480e+02, 2.9571215e+01, -3.9096990e+00, ...,\n",
" -1.2455600e-01, -1.8519229e+00, 7.0855033e-01]], dtype=float32)"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"X"
]
},
{
"cell_type": "code",
"execution_count": 96,
"id": "e345782b-935d-46d3-9a52-88cdcb3ed44c",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[0 0 0 ... 7 7 7]\n",
"[[1. 0. 0. ... 0. 0. 0.]\n",
" [1. 0. 0. ... 0. 0. 0.]\n",
" [1. 0. 0. ... 0. 0. 0.]\n",
" ...\n",
" [0. 0. 0. ... 0. 0. 1.]\n",
" [0. 0. 0. ... 0. 0. 1.]\n",
" [0. 0. 0. ... 0. 0. 1.]]\n"
]
}
],
"source": [
"### Label Encoding\n",
"label_encoder = LabelEncoder()\n",
"y_encoded = label_encoder.fit_transform(y)\n",
"print(y_encoded)\n",
"##One_hot encoding\n",
"y_cat=to_categorical (y_encoded)\n",
"print(y_cat)"
]
},
{
"cell_type": "markdown",
"id": "c09420cf-a9a1-464e-969b-ebfcc3d828ae",
"metadata": {},
"source": [
"## Train-Test Split"
]
},
{
"cell_type": "code",
"execution_count": 40,
"id": "d25650c5-5681-4a24-ab42-380bb19f82a8",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Training set size: 2304 samples\n",
"Testing set size: 576 samples\n",
"X_train shape: (2304, 40)\n",
"X_test shape: (576, 40)\n"
]
}
],
"source": [
"\n",
"X_train, X_test, y_train, y_test = train_test_split(X, y_cat, test_size=0.2, random_state=9)\n",
"\n",
"# Print shapes to verify\n",
"print(f\"Training set size: {X_train.shape[0]} samples\")\n",
"print(f\"Testing set size: {X_test.shape[0]} samples\")\n",
"print(f\"X_train shape: {X_train.shape}\")\n",
"print(f\"X_test shape: {X_test.shape}\")\n"
]
},
{
"cell_type": "code",
"execution_count": 13,
"id": "498fc255-2b1c-4ff8-afdf-7f5b8998dd16",
"metadata": {},
"outputs": [],
"source": [
"X_train = X_train[..., np.newaxis]\n",
"X_test=X_test[...,np.newaxis]"
]
},
{
"cell_type": "markdown",
"id": "69089f82-0783-441b-9800-01dbb591fa75",
"metadata": {},
"source": [
"# Conv1D Model for Emotion Recognition\n",
"\n",
"### Model Architecture\n",
"\n",
"The Conv1D model is designed to classify emotions based on audio features extracted using Mel-Frequency Cepstral Coefficients (MFCCs). The architecture consists of:\n",
"\n",
"- **Conv1D Layer**: The first layer has 64 filters with a kernel size of 3 and uses the ReLU activation function. It takes input with a shape corresponding to the MFCC feature dimensions.\n",
"- **MaxPooling1D Layer**: This layer with a pool size of 2 reduces the dimensionality of the feature maps.\n",
"- **Dropout Layer**: A dropout rate of 0.3 is used to prevent overfitting.\n",
"- **Conv1D Layer**: The second Conv1D layer has 128 filters with a kernel size of 3 and uses the ReLU activation function.\n",
"- **MaxPooling1D Layer**: Another pooling layer with a pool size of 2.\n",
"- **Dropout Layer**: Another dropout layer with a rate of 0.3.\n",
"- **Flatten Layer**: Flattens the feature maps into a single vector.\n",
"- **Dense Layer**: A dense layer with 128 units and ReLU activation.\n",
"- **Dropout Layer**: Dropout with a rate of 0.3.\n",
"- **Output Layer**: A dense layer with a number of units equal to the number of emotion classes, using the softmax activation function to output probability distributions for each class.\n",
"\n",
"### Model Compilation\n",
"\n",
"The model is compiled using the Adam optimizer and categorical cross-entropy loss function, which is suitable for multi-class classification problems. The performance metric used is accuracy."
]
},
{
"cell_type": "code",
"execution_count": 72,
"id": "04b67f11-b07f-42f9-9f23-73eb9b66ddfd",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"C:\\Users\\KIIT\\anaconda3\\Lib\\site-packages\\keras\\src\\layers\\convolutional\\base_conv.py:107: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n",
" super().__init__(activity_regularizer=activity_regularizer, **kwargs)\n"
]
}
],
"source": [
"\n",
"model = Sequential([\n",
" Conv1D(64, kernel_size=3, activation='relu', input_shape=(X.shape[1], 1)),\n",
" MaxPooling1D(pool_size=2),\n",
" Dropout(0.3),\n",
" Conv1D(128, kernel_size=3, activation='relu'),\n",
" MaxPooling1D(pool_size=2),\n",
" Dropout(0.3),\n",
" Flatten(),\n",
" Dense(128, activation='relu'),\n",
" Dropout(0.3),\n",
" Dense(len(label_encoder.classes_), activation='softmax')\n",
"])\n",
"\n",
"# Compile the model\n",
"model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])\n",
"\n",
"\n"
]
},
{
"cell_type": "markdown",
"id": "3d38d1f3-80a9-4343-af19-afcd416c0017",
"metadata": {},
"source": [
"### Adding Channel Dimension for Conv1D\r\n",
"\r\n",
"In Conv1D neural networks, adding an additional channel dimension (`np.newaxis`) to input data like MFCC features (`X_train` and `X_test`) is essential. This adjustment ensures compatibility with Conv1D layers, which expect input shapes in the form `(batch_size, sequence_length, num_channels)`. The channel dimension allows Conv1D to effectively apply filters across the temporal dimension of the data, facilitating feature extraction and learning temporal patterns.\r\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "19484b99-108b-4a2f-8606-85eb8effeb41",
"metadata": {},
"outputs": [],
"source": [
"X_train = X_train[..., np.newaxis]\n",
"X_test=X_test[...,np.newaxis]"
]
},
{
"cell_type": "markdown",
"id": "9d45be4e-fb81-4331-b40b-357490e9c6c5",
"metadata": {},
"source": [
"### Training the Model\n",
"\n",
"The model is trained using the training data (`X_train` and `y_train`). The training process involves:\n",
"\n",
"- **Epochs**: The model is trained for 100 epochs.\n",
"- **Batch Size**: A batch size of 32 is used for each training step.\n",
"- **Validation Data**: The model's performance is validated on the validation set (`X_test` and `y_test`) during training to monitor and evaluate its accuracy and loss.\n"
]
},
{
"cell_type": "code",
"execution_count": 74,
"id": "aee62dc0-cf5b-47b6-ab28-029e63ae9f21",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 8ms/step - accuracy: 0.1157 - loss: 7.4988 - val_accuracy: 0.1441 - val_loss: 2.0782\n",
"Epoch 2/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.1531 - loss: 2.1160 - val_accuracy: 0.1441 - val_loss: 2.0756\n",
"Epoch 3/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.1892 - loss: 2.0787 - val_accuracy: 0.1181 - val_loss: 2.0707\n",
"Epoch 4/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.1655 - loss: 2.0423 - val_accuracy: 0.1910 - val_loss: 1.9951\n",
"Epoch 5/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.1787 - loss: 2.0074 - val_accuracy: 0.1823 - val_loss: 1.9416\n",
"Epoch 6/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.1839 - loss: 1.9711 - val_accuracy: 0.1979 - val_loss: 1.9443\n",
"Epoch 7/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 7ms/step - accuracy: 0.1942 - loss: 1.9733 - val_accuracy: 0.2153 - val_loss: 1.9185\n",
"Epoch 8/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.1962 - loss: 1.9873 - val_accuracy: 0.2378 - val_loss: 1.8989\n",
"Epoch 9/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.2008 - loss: 1.9777 - val_accuracy: 0.2674 - val_loss: 1.8965\n",
"Epoch 10/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.2237 - loss: 1.9245 - val_accuracy: 0.3125 - val_loss: 1.8432\n",
"Epoch 11/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.2363 - loss: 1.9118 - val_accuracy: 0.3351 - val_loss: 1.8177\n",
"Epoch 12/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.2540 - loss: 1.8811 - val_accuracy: 0.3299 - val_loss: 1.8050\n",
"Epoch 13/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.2626 - loss: 1.8602 - val_accuracy: 0.3229 - val_loss: 1.7773\n",
"Epoch 14/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.2709 - loss: 1.8392 - val_accuracy: 0.3542 - val_loss: 1.7595\n",
"Epoch 15/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.2832 - loss: 1.8347 - val_accuracy: 0.3837 - val_loss: 1.6992\n",
"Epoch 16/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.3186 - loss: 1.7459 - val_accuracy: 0.3976 - val_loss: 1.6929\n",
"Epoch 17/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.3391 - loss: 1.7321 - val_accuracy: 0.4115 - val_loss: 1.6606\n",
"Epoch 18/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.3241 - loss: 1.7384 - val_accuracy: 0.3872 - val_loss: 1.6148\n",
"Epoch 19/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.3558 - loss: 1.6897 - val_accuracy: 0.4358 - val_loss: 1.5433\n",
"Epoch 20/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.3464 - loss: 1.6552 - val_accuracy: 0.4358 - val_loss: 1.5463\n",
"Epoch 21/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.3539 - loss: 1.6739 - val_accuracy: 0.4375 - val_loss: 1.5155\n",
"Epoch 22/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.3673 - loss: 1.6462 - val_accuracy: 0.4688 - val_loss: 1.5297\n",
"Epoch 23/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.3828 - loss: 1.6084 - val_accuracy: 0.4809 - val_loss: 1.4706\n",
"Epoch 24/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.4081 - loss: 1.5665 - val_accuracy: 0.4670 - val_loss: 1.4687\n",
"Epoch 25/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.3772 - loss: 1.5951 - val_accuracy: 0.4809 - val_loss: 1.4169\n",
"Epoch 26/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.4141 - loss: 1.5487 - val_accuracy: 0.4844 - val_loss: 1.3917\n",
"Epoch 27/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.4212 - loss: 1.5296 - val_accuracy: 0.5069 - val_loss: 1.3467\n",
"Epoch 28/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.4191 - loss: 1.4857 - val_accuracy: 0.5347 - val_loss: 1.3098\n",
"Epoch 29/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.4203 - loss: 1.4929 - val_accuracy: 0.5278 - val_loss: 1.2994\n",
"Epoch 30/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.4440 - loss: 1.4514 - val_accuracy: 0.5451 - val_loss: 1.2825\n",
"Epoch 31/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.4556 - loss: 1.4015 - val_accuracy: 0.5642 - val_loss: 1.2389\n",
"Epoch 32/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.4843 - loss: 1.3758 - val_accuracy: 0.5503 - val_loss: 1.2232\n",
"Epoch 33/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 7ms/step - accuracy: 0.4633 - loss: 1.3930 - val_accuracy: 0.5590 - val_loss: 1.2229\n",
"Epoch 34/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.4870 - loss: 1.3590 - val_accuracy: 0.5764 - val_loss: 1.2046\n",
"Epoch 35/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.4817 - loss: 1.3551 - val_accuracy: 0.5938 - val_loss: 1.1620\n",
"Epoch 36/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5026 - loss: 1.2994 - val_accuracy: 0.6007 - val_loss: 1.1363\n",
"Epoch 37/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.5011 - loss: 1.3065 - val_accuracy: 0.6181 - val_loss: 1.1233\n",
"Epoch 38/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5183 - loss: 1.2644 - val_accuracy: 0.5990 - val_loss: 1.0996\n",
"Epoch 39/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5373 - loss: 1.2638 - val_accuracy: 0.6372 - val_loss: 1.0709\n",
"Epoch 40/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5638 - loss: 1.1969 - val_accuracy: 0.6493 - val_loss: 1.0466\n",
"Epoch 41/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5522 - loss: 1.1764 - val_accuracy: 0.6510 - val_loss: 1.0343\n",
"Epoch 42/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5634 - loss: 1.1854 - val_accuracy: 0.6458 - val_loss: 0.9844\n",
"Epoch 43/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5634 - loss: 1.1475 - val_accuracy: 0.6875 - val_loss: 0.9832\n",
"Epoch 44/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5888 - loss: 1.0973 - val_accuracy: 0.6771 - val_loss: 0.9465\n",
"Epoch 45/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5933 - loss: 1.1037 - val_accuracy: 0.6944 - val_loss: 0.8992\n",
"Epoch 46/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5968 - loss: 1.0746 - val_accuracy: 0.6840 - val_loss: 0.9157\n",
"Epoch 47/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.6292 - loss: 1.0097 - val_accuracy: 0.7222 - val_loss: 0.8362\n",
"Epoch 48/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.6108 - loss: 1.0316 - val_accuracy: 0.7344 - val_loss: 0.8298\n",
"Epoch 49/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.6361 - loss: 0.9816 - val_accuracy: 0.7483 - val_loss: 0.8078\n",
"Epoch 50/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.6402 - loss: 0.9807 - val_accuracy: 0.7326 - val_loss: 0.7480\n",
"Epoch 51/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.6390 - loss: 0.9621 - val_accuracy: 0.7691 - val_loss: 0.7645\n",
"Epoch 52/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6421 - loss: 0.9360 - val_accuracy: 0.7361 - val_loss: 0.7458\n",
"Epoch 53/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.6702 - loss: 0.8742 - val_accuracy: 0.7778 - val_loss: 0.7248\n",
"Epoch 54/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6831 - loss: 0.8708 - val_accuracy: 0.7743 - val_loss: 0.7021\n",
"Epoch 55/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.6895 - loss: 0.8408 - val_accuracy: 0.7691 - val_loss: 0.6989\n",
"Epoch 56/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.6986 - loss: 0.8491 - val_accuracy: 0.7483 - val_loss: 0.7095\n",
"Epoch 57/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.6966 - loss: 0.7870 - val_accuracy: 0.7986 - val_loss: 0.6644\n",
"Epoch 58/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.7010 - loss: 0.7923 - val_accuracy: 0.7830 - val_loss: 0.6632\n",
"Epoch 59/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.6859 - loss: 0.8124 - val_accuracy: 0.7847 - val_loss: 0.6348\n",
"Epoch 60/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7143 - loss: 0.7737 - val_accuracy: 0.8003 - val_loss: 0.6152\n",
"Epoch 61/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.7154 - loss: 0.7969 - val_accuracy: 0.7934 - val_loss: 0.6460\n",
"Epoch 62/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7492 - loss: 0.7217 - val_accuracy: 0.7847 - val_loss: 0.6135\n",
"Epoch 63/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.7147 - loss: 0.7626 - val_accuracy: 0.8194 - val_loss: 0.5998\n",
"Epoch 64/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.7455 - loss: 0.6917 - val_accuracy: 0.8229 - val_loss: 0.5715\n",
"Epoch 65/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7447 - loss: 0.7022 - val_accuracy: 0.8316 - val_loss: 0.5726\n",
"Epoch 66/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7659 - loss: 0.6705 - val_accuracy: 0.8229 - val_loss: 0.5652\n",
"Epoch 67/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7592 - loss: 0.6647 - val_accuracy: 0.8281 - val_loss: 0.5232\n",
"Epoch 68/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7637 - loss: 0.6515 - val_accuracy: 0.8316 - val_loss: 0.5361\n",
"Epoch 69/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7422 - loss: 0.6811 - val_accuracy: 0.8351 - val_loss: 0.5108\n",
"Epoch 70/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.7770 - loss: 0.6319 - val_accuracy: 0.8403 - val_loss: 0.4960\n",
"Epoch 71/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.7674 - loss: 0.6403 - val_accuracy: 0.8455 - val_loss: 0.5008\n",
"Epoch 72/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7764 - loss: 0.6190 - val_accuracy: 0.8559 - val_loss: 0.4857\n",
"Epoch 73/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7797 - loss: 0.6008 - val_accuracy: 0.8385 - val_loss: 0.4882\n",
"Epoch 74/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7803 - loss: 0.5882 - val_accuracy: 0.8507 - val_loss: 0.4673\n",
"Epoch 75/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7957 - loss: 0.5663 - val_accuracy: 0.8472 - val_loss: 0.4710\n",
"Epoch 76/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7779 - loss: 0.6297 - val_accuracy: 0.8507 - val_loss: 0.4922\n",
"Epoch 77/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.7996 - loss: 0.5509 - val_accuracy: 0.8681 - val_loss: 0.4591\n",
"Epoch 78/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.7974 - loss: 0.5581 - val_accuracy: 0.8646 - val_loss: 0.4398\n",
"Epoch 79/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 8ms/step - accuracy: 0.7955 - loss: 0.5555 - val_accuracy: 0.8663 - val_loss: 0.4259\n",
"Epoch 80/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.8027 - loss: 0.5486 - val_accuracy: 0.8611 - val_loss: 0.4192\n",
"Epoch 81/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.8128 - loss: 0.5246 - val_accuracy: 0.8750 - val_loss: 0.4238\n",
"Epoch 82/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8240 - loss: 0.4724 - val_accuracy: 0.8872 - val_loss: 0.4158\n",
"Epoch 83/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8174 - loss: 0.5094 - val_accuracy: 0.8854 - val_loss: 0.3948\n",
"Epoch 84/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.8240 - loss: 0.4929 - val_accuracy: 0.8958 - val_loss: 0.3806\n",
"Epoch 85/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.8212 - loss: 0.4963 - val_accuracy: 0.9028 - val_loss: 0.3706\n",
"Epoch 86/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8229 - loss: 0.4730 - val_accuracy: 0.8872 - val_loss: 0.4016\n",
"Epoch 87/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8183 - loss: 0.4667 - val_accuracy: 0.8906 - val_loss: 0.3899\n",
"Epoch 88/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8255 - loss: 0.4798 - val_accuracy: 0.8854 - val_loss: 0.3904\n",
"Epoch 89/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8397 - loss: 0.4399 - val_accuracy: 0.8941 - val_loss: 0.3561\n",
"Epoch 90/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8457 - loss: 0.4585 - val_accuracy: 0.9010 - val_loss: 0.3516\n",
"Epoch 91/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.8544 - loss: 0.4099 - val_accuracy: 0.8976 - val_loss: 0.3647\n",
"Epoch 92/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.8394 - loss: 0.4442 - val_accuracy: 0.9010 - val_loss: 0.3880\n",
"Epoch 93/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8546 - loss: 0.4011 - val_accuracy: 0.9062 - val_loss: 0.3315\n",
"Epoch 94/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.8570 - loss: 0.4066 - val_accuracy: 0.8837 - val_loss: 0.3517\n",
"Epoch 95/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.8570 - loss: 0.4290 - val_accuracy: 0.8993 - val_loss: 0.3558\n",
"Epoch 96/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8609 - loss: 0.3896 - val_accuracy: 0.8906 - val_loss: 0.3549\n",
"Epoch 97/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.8593 - loss: 0.4013 - val_accuracy: 0.9010 - val_loss: 0.3377\n",
"Epoch 98/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8638 - loss: 0.3873 - val_accuracy: 0.9115 - val_loss: 0.3193\n",
"Epoch 99/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.8615 - loss: 0.3948 - val_accuracy: 0.9010 - val_loss: 0.3076\n",
"Epoch 100/100\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8638 - loss: 0.3963 - val_accuracy: 0.9010 - val_loss: 0.3422\n"
]
}
],
"source": [
"\n",
"history = model.fit(X_train, y_train, epochs=100, batch_size=32, validation_data=(X_test, y_test))"
]
},
{
"cell_type": "markdown",
"id": "669a85fe-8a7b-438c-8524-ae3ccec2be1f",
"metadata": {},
"source": [
"### Model Evaluation\n",
"\n",
"After training, the model is evaluated on the test set to determine its final accuracy. The test accuracy provides an indication of how well the model generalizes to unseen data."
]
},
{
"cell_type": "code",
"execution_count": 75,
"id": "6141387a-448a-4ba5-a8c3-0957658002b3",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[1m18/18\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.9008 - loss: 0.3352\n",
"Test accuracy: 0.9010416865348816\n",
"\u001b[1m18/18\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step\n",
"True label: 5, Predicted label: 5\n",
"True label: 2, Predicted label: 2\n",
"True label: 3, Predicted label: 3\n",
"True label: 1, Predicted label: 1\n",
"True label: 1, Predicted label: 1\n",
"True label: 1, Predicted label: 1\n",
"True label: 7, Predicted label: 7\n",
"True label: 0, Predicted label: 0\n",
"True label: 3, Predicted label: 1\n",
"True label: 2, Predicted label: 2\n"
]
}
],
"source": [
"\n",
"test_loss, test_acc = model.evaluate(X_test, y_test)\n",
"print(f\"Test accuracy: {test_acc}\")\n",
"predictions = model.predict(X_test)\n",
"predicted_labels = np.argmax(predictions, axis=1)\n",
"true_labels = np.argmax(y_test, axis=1)\n",
"\n",
"# Print some predictions\n",
"for i in range(10):\n",
" print(f\"True label: {true_labels[i]}, Predicted label: {predicted_labels[i]}\")\n"
]
},
{
"cell_type": "markdown",
"id": "8c2536e4-e80a-4168-9cd4-6da3ff687b49",
"metadata": {},
"source": [
"\n",
"# Model Training and Test Accuracy\r\n",
"\r\n",
"The following plot shows the accuracy of the Conv1D model over 100 epochs for both the training and test datasets.ndt)\r\n",
"\r\n",
"### Description\r\n",
"\r\n",
"- **X-axis (Epoch)**: Represents the number of training epochs. Each epoch is one complete pass through the training dataset.\r\n",
"- **Y-axis (Accuracy)**: Represents the accuracy of the model. Accuracy is the proportion of correctly predicted instances out of the total instances.\r\n",
"\r\n",
"### Observations\r\n",
"\r\n",
"- **Training Accuracy (Blue Line)**: The training accuracy shows a steady increase as the number of epochs increases. This indicates that the model is learning and improving its performance on the training data.\r\n",
"- **Test Accuracy (Orange Line)**: The test accuracy also shows an increase and follows a similar trend to the training accuracy. This indicates that the model generalizes well to unseen data.\r\n",
"- **Model Performance**: The model achieves over 90% accuracy on both the training and test datasets after 100 epochs, demonstrating its effectiveness in classifying emotions from auficant overfitting.\r\n"
]
},
{
"cell_type": "code",
"execution_count": 60,
"id": "2d7647a2-cc62-4c55-9f04-8c1705c53f66",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"<matplotlib.legend.Legend at 0x1e6d7d53a10>"
]
},
"execution_count": 60,
"metadata": {},
"output_type": "execute_result"
},
{
"data": {
"image/png": "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",
"text/plain": [
"<Figure size 1200x400 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plt.figure(figsize=(12, 4))\n",
"plt.subplot(1, 2, 1)\n",
"plt.plot(history.history['accuracy'])\n",
"plt.plot(history.history['val_accuracy'])\n",
"plt.title('Model accuracy')\n",
"plt.ylabel('Accuracy')\n",
"plt.xlabel('Epoch')\n",
"plt.legend(['Train', 'Test'], loc='upper left')"
]
},
{
"cell_type": "markdown",
"id": "31f7b8ff-465e-4c9f-884b-86fe27d63f2f",
"metadata": {},
"source": [
"### Conclusion\n",
"\n",
"The graph indicates that the Conv1D model is effective in learning and generalizing the task of emotion recognition from audio features. Both the training and test accuracies show a consistent improvement over the epochs, with the test accuracy closely following the training accuracy, suggesting that the model does not suffer from significant overfitting."
]
},
{
"cell_type": "markdown",
"id": "187e8084-a43c-4b04-8eb8-1f155f8777b7",
"metadata": {},
"source": [
"# Confusion Matrix\n",
"\n",
"The following plot shows the confusion matrix for the Conv1D model's predictions on the test dataset.\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 68,
"id": "eed99318-753a-4196-a9ea-e41173335c57",
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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",
"text/plain": [
"<Figure size 640x480 with 2 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"import seaborn as sns\n",
"from sklearn.metrics import confusion_matrix\n",
"\n",
"# Compute confusion matrix\n",
"conf_matrix = confusion_matrix(true_labels, predicted_labels)\n",
"sns.heatmap(conf_matrix, annot=True, fmt=\"d\", cmap=\"Blues\", xticklabels=label_encoder.classes_, yticklabels=label_encoder.classes_)\n",
"plt.ylabel('Actual')\n",
"plt.xlabel('Predicted')\n",
"plt.show()\n"
]
},
{
"cell_type": "markdown",
"id": "75016e46-b833-4114-95c6-212bed312c57",
"metadata": {},
"source": [
"\n",
"\n",
"### Description\n",
"\n",
"- **X-axis (Predicted)**: Represents the predicted emotion labels by the model.\n",
"- **Y-axis (Actual)**: Represents the actual emotion labels.\n",
"- **Color Scale**: The intensity of the color represents the number of instances classified in each category, with darker colors indicating a higher number of instances.\n",
"\n",
"### Emotion Labels\n",
"The numerical labels correspond to the following emotions:\n",
"- 0: Neutral\n",
"- 1: Calm\n",
"- 2: Happy\n",
"- 3: Sad\n",
"- 4: Angry\n",
"- 5: Fearful\n",
"- 6: Disgust\n",
"- 7: Surprised\n",
"\n",
"### Observations\n",
"\n",
"- **High Accuracy**: Most predictions are concentrated along the diagonal, indicating correct classifications.\n",
"- **Misclassifications**: Some off-diagonal elements show misclassifications, such as a few instances where 'Neutral' is predicted as 'Calm' or 'Happy'.\n",
"- **Confusion**: Specific emotions like 'Sad' and 'Angry' show some level of confusion with other emotions, as indicated by the lighter cells off the diagonal.\n",
"\n",
"### Conclusion\n",
"\n",
"The confusion matrix indicates that the Conv1D model performs well in classifying most of the emotions correctly, with a high number of correct predictions for 'Neutral', 'Sad', and 'Angry'. Some emotions like 'Calm' and 'Happy' have slight confusion with other emotions, but overall, the model demonstrates strong performance in emotion recognition."
]
},
{
"cell_type": "code",
"execution_count": 61,
"id": "4dc5b551-0161-4b65-80a6-95572d58486e",
"metadata": {},
"outputs": [],
"source": [
"# Save the model in the recommended Keras format\n",
"model.save('emotion_recognition_model.keras')\n"
]
},
{
"cell_type": "markdown",
"id": "16a6eed0-1f41-4865-a630-fb5947b5627c",
"metadata": {},
"source": [
"\n",
"# LSTM Model for Emotion Classification\r\n",
"\r\n",
"## Model Architecture\r\n",
"\r\n",
"- **Input Layer**: The model accepts input with the shape (40, 1), where 40 represents the number of MFCC features extracted from the audio signals.\r\n",
"- **LSTM Layer**: A Long Short-Term Memory (LSTM) layer with 128 units is used. This layer processes the input sequence and outputs the last hidden state.\r\n",
"- **Dense Layer 1**: A fully connected layer with 64 units.\r\n",
"- **Dropout Layer 1**: A dropout layer with a dropout rate of 0.4 is added after the first dense layer to reduce overfitting by randomly setting 40% of the input units to 0 during training.\r\n",
"- **Activation Layer 1**: A ReLU activation function is applied to introduce non-linearity.\r\n",
"- **Dense Layer 2**: Another fully connected layer with 32 units.\r\n",
"- **Dropout Layer 2**: Another dropout layer with a dropout rate of 0.4 is added after the second dense layer to further reduce overfitting.\r\n",
"- **Activation Layer 2**: Another ReLU activation function is applied.\r\n",
"- **Output Layer**: The final output layer consists of 8 units, corresponding to the 8 emotion categories. A softmax activation function is used to output a probability distribution over the emotion categories.\r\n",
"\r\n",
"## Model Compilation\r\n",
"\r\n",
"- **Loss Function**: The model uses `categorical_crossentropy` as the loss function, which is suitable for multi-class classification problems.\r\n",
"- **Optimizer**: The `Adam` optimizer is used to update the model weights during training. Adam is known for its efficiency and effectiveness in training deep learning models.\r\n",
"- **Metrics**: The `accuracy` metric is used to evaluate the model's performance on the training and validation datasets.\r\n",
"\r\n",
"## Usage\r\n",
"\r\n",
"The function `create_model_LSTM` initializes and compiles the LSTM model based on the defined architecture. This function can be called to get a compiled model ready for training on the preprocessed dataset.\r\n"
]
},
{
"cell_type": "code",
"execution_count": 53,
"id": "bc569d3f-f20f-4a5e-a4d2-91de0d483ce8",
"metadata": {},
"outputs": [],
"source": [
"def create_model_LSTM():\n",
" model = Sequential()\n",
" model.add(LSTM(128, return_sequences=False, input_shape=(40, 1)))\n",
" model.add (Dense (64))\n",
" model.add(Dropout (0.4))\n",
" model.add(Activation('relu'))\n",
" model.add(Dense(32))\n",
" model.add(Dropout (0.4))\n",
" model.add(Activation('relu'))\n",
" model.add(Dense(8))\n",
" model.add(Activation('softmax'))\n",
" # Configures the model for training\n",
" model.compile(loss='categorical_crossentropy', optimizer='Adam', metrics=['accuracy'])\n",
" return model"
]
},
{
"cell_type": "code",
"execution_count": 76,
"id": "eda58f26-9557-4d6f-872b-a4fcdb0b1fd2",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"(2880, 40)"
]
},
"execution_count": 76,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"X.shape"
]
},
{
"cell_type": "markdown",
"id": "d50bc884-a1c6-48f1-8327-0950ea81ecb4",
"metadata": {},
"source": [
"## Data Preparation for Model Training\r\n",
"\r\n",
"### Splitting the Dataset\r\n",
"\r\n",
"1. **Determine the Number of Samples**: \r\n",
" - `X.shape[0]` gives the total number of samples in the dataset.\r\n",
"\r\n",
"2. **Calculate the Number of Training, Validation, and Test Samples**:\r\n",
" - **Training Samples**: 80% of the total samples are allocated for training.\r\n",
" - **Validation Samples**: 10% of the total samples are allocated for validation.\r\n",
" - **Test Samples**: 10% of the total samples are allocated for tef_samples * 0.1)\r\n"
]
},
{
"cell_type": "code",
"execution_count": 77,
"id": "e54f28e0-43a4-42e5-ab72-818b97a8b4d5",
"metadata": {},
"outputs": [],
"source": [
"number_of_samples = X.shape[0]\n",
"training_samples = int(number_of_samples * 0.8)\n",
"validation_samples = int(number_of_samples * 0.1)\n",
"test_samples = int(number_of_samples * 0.1)\r\n"
]
},
{
"cell_type": "markdown",
"id": "00ccfac3-a63b-4949-ac64-ec0ca3399897",
"metadata": {},
"source": [
"#### Reshaping the Training Data\n",
"\n",
"**Expand Dimensions of Training Data**:\n",
"\n",
"- **Reshape the Training Data**: \n",
" - The training data, `X[:training_samples]`, is reshaped by adding an additional dimension using `np.expand_dims`.\n",
" - This step is crucial to match the input shape required by the LSTM model, which expects a 3D input.\n",
" - The reshaped data ensures compatibility with the LSTM model's input requirements.\n"
]
},
{
"cell_type": "code",
"execution_count": 81,
"id": "57738349-717f-443d-8480-2f51640b5a18",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"(2304, 40, 1)"
]
},
"execution_count": 81,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"w = np.expand_dims(X[:training_samples],-1)\n",
"w.shape"
]
},
{
"cell_type": "markdown",
"id": "7db51c95-8f30-4578-90f2-5f732a803b10",
"metadata": {},
"source": [
"#### Training the LSTM Model\r\n",
"\r\n",
"**Model Initialization**:\r\n",
"- Create the LSTM model by calling the `create_model_LSTM()` function and assign it to `model_A`.\r\n",
"\r\n",
"**Model Training**:\r\n",
"- Train the model using the `fit` method:\r\n",
" - **Training Data**: `w`, which is the reshaped training data.\r\n",
" - **Training Labels**: `y_cat[:training_samples]`, which are the categorical labels for the training data.\r\n",
" - **Validation Data**: \r\n",
" - Inputs: `np.expand_dims(X[training_samples:training_samples + validation_samples], -1)`, reshaped validation data.\r\n",
" - Labels: `y_cat[training_samples:training_samples + validation_samples]`, categorical labels for the validation data.\r\n",
" - **Epochs**: Train the model for 130 epochs.\r\n",
" - **Shuffle**: Enable shuffling of training data at each epoch to improve generalization.\r\n"
]
},
{
"cell_type": "code",
"execution_count": 83,
"id": "9b377dbb-d02c-42f7-8082-6de1aba76985",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"C:\\Users\\KIIT\\anaconda3\\Lib\\site-packages\\keras\\src\\layers\\rnn\\rnn.py:204: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n",
" super().__init__(**kwargs)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m7s\u001b[0m 30ms/step - accuracy: 0.1385 - loss: 2.0760 - val_accuracy: 0.2014 - val_loss: 2.0433\n",
"Epoch 2/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 22ms/step - accuracy: 0.1759 - loss: 2.0426 - val_accuracy: 0.2326 - val_loss: 1.9693\n",
"Epoch 3/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 21ms/step - accuracy: 0.2172 - loss: 1.9757 - val_accuracy: 0.2465 - val_loss: 1.9313\n",
"Epoch 4/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 21ms/step - accuracy: 0.2460 - loss: 1.9279 - val_accuracy: 0.3264 - val_loss: 1.8912\n",
"Epoch 5/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 21ms/step - accuracy: 0.2609 - loss: 1.8951 - val_accuracy: 0.3194 - val_loss: 1.8512\n",
"Epoch 6/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 22ms/step - accuracy: 0.2863 - loss: 1.8695 - val_accuracy: 0.3264 - val_loss: 1.8080\n",
"Epoch 7/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 22ms/step - accuracy: 0.2855 - loss: 1.8637 - val_accuracy: 0.2986 - val_loss: 1.8396\n",
"Epoch 8/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 23ms/step - accuracy: 0.2954 - loss: 1.8291 - val_accuracy: 0.3056 - val_loss: 1.8039\n",
"Epoch 9/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 22ms/step - accuracy: 0.3015 - loss: 1.8223 - val_accuracy: 0.3264 - val_loss: 1.7575\n",
"Epoch 10/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.2902 - loss: 1.8003 - val_accuracy: 0.3021 - val_loss: 1.7344\n",
"Epoch 11/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.3098 - loss: 1.7580 - val_accuracy: 0.3125 - val_loss: 1.7068\n",
"Epoch 12/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.3321 - loss: 1.7474 - val_accuracy: 0.3125 - val_loss: 1.7494\n",
"Epoch 13/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.2942 - loss: 1.7550 - val_accuracy: 0.3299 - val_loss: 1.7023\n",
"Epoch 14/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.3294 - loss: 1.7053 - val_accuracy: 0.3299 - val_loss: 1.6989\n",
"Epoch 15/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.3304 - loss: 1.6865 - val_accuracy: 0.3438 - val_loss: 1.6837\n",
"Epoch 16/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.3505 - loss: 1.6672 - val_accuracy: 0.3507 - val_loss: 1.6803\n",
"Epoch 17/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.3555 - loss: 1.6649 - val_accuracy: 0.3785 - val_loss: 1.6393\n",
"Epoch 18/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 15ms/step - accuracy: 0.3635 - loss: 1.6090 - val_accuracy: 0.3611 - val_loss: 1.6099\n",
"Epoch 19/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.3768 - loss: 1.6106 - val_accuracy: 0.3681 - val_loss: 1.5997\n",
"Epoch 20/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.3791 - loss: 1.5797 - val_accuracy: 0.3681 - val_loss: 1.5913\n",
"Epoch 21/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.3745 - loss: 1.5890 - val_accuracy: 0.4062 - val_loss: 1.5238\n",
"Epoch 22/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.4029 - loss: 1.5232 - val_accuracy: 0.4340 - val_loss: 1.4941\n",
"Epoch 23/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.3985 - loss: 1.5467 - val_accuracy: 0.4132 - val_loss: 1.5048\n",
"Epoch 24/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.4359 - loss: 1.4789 - val_accuracy: 0.3958 - val_loss: 1.5107\n",
"Epoch 25/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.4214 - loss: 1.5054 - val_accuracy: 0.4167 - val_loss: 1.4671\n",
"Epoch 26/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.4240 - loss: 1.4624 - val_accuracy: 0.4167 - val_loss: 1.4613\n",
"Epoch 27/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.4437 - loss: 1.4796 - val_accuracy: 0.4375 - val_loss: 1.4436\n",
"Epoch 28/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.4420 - loss: 1.4239 - val_accuracy: 0.4340 - val_loss: 1.5034\n",
"Epoch 29/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.4631 - loss: 1.4298 - val_accuracy: 0.4653 - val_loss: 1.4203\n",
"Epoch 30/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.4888 - loss: 1.3788 - val_accuracy: 0.4236 - val_loss: 1.4395\n",
"Epoch 31/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.4805 - loss: 1.3585 - val_accuracy: 0.4375 - val_loss: 1.4169\n",
"Epoch 32/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.4768 - loss: 1.3328 - val_accuracy: 0.4340 - val_loss: 1.4245\n",
"Epoch 33/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.4883 - loss: 1.3612 - val_accuracy: 0.4444 - val_loss: 1.3808\n",
"Epoch 34/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 15ms/step - accuracy: 0.5091 - loss: 1.2764 - val_accuracy: 0.4549 - val_loss: 1.3646\n",
"Epoch 35/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.4976 - loss: 1.2810 - val_accuracy: 0.4653 - val_loss: 1.3463\n",
"Epoch 36/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.5108 - loss: 1.2716 - val_accuracy: 0.4653 - val_loss: 1.3406\n",
"Epoch 37/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.5464 - loss: 1.1851 - val_accuracy: 0.4688 - val_loss: 1.3320\n",
"Epoch 38/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.5583 - loss: 1.1685 - val_accuracy: 0.4583 - val_loss: 1.3303\n",
"Epoch 39/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.5531 - loss: 1.1702 - val_accuracy: 0.4861 - val_loss: 1.3228\n",
"Epoch 40/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.5678 - loss: 1.1451 - val_accuracy: 0.4965 - val_loss: 1.2761\n",
"Epoch 41/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.5576 - loss: 1.1225 - val_accuracy: 0.5035 - val_loss: 1.2133\n",
"Epoch 42/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.6005 - loss: 1.0755 - val_accuracy: 0.4722 - val_loss: 1.2877\n",
"Epoch 43/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.5999 - loss: 1.0849 - val_accuracy: 0.5486 - val_loss: 1.1625\n",
"Epoch 44/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.6124 - loss: 1.0376 - val_accuracy: 0.5139 - val_loss: 1.2740\n",
"Epoch 45/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.5941 - loss: 1.0931 - val_accuracy: 0.5243 - val_loss: 1.2551\n",
"Epoch 46/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.6204 - loss: 1.0023 - val_accuracy: 0.5382 - val_loss: 1.2210\n",
"Epoch 47/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 15ms/step - accuracy: 0.6260 - loss: 0.9726 - val_accuracy: 0.5486 - val_loss: 1.1852\n",
"Epoch 48/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 15ms/step - accuracy: 0.6579 - loss: 0.9275 - val_accuracy: 0.5590 - val_loss: 1.1390\n",
"Epoch 49/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.6578 - loss: 0.9323 - val_accuracy: 0.5625 - val_loss: 1.1475\n",
"Epoch 50/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.6344 - loss: 1.0003 - val_accuracy: 0.5938 - val_loss: 1.0751\n",
"Epoch 51/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.6757 - loss: 0.8811 - val_accuracy: 0.5903 - val_loss: 1.1043\n",
"Epoch 52/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.6650 - loss: 0.9288 - val_accuracy: 0.5972 - val_loss: 1.0543\n",
"Epoch 53/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.6940 - loss: 0.8132 - val_accuracy: 0.6146 - val_loss: 0.9922\n",
"Epoch 54/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.6904 - loss: 0.8325 - val_accuracy: 0.5833 - val_loss: 1.1041\n",
"Epoch 55/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.6828 - loss: 0.8567 - val_accuracy: 0.6319 - val_loss: 0.9416\n",
"Epoch 56/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.6997 - loss: 0.8117 - val_accuracy: 0.6215 - val_loss: 1.0143\n",
"Epoch 57/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 19ms/step - accuracy: 0.7020 - loss: 0.8106 - val_accuracy: 0.6319 - val_loss: 0.9540\n",
"Epoch 58/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.7438 - loss: 0.7429 - val_accuracy: 0.6215 - val_loss: 0.9677\n",
"Epoch 59/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.7177 - loss: 0.7774 - val_accuracy: 0.6597 - val_loss: 0.9305\n",
"Epoch 60/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.7146 - loss: 0.7585 - val_accuracy: 0.6528 - val_loss: 0.9641\n",
"Epoch 61/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.7481 - loss: 0.6814 - val_accuracy: 0.6910 - val_loss: 0.7954\n",
"Epoch 62/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.7523 - loss: 0.6655 - val_accuracy: 0.6806 - val_loss: 0.8791\n",
"Epoch 63/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.7756 - loss: 0.6341 - val_accuracy: 0.6771 - val_loss: 0.8422\n",
"Epoch 64/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.7310 - loss: 0.7380 - val_accuracy: 0.7083 - val_loss: 0.8076\n",
"Epoch 65/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 19ms/step - accuracy: 0.7553 - loss: 0.6756 - val_accuracy: 0.7396 - val_loss: 0.6976\n",
"Epoch 66/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.8022 - loss: 0.5456 - val_accuracy: 0.7257 - val_loss: 0.6977\n",
"Epoch 67/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 19ms/step - accuracy: 0.7832 - loss: 0.6260 - val_accuracy: 0.7014 - val_loss: 0.7721\n",
"Epoch 68/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 21ms/step - accuracy: 0.7536 - loss: 0.6623 - val_accuracy: 0.7396 - val_loss: 0.7405\n",
"Epoch 69/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 22ms/step - accuracy: 0.7990 - loss: 0.6290 - val_accuracy: 0.7188 - val_loss: 0.8257\n",
"Epoch 70/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 22ms/step - accuracy: 0.7708 - loss: 0.6380 - val_accuracy: 0.7812 - val_loss: 0.6389\n",
"Epoch 71/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 23ms/step - accuracy: 0.8145 - loss: 0.5247 - val_accuracy: 0.7049 - val_loss: 0.8497\n",
"Epoch 72/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.7706 - loss: 0.6614 - val_accuracy: 0.7326 - val_loss: 0.7003\n",
"Epoch 73/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.8195 - loss: 0.5402 - val_accuracy: 0.7847 - val_loss: 0.5694\n",
"Epoch 74/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.8498 - loss: 0.4310 - val_accuracy: 0.8125 - val_loss: 0.5392\n",
"Epoch 75/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.8564 - loss: 0.4021 - val_accuracy: 0.8090 - val_loss: 0.5493\n",
"Epoch 76/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.8385 - loss: 0.4566 - val_accuracy: 0.8160 - val_loss: 0.4707\n",
"Epoch 77/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.8703 - loss: 0.3916 - val_accuracy: 0.8125 - val_loss: 0.5303\n",
"Epoch 78/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.8670 - loss: 0.4032 - val_accuracy: 0.8229 - val_loss: 0.5355\n",
"Epoch 79/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.7980 - loss: 0.5889 - val_accuracy: 0.8090 - val_loss: 0.5217\n",
"Epoch 80/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.8532 - loss: 0.4392 - val_accuracy: 0.7465 - val_loss: 0.7366\n",
"Epoch 81/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.8216 - loss: 0.5801 - val_accuracy: 0.8542 - val_loss: 0.4492\n",
"Epoch 82/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 20ms/step - accuracy: 0.8776 - loss: 0.3518 - val_accuracy: 0.8611 - val_loss: 0.4665\n",
"Epoch 83/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 21ms/step - accuracy: 0.8704 - loss: 0.3913 - val_accuracy: 0.8681 - val_loss: 0.3597\n",
"Epoch 84/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 21ms/step - accuracy: 0.8794 - loss: 0.3654 - val_accuracy: 0.8819 - val_loss: 0.3734\n",
"Epoch 85/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 19ms/step - accuracy: 0.9052 - loss: 0.2984 - val_accuracy: 0.8924 - val_loss: 0.3333\n",
"Epoch 86/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9122 - loss: 0.2744 - val_accuracy: 0.8854 - val_loss: 0.3645\n",
"Epoch 87/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.8811 - loss: 0.3982 - val_accuracy: 0.8507 - val_loss: 0.4365\n",
"Epoch 88/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.8503 - loss: 0.4938 - val_accuracy: 0.8472 - val_loss: 0.4148\n",
"Epoch 89/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.8694 - loss: 0.4349 - val_accuracy: 0.8507 - val_loss: 0.4137\n",
"Epoch 90/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.8829 - loss: 0.3551 - val_accuracy: 0.8889 - val_loss: 0.3069\n",
"Epoch 91/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9086 - loss: 0.2807 - val_accuracy: 0.9271 - val_loss: 0.2338\n",
"Epoch 92/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.8819 - loss: 0.3671 - val_accuracy: 0.8611 - val_loss: 0.4722\n",
"Epoch 93/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.8720 - loss: 0.4260 - val_accuracy: 0.8750 - val_loss: 0.3514\n",
"Epoch 94/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9211 - loss: 0.2517 - val_accuracy: 0.9410 - val_loss: 0.2341\n",
"Epoch 95/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9250 - loss: 0.2363 - val_accuracy: 0.8750 - val_loss: 0.4733\n",
"Epoch 96/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 19ms/step - accuracy: 0.8295 - loss: 0.7536 - val_accuracy: 0.7812 - val_loss: 0.7854\n",
"Epoch 97/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.8780 - loss: 0.3910 - val_accuracy: 0.8819 - val_loss: 0.3962\n",
"Epoch 98/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.9286 - loss: 0.2540 - val_accuracy: 0.9097 - val_loss: 0.2751\n",
"Epoch 99/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.9195 - loss: 0.2469 - val_accuracy: 0.8889 - val_loss: 0.2868\n",
"Epoch 100/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9285 - loss: 0.2064 - val_accuracy: 0.9340 - val_loss: 0.2146\n",
"Epoch 101/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9480 - loss: 0.1748 - val_accuracy: 0.9375 - val_loss: 0.2259\n",
"Epoch 102/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.9600 - loss: 0.1475 - val_accuracy: 0.9201 - val_loss: 0.2264\n",
"Epoch 103/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9378 - loss: 0.1833 - val_accuracy: 0.9722 - val_loss: 0.1391\n",
"Epoch 104/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9416 - loss: 0.2220 - val_accuracy: 0.8889 - val_loss: 0.3331\n",
"Epoch 105/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.8693 - loss: 0.4806 - val_accuracy: 0.8958 - val_loss: 0.4250\n",
"Epoch 106/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 19ms/step - accuracy: 0.8749 - loss: 0.3716 - val_accuracy: 0.9479 - val_loss: 0.1862\n",
"Epoch 107/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 19ms/step - accuracy: 0.9430 - loss: 0.1981 - val_accuracy: 0.9479 - val_loss: 0.1678\n",
"Epoch 108/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.9491 - loss: 0.1635 - val_accuracy: 0.9583 - val_loss: 0.1308\n",
"Epoch 109/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 19ms/step - accuracy: 0.9563 - loss: 0.1434 - val_accuracy: 0.9236 - val_loss: 0.2487\n",
"Epoch 110/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.9105 - loss: 0.2958 - val_accuracy: 0.9340 - val_loss: 0.2683\n",
"Epoch 111/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9358 - loss: 0.2196 - val_accuracy: 0.9549 - val_loss: 0.1811\n",
"Epoch 112/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9491 - loss: 0.2278 - val_accuracy: 0.9549 - val_loss: 0.1639\n",
"Epoch 113/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.9447 - loss: 0.1797 - val_accuracy: 0.9271 - val_loss: 0.2196\n",
"Epoch 114/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9286 - loss: 0.2494 - val_accuracy: 0.9167 - val_loss: 0.3452\n",
"Epoch 115/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9184 - loss: 0.3099 - val_accuracy: 0.8646 - val_loss: 0.4911\n",
"Epoch 116/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 19ms/step - accuracy: 0.8553 - loss: 0.6096 - val_accuracy: 0.9097 - val_loss: 0.3920\n",
"Epoch 117/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9028 - loss: 0.3418 - val_accuracy: 0.9132 - val_loss: 0.3387\n",
"Epoch 118/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9131 - loss: 0.3048 - val_accuracy: 0.9514 - val_loss: 0.1927\n",
"Epoch 119/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9506 - loss: 0.1871 - val_accuracy: 0.9722 - val_loss: 0.1258\n",
"Epoch 120/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.9440 - loss: 0.2410 - val_accuracy: 0.9653 - val_loss: 0.1218\n",
"Epoch 121/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.9566 - loss: 0.1540 - val_accuracy: 0.9896 - val_loss: 0.0646\n",
"Epoch 122/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 18ms/step - accuracy: 0.9664 - loss: 0.1116 - val_accuracy: 0.9861 - val_loss: 0.0637\n",
"Epoch 123/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9732 - loss: 0.0885 - val_accuracy: 0.9826 - val_loss: 0.0530\n",
"Epoch 124/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.9746 - loss: 0.1062 - val_accuracy: 0.9792 - val_loss: 0.0599\n",
"Epoch 125/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9544 - loss: 0.1428 - val_accuracy: 0.9340 - val_loss: 0.1840\n",
"Epoch 126/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9577 - loss: 0.1525 - val_accuracy: 0.9653 - val_loss: 0.1477\n",
"Epoch 127/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.9581 - loss: 0.1310 - val_accuracy: 0.9896 - val_loss: 0.0937\n",
"Epoch 128/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.9600 - loss: 0.1122 - val_accuracy: 0.9792 - val_loss: 0.0769\n",
"Epoch 129/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 17ms/step - accuracy: 0.9633 - loss: 0.1239 - val_accuracy: 0.9583 - val_loss: 0.2492\n",
"Epoch 130/130\n",
"\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 16ms/step - accuracy: 0.9667 - loss: 0.1483 - val_accuracy: 0.9618 - val_loss: 0.1356\n"
]
}
],
"source": [
"model_A = create_model_LSTM()\n",
"history = model_A.fit(w, y_cat[:training_samples], validation_data=(np.expand_dims(X[training_samples: training_samples+validation_samples], -1),y_cat[training_samples: training_samples+validation_samples]), epochs=130, shuffle=True)"
]
},
{
"cell_type": "markdown",
"id": "bb3bf8c7-1d41-464b-83c9-9926ceee4d62",
"metadata": {},
"source": [
"### Model Evaluation\n",
"\n",
"After training, the model is evaluated on the test set to determine its final accuracy. The test accuracy provides an indication of how well the model generalizes to unseen data."
]
},
{
"cell_type": "code",
"execution_count": 85,
"id": "74caa6ca-d114-4d23-bf42-a80728f81ae8",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[1m81/81\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 12ms/step - accuracy: 0.9899 - loss: 0.0427\n",
"Evaluation Metrics:\n",
"Loss: 0.05488839000463486\n",
"Accuracy: 0.9868826866149902\n",
"\u001b[1m18/18\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 12ms/step\n",
"True label: 5, Predicted label: 2\n",
"True label: 2, Predicted label: 2\n",
"True label: 3, Predicted label: 3\n",
"True label: 1, Predicted label: 1\n",
"True label: 1, Predicted label: 1\n",
"True label: 1, Predicted label: 1\n",
"True label: 7, Predicted label: 7\n",
"True label: 0, Predicted label: 0\n",
"True label: 3, Predicted label: 3\n",
"True label: 2, Predicted label: 2\n"
]
}
],
"source": [
"\n",
"\n",
"# Evaluation on test data\n",
"evaluation = model_A.evaluate(\n",
" np.expand_dims(X[test_samples:], -1), # Test data (assuming X is your test data)\n",
" y_cat[test_samples:], # Test labels in categorical format (assuming y_cat is your test labels)\n",
")\n",
"\n",
"# Printing evaluation metrics\n",
"print(\"Evaluation Metrics:\")\n",
"print(\"Loss:\", evaluation[0])\n",
"print(\"Accuracy:\", evaluation[1])\n",
"\n",
"predictions = model_A.predict(X_test)\n",
"predicted_labels = np.argmax(predictions, axis=1)\n",
"true_labels = np.argmax(y_test, axis=1)\n",
"\n",
"# Print some predictions\n",
"for i in range(10):\n",
" print(f\"True label: {true_labels[i]}, Predicted label: {predicted_labels[i]}\")\n"
]
},
{
"cell_type": "markdown",
"id": "afb7c218-949d-4729-9beb-63516a267d7e",
"metadata": {},
"source": [
"\n",
"# Model Training and Test Accuracy\n",
"\n",
"The following plot shows the accuracy of the Conv1D model over 100 epochs for both the training and test datasets.\n"
]
},
{
"cell_type": "code",
"execution_count": 88,
"id": "0c09a48c-9541-48b7-a14d-255d02d6feb4",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"<matplotlib.legend.Legend at 0x1e6e7111350>"
]
},
"execution_count": 88,
"metadata": {},
"output_type": "execute_result"
},
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAe4AAAGHCAYAAAB27LHEAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAAB6DElEQVR4nO3deVxU5f7A8c/MAMO+78jqiruC+66pqWVW5lYurddsM6tbXn83yyy7dStbtcX0lpVmWVlZilru+4r7DiggArLLMjPn98eBgRFQ0EEEvu/Xa15z5sxzznkeF748u0ZRFAUhhBBC1Ana2s6AEEIIIapOArcQQghRh0jgFkIIIeoQCdxCCCFEHSKBWwghhKhDJHALIYQQdYgEbiGEEKIOkcAthBBC1CESuIUQQog6RAK3EDdo0aJFaDQaNBoNf//9d7nvFUWhSZMmaDQa+vbta9VnazQaXnnllWpfd/bsWTQaDYsWLbJqfoQQNU8CtxBW4uLiwoIFC8qdX79+PadOncLFxaUWciWEqG8kcAthJaNHj+bHH38kKyvL4vyCBQvo1q0bISEhtZSzhqOoqAiDwVDb2RCiRkngFsJKxo4dC8B3331nPpeZmcmPP/7IQw89VOE16enpTJkyhaCgIOzs7IiIiGDGjBkUFBRYpMvKyuLRRx/Fy8sLZ2dnbr/9do4fP17hPU+cOMG4cePw9fVFr9cTGRnJxx9/fF1lys/P57nnnqN9+/a4ubnh6elJt27d+OWXX8qlNZlMfPjhh7Rv3x4HBwfc3d3p2rUrK1assEj37bff0q1bN5ydnXF2dqZ9+/YWLRVhYWFMmjSp3P379u1r0dXw999/o9Fo+Prrr3nuuecICgpCr9dz8uRJLl68yJQpU2jZsiXOzs74+vrSv39/Nm7cWO6+BQUFzJo1i8jISOzt7fHy8qJfv35s2bIFgAEDBtCiRQuu3I+ppAtk2LBh1fkjFeKG2dR2BoSoL1xdXRk5ciRffvkl//jHPwA1iGu1WkaPHs3cuXMt0ufn59OvXz9OnTrFq6++Stu2bdm4cSNz5sxh3759/P7774AaIEaMGMGWLVt4+eWX6dSpE5s3b2bIkCHl8nD48GG6d+9OSEgI77zzDv7+/qxatYqnn36a1NRUZs6cWa0yFRQUkJ6ezvPPP09QUBCFhYWsWbOGe+65h4ULFzJhwgRz2kmTJrF48WIefvhhZs2ahZ2dHXv27OHs2bPmNC+//DKvvfYa99xzD8899xxubm4cPHiQuLi4auWrrOnTp9OtWzfmz5+PVqvF19eXixcvAjBz5kz8/f3Jycnhp59+om/fvqxdu9b8C4DBYGDIkCFs3LiRqVOn0r9/fwwGA9u2bSM+Pp7u3bvzzDPPcNddd7F27Vpuu+0283P/+OMPTp06xQcffHDdeRfiuihCiBuycOFCBVB27typ/PXXXwqgHDx4UFEURenUqZMyadIkRVEUpVWrVkqfPn3M182fP18BlO+//97ifv/5z38UQFm9erWiKIryxx9/KIDy/vvvW6R7/fXXFUCZOXOm+dzgwYOVRo0aKZmZmRZpn3zyScXe3l5JT09XFEVRzpw5owDKwoULq1VWg8GgFBUVKQ8//LDSoUMH8/kNGzYogDJjxoxKrz19+rSi0+mU+++//6rPCA0NVSZOnFjufJ8+fSz+/Er+rHv37l3lfA8YMEC5++67zee/+uorBVA+//zzSq81Go1KRESEctddd1mcHzJkiNK4cWPFZDJd8/lCWJM0lQthRX369KFx48Z8+eWXxMbGsnPnzkqbydetW4eTkxMjR460OF/STLx27VoA/vrrLwDuv/9+i3Tjxo2z+Jyfn8/atWu5++67cXR0xGAwmF9Dhw4lPz+fbdu2VbtMy5Yto0ePHjg7O2NjY4OtrS0LFizgyJEj5jR//PEHAE888USl94mJicFoNF41zfW49957Kzw/f/58OnbsiL29vTnfa9euLZdve3v7Sv+OALRaLU8++SS//fYb8fHxAJw6dYo///yTKVOmoNForFoeIa5FArcQVqTRaHjwwQdZvHgx8+fPp1mzZvTq1avCtGlpafj7+5f7we/r64uNjQ1paWnmdDY2Nnh5eVmk8/f3L3c/g8HAhx9+iK2trcVr6NChAKSmplarPMuXL2fUqFEEBQWxePFitm7dav5lJD8/35zu4sWL6HS6cnkqq6T5ulGjRtXKw7UEBASUO/fuu+/y+OOP06VLF3788Ue2bdvGzp07uf3227l8+bJFngIDA9Fqr/6j8KGHHsLBwYH58+cD8PHHH+Pg4HDVgC9ETZE+biGsbNKkSbz88svMnz+f119/vdJ0Xl5ebN++HUVRLIJ3SkoKBoMBb29vczqDwUBaWppF8E5OTra4n4eHBzqdjvHjx1daqw0PD69WWRYvXkx4eDhLly61yOOVg+d8fHwwGo0kJydXGEhL0gCcO3eO4ODgSp9pb29f7v6g/tJR8mdSVkU13sWLF9O3b1/mzZtncT47O7tcnjZt2oTJZLpq8HZzc2PixIl88cUXPP/88yxcuJBx48bh7u5e6TVC1BSpcQthZUFBQbzwwgvceeedTJw4sdJ0AwYMICcnh59//tni/FdffWX+HqBfv34AfPPNNxbpvv32W4vPjo6O9OvXj71799K2bVuio6PLva6stV+LRqPBzs7OIjgmJyeXG1VeMlDuykBZ1qBBg9DpdFdNA+qo8gMHDlicO378OMeOHatWvvV6vcW5AwcOsHXr1nL5zs/Pr9JCNCUD/EaOHElGRgZPPvlklfMjhDVJjVuIGvDmm29eM82ECRP4+OOPmThxImfPnqVNmzZs2rSJN954g6FDh5pHMA8aNIjevXvzz3/+k9zcXKKjo9m8eTNff/11uXu+//779OzZk169evH4448TFhZGdnY2J0+e5Ndff2XdunXVKscdd9zB8uXLmTJlCiNHjiQhIYHXXnuNgIAATpw4YU7Xq1cvxo8fz+zZs7lw4QJ33HEHer2evXv34ujoyFNPPUVYWBj/+te/eO2117h8+TJjx47Fzc2Nw4cPk5qayquvvgrA+PHjeeCBB5gyZQr33nsvcXFxvPXWW+Yae1Xz/dprrzFz5kz69OnDsWPHmDVrFuHh4RbzvMeOHcvChQuZPHkyx44do1+/fphMJrZv305kZCRjxowxp23WrBm33347f/zxBz179qRdu3bV+rMUwmpqe3ScEHVd2VHlV3PlqHJFUZS0tDRl8uTJSkBAgGJjY6OEhoYq06dPV/Lz8y3SZWRkKA899JDi7u6uODo6KgMHDlSOHj1ablS5oqgjxh966CElKChIsbW1VXx8fJTu3bsrs2fPtkhDFUeVv/nmm0pYWJii1+uVyMhI5fPPP1dmzpypXPnjw2g0Ku+9957SunVrxc7OTnFzc1O6deum/PrrrxbpvvrqK6VTp06Kvb294uzsrHTo0MEiHyaTSXnrrbeUiIgIxd7eXomOjlbWrVtX6ajyZcuWlctzQUGB8vzzzytBQUGKvb290rFjR+Xnn39WJk6cqISGhlqkvXz5svLyyy8rTZs2Vezs7BQvLy+lf//+ypYtW8rdd9GiRQqgLFmy5Jp/bkLUFI2iXLGqgBBCiArde++9bNu2jbNnz2Jra1vb2RENlDSVCyHEVRQUFLBnzx527NjBTz/9xLvvvitBW9QqqXELIcRVnD17lvDwcFxdXRk3bhwfffQROp2utrMlGjAJ3EIIIUQdItPBhBBCiDpEArcQQghRh0jgFkIIIeqQBjeq3GQykZiYiIuLi2wOIIQQ4pagKArZ2dlVWju/wQXuxMTEq66TLIQQQtSWhISEa27E0+ACt4uLC6D+4bi6utZyboQQQgjIysoiODjYHKOupsEF7pLmcVdXVwncQgghbilV6cKVwWlCCCFEHSKBWwghhKhDJHALIYQQdUit9nFv2LCBt99+m927d5OUlMRPP/3EiBEjrnrN+vXrmTZtGocOHSIwMJB//vOfTJ482ar5UhQFg8GA0Wi06n0bEp1Oh42NjUy5E0IIK6vVwJ2bm0u7du148MEHuffee6+Z/syZMwwdOpRHH32UxYsXs3nzZqZMmYKPj0+Vrq+KwsJCkpKSyMvLs8r9GjJHR0cCAgKws7Or7awIIUS9UauBe8iQIQwZMqTK6efPn09ISAhz584FIDIykl27dvHf//7XKoHbZDJx5swZdDodgYGB2NnZSY3xOiiKQmFhIRcvXuTMmTM0bdr0mgsKCCGEqJo6NR1s69atDBo0yOLc4MGDWbBgAUVFRRXukVtQUEBBQYH5c1ZWVqX3LywsxGQyERwcjKOjo/Uy3gA5ODhga2tLXFwchYWF2Nvb13aWhBCiXqhT1aDk5GT8/Pwszvn5+WEwGEhNTa3wmjlz5uDm5mZ+VWXVNKkdWof8OQohhPXVuZ+sVzZdl2wnXlmT9vTp08nMzDS/EhISajyPQgghRE2pU03l/v7+JCcnW5xLSUnBxsYGLy+vCq/R6/Xo9fqbkT0hhBC1LSsJcpIhoD3cyBiltFOg1YFHmLVyZjV1qsbdrVs3YmJiLM6tXr2a6OjoCvu3xY3p27cvU6dOre1sCCFE1RiLYMEg+KwvfNgRNr6jBvLqyk2DT/vAJ93g/B6rZ/NG1WrgzsnJYd++fezbtw9Qp3vt27eP+Ph4QG3mnjBhgjn95MmTiYuLY9q0aRw5coQvv/ySBQsW8Pzzz9dG9m8ZGo3mqq9JkyZd132XL1/Oa6+9Zt3MCiFETTm5FjLV+EH6aVg7C95rBd+OgaO/q4G9Kg7+CIXZUJQH342BjPiay/N1qNWm8l27dtGvXz/z52nTpgEwceJEFi1aRFJSkjmIA4SHh7Ny5UqeffZZPv74YwIDA/nggw+sNoe7rkpKKv2NcunSpbz88sscO3bMfM7BwcEifWUj8K/k6elpvUwKIURN2/+d+h71IDSKhj1fQ8I2OP6H+nLyhfZjocME8G5ylft8q77bOEDOBfh2NDz0J9i71XwZqqBWa9x9+/ZFUZRyr0WLFgGwaNEi/v77b4tr+vTpw549eygoKODMmTNWXzXtSoqikFdoqJVXycC7a/H39ze/3Nzc0Gg05s/5+fm4u7vz/fff07dvX+zt7Vm8eDFpaWmMHTuWRo0a4ejoSJs2bfjuu+8s7ntlU3lYWBhvvPEGDz30EC4uLoSEhPDZZ59Z849bCCGuz+VLcGylehz9EHR4AB5eBU/shO5Pg5MP5KbA5vfhoyj4bhyYTOXvk3IUEveC1kYN1s7+kHIYfngYqvgzuabVqcFpteFykZGWL6+qlWcfnjUYRzvr/BW9+OKLvPPOOyxcuBC9Xk9+fj5RUVG8+OKLuLq68vvvvzN+/HgiIiLo0qVLpfd55513eO211/jXv/7FDz/8wOOPP07v3r1p0aKFVfIphBDX5dBPYCwE31bg36b0vE8zGPQaDHgZjq+CPV/BidVw7HdIPQ6+V/zsOrBEfW8yEALbw7ilar/5yRhI2AEhlf98vFnq1OA0cf2mTp3KPffcQ3h4OIGBgQQFBfH888/Tvn17IiIieOqppxg8eDDLli276n2GDh3KlClTaNKkCS+++CLe3t7lWkWEEHVA9gXYNFcdiFUXJexUm8JL+q33FbcYth9b4WjyxGwDKwo78oH/65x2aA2AkrjXMpHJCPuXlt4H1ODd9j71ePt8Kxfi+kiN+xocbHUcnjW41p5tLdHR0RafjUYjb775JkuXLuX8+fPmFeacnJyuep+2bduaj0ua5FNSUqyWTyHETbLpPdg+D4ouQ7/ptZ2b6snPgsX3QEEW7F0MA/4N53aARgtt7rNIejIlh0/+Pskv+xIxmtSmbhebQCJsYrlwbDv+JQEa4MwGyE4Ee3dodnvp+S6T1ecc/gUyz4Nb0E0oZOUkcF+DRqOxWnN1bboyIL/zzju89957zJ07lzZt2uDk5MTUqVMpLCy86n2uHNSm0WgwVdRPJIS4tSUfUN8vnandfFyP/d+pQRvUwWeL7lCPGw8AF38ALmYXMGflEX7ad97cNd2ukRtN/VzIPdcGMleRc3ZX+fsCtL4XbMqs/+HfBkJ7Qtwm2LVAbXavRXU/IonrsnHjRu666y4eeOABQN1g5cSJE0RGRtZyzoQQNU5R1AFXAFmJtZuX6jKZYPun6nH3p9UpYCmH1M/txmAyKXy/K4E5fxwl87LajH5bpB9P9W9Cu2B3AE4eLIAf/ktA3nEuZOTi5+4EhXlw5Ff1Pu3HlXtsTvuHcY7bxOVtC/i8aAT2jk64afPxNSTRt8+Am7ohlQTuBqpJkyb8+OOPbNmyBQ8PD959912Sk5MlcAvREORcUEdhQ90L3KfWQvop0LtBnxeh73SIeVldLS3yTp5espffDqhTZFsFuvLG3W3MAbtEk5ZR5Gv0OFHAD+s3MfGuwXB2kzpv2y0YgqLYG3+JDcdTOZiYyeHELJIzbFiv96ZRUSp2m97CU5PJUO12UvBA0+foTf0jkMDdQP373//mzJkzDB48GEdHRx577DFGjBhBZmZmbWdNCFHTSmrbANlJag28OjXGbfNg5xcwejH43uRf9ksGiHUcD3pn9XjYfwE4nJjFbweSsNFqeGlICyZ1D8NGV8EYbK2OPI+W2Kfv5eSBzRQOG4jdidUAFEUM4JWfD/LN9isXXdHxm91QJhd9xWSbX81n9bZ2kJ0MrgHWLmmlJHDXM5MmTbJYKS0sLKzC+eCenp78/PPPV73XlaPFz549Wy5Nyap3Qog6JOVI6XFRHuRngINH1a49sAz+fAkAZc/XaG5/o+J0h3+BLR+BYlQ/O3rDiHngVPG+ElWSegJOrgE00OmRcl8v2akG28Gt/HmkV8RVb+XWuBOk7yW04Dh/HkzijhMxaIF/HwpgSZZ6n6Ft/IkO9aRVoCstA11xMXWDeTHq4LjWd0OHCQQEd76xNdGvgwRuIYRoaMrWuEFdz7sqgTtuK8ovUygJUxcOxOBfUeA2FMDvz6sLnpS19yvo+ex1ZRmAHcULPjUfAp7hFl9dLjTy097zAIztHHLNW+mCOsBOaKM9w+yfVjNcc5YCxYYVWU0Jcnfg7ZFt6d7E+4qrPOHpvYAGbO2vvxw3SAK3EELUc5l5RRxKzORQYhbnMy4zPfEQFnsmZiWCX8vyF+akwIGlaiBWFNj2CRpjIRuMbeiti8U/7zhrdx9hQNQVzeWHflaDtksA3DEXzm6ErR/B4RXXH7hNJoj9QT3u/Gi5r3+PTSI730CwpwPdG1ehVh/QHoBWmjg6G3aCLezXtmR0jxZMG9gMF/tKloW2daj4/E0kgVsIIeqxX/cn8tz3+yk0qtM2NZh40b64qdwjDC6dhazzFV+8bjbs+Z/FqX2mCB4rmkaM/lWCDWf5fcX3NAl7nlCv4imniqLODwfo9DA0vx2COsLWjyFxD2QkgHtw9QuSdgIup6vrh4f25MSFbM6k5jIg0g+dVsN3O9Tm7TGdQtBqq9B07d0MbBxwNlxmmtsGyIPOA0fTuXur6uftJpOV04QQ4laSm2a1NbFTcwr4v58PUmg0EeTuwJDW/rR3ycaBfIqwRQntoSbMrmTry5KVxZoMJLvlOD5XRvBQ4T+Z2CeSwI7qwlQdjAd4fPEe8ouK+7LP7VKv0+k5EzaK99ecYNDnR9ilNAcgdu035BQYql+Y+K3qe6No8hUdYz/fxmNf72bMZ1tZfSiZ3XGXsNFquC+6UdXup7MxL43qlJegnms6qPr5qgUSuIUQ4lYRvx3ebgyr/mWV272x8giZl4toFejK+hf6Mu+BKN7pawfACVMg29OLR2VXVOM2GeFi8S6DQ/7Dy8o/eL1gFE3Cw3hhUHN0EX0A6GlzmMNJWaWjsItHfR/xGUy/Tw7y3prjHL+Qw++GTgDk7f+JHm+u4+D5as5gid+mvod0ZdWhZFJz1MWidp69xGNf7wZgQKQvvi7V6HsOaFd67B4C3k2rl6daIoFbCCFuFXGbAEXtE75BW08kkbPvZxbYvs2K7LHYHF4OQIQpDoBjSiN+OlW86mFWBTXu9DNgLFCbpj3C2B2nzvt+qn8TdYpVaA/QaAknET/SWbozHiXzPBz+GYAZ57sD0LuZD/+9rx33PfA4AJ20x7C9nMrUpftKa+lVEbdFfQ/pam4WH9s5mK4RpdsPj6nCoDQLge1Lj5sOuumjw6+X9HELIcStIr14+dFLcdWfW11GYfxOmn17H5/ZFS+yUoRai28+1DwVzDGoNUnx7gBkX4zH5cqblIw892nOpctG4tPzAGgbpF6Dg7s6wCtxD33sDvP9BU9SYubiZzKQ7N6RPckhNPdz4X8PdipeVawRBHZEm7iHexz38VmKG2+vOsa/76hgUNyVshIhIw40Ws44tGLb6T1oNfBk/6YEuNqzfO95cgsM9G3mU70/qOIBaoC6G1gdITVuIYS4VVw6q74XZEFe+nXf5tgf8/FSLpGGGwVdngL3UHW1tJ2fmwP3bX360qyp2u9syDjP38eumLpVMtfbtyWxxc3aYV6OuDmWGW0d3huAkR6nGaLdjt9BdbrWJ/lq//cD3UItlwJtORyAyT4HAViw6QxbTqVeu0AlzeR+rVmyPwOAvs19CXJ3QKvVMDKqERO7h1V/2VGfFuARDq5BEN6retfWIgncQghxMygK7PrScvGTK5UEbrjuzT/i0nIxFQ8qOxP9b/RDZkNfdcEUNr2n7kEN6AJa8+Ko/gB4aHJ4+ustxJ4r0+9cUuP2jeTAuQwA2jZyt3xYceDumL+N92w/ASC+yXi+ymiDk52OuztcsYtWpBq4PS9u56GO6r1eWHaAC1n5Vy9UceA2Bnflh93nABjT6TpGpl9JZwP/2ACPbwa7q++MeCuRwC2EEDfDkRXw27Pw+3MVf28ogMxzpZ/Tqx+4FUXh1Z/30wK1Dziqaz/1izajwKupuj65sRDsnMEtGFtHdxRbNWB5GlP5LbbMuuVlatz7iwN620Zulg8M6QpaW2wKMrDXFBFj7MhdJ4cBcHfHIJz1V/TGejUGv9ZgMvAv5XPCPO05n3GZYR9sZOupq+wLXjyifL8mkrTcQnxd9PRv4VvtP58K2btWfdW4W4QE7npAo9Fc9VV2CdTqCgsLY+7cuVbLqxAN1pniAWeV1bgz4oHSaWCL//iLR/63kz3xlyq95eVCI19sPM2mE6mYTAorY5NJOrkPvaYIk50LGs/iZT91NpZ7bvtGqv3nGg2a4jW2/TWXSMkqUL83FEDaSfXYr2XlNW47J2ikjha/6NyCZ4qe5FK+OuDtga6hFWf69jdBa4PNkZ/4peXftPB3ITWnkPu/2MZnG06VT5+fBRfUpvVFCX4AjO4UXPEa5A2EDE6rB5KSSkeELl26lJdffpljx46Zzzk41P5KP0I0eCX9tJfTIT8T7K+ovV5Rw9ZnJ7Dmktrv/MXEThXe8pO/T/LhOjXANvJw4HKhkf5a9T7awPagLRPcWt4Nfu+qQdCnRel510BIO4k/6aVN1qkn1DXG7d24oHhwIasArQZaB7mWz8Sg2XBgCdoOT1H44UEwKXQO86SFfwVpQe1LvvMD+GUKbrs+YMWwJrx0ph3L95znjZVH6RzuRfuyu3md2wmKCZN7KCvjNIDCvR2rOFe7nmq4v7JUlaJAYW7tvKq4CIO/v7/55ebmhkajsTi3YcMGoqKisLe3JyIigldffRWDoXQBhFdeeYWQkBD0ej2BgYE8/fTTAPTt25e4uDieffZZc+1dCHEd8jPNtUag4mbw4v7tIkUHQHfPbAAuZhdUeEtFUfh1v9q0bavTcO7SZdJyC+nuUNzcXnaOMqhBfPgHENHXcslQl0AA/DVlAncFzeRNfV1wtKugrtcoCoa+jVdACMPbqfd6pFd4+XRldbgfer8AgN0f03innyO9i0eEl9TuzYp/4Ulx74DBpBDs6UCYd93pj64JUuO+lqI8eCOwdp79r8QbHjCxatUqHnjgAT744AN69erFqVOneOyxxwCYOXMmP/zwA++99x5LliyhVatWJCcns3//fgCWL19Ou3bteOyxx3j00fJrAwshqihhJ2Wbwbl01nIOMaCkn0YD7FGa0kVzFB+DGpRLFhq50qHELM6m5WFvq2XTi/3ZcPwim0+mMTg1GVKAwA7lLwqKggm/WJ5zLRu4i39JsBiYVkn/dgXeuKcNT/RvQmMf52umpd8MtfsgYRuauE1EBnRjw/GLnEzJUb8vzFV3GNu/BIA9xSuv9Wh85cYfDY8E7nru9ddf56WXXmLixIkARERE8Nprr/HPf/6TmTNnEh8fj7+/P7fddhu2traEhITQuXNnQN36U6fT4eLigr+/f20WQ4i6rWS5zhIVjBjPTzmFA7DZ1IYu2qPY5V1ATyFpuVoURSnX4vXbgSQaaS7yiesSvNPcuadjN+5p5w9zimvLZecoX405cF8ip8BAToEB57I17tgMANqWbb6uhL2trmpBG9Q+9uDOkLANUo7QxFedR33qYo661/eaV9VpcQA2Dnx/SQ3c5XfsangkcF+LraNa862tZ9+g3bt3s3PnTl5//XXzOaPRSH5+Pnl5edx3333MnTuXiIgIbr/9doYOHcqdd96JjY380xDCWkxxW9ECiYongZp08i6c5Mr/3YbU0wBkeLSFwlVQkEWwJoWTRY3IKzTiVGaEtqIo/HYgkcd1K2ibuxV+nwaPb4HUY2C4DHYuUDIw7VqKA3eQVp03npKVj3PKIfU5Pi3Mc7jbBl27xl1tvsWLr6QcoUlrNeCfvJANf/9HDdoeYdDhAS41G8Xfc9Wuhirt/FXPyU/na9Fo6tT8viuZTCZeffVV7rnnnnLf2dvbExwczLFjx4iJiWHNmjVMmTKFt99+m/Xr12NrW8m2dkI0UPFpeUxdupdujb14YXCLa18AYChAObcLgJ+MPXnCZgU5SScsA7fJhH2OutGFe6NmcCkMkg/Q1OYiJ4sakZZTaBG495/LJPFSLoP16n1JOawuk5pRvFlGQFvLgWlXUxy4A4oD98W0NCIy1Olk523DyMg7gK1OQ4uAcmur3Tjf4u1ALxyisU/xz9nsZLBPAY0OHt8Kdo5sOaAOwG3h74K3s76SmzUcMjitnuvYsSPHjh2jSZMm5V7a4v/YDg4ODB8+nA8++IC///6brVu3EhsbC4CdnR1GYzXWExainkpIz2Ps59vYE5/B5xvPUGgwVek6JXEfOlMBaYoLe+yiAbDJjLNMlJOMrVKAQdESEt4MPNXBXS306tzm1FzLAWq/H0gkSnMcb02ZBVO2zYekfepxVZvJwTw4zVPJwAYDBYlqbRtnP/amqQPlIgNc0dvoqn7PqvJpDmjgcjquxgx8XfS0KR4Vj08LsFN/vdlcvLpad+nfBqTGXe+9/PLL3HHHHQQHB3Pfffeh1Wo5cOAAsbGxzJ49m0WLFmE0GunSpQuOjo58/fXXODg4EBqqzsEMCwtjw4YNjBkzBr1ej7e3/McRDU9ixmXGfbGN8xmXASg0mDiYmEnHkGsv3BG3dy1hwF5aMGpwH/gD3IpSUAwFaGzU2qMh7TQ2QKLiRbtQH8gMA6CxjTodLK3MADWTSeH3A0k8otuhnmjUGc7tgGMr1R2uoNzAt6ty8gGtDVqTAR8yUVIuquctVkyrgWZyAFsHtUk//RSkHKaJrzNt4ooDd5lR8VtOlgRuaSYHqXHXe4MHD+a3334jJiaGTp060bVrV959911zYHZ3d+fzzz+nR48etG3blrVr1/Lrr7/i5aX+B5k1axZnz56lcePG+PhUcwF/IeqBrPwixn22lYT0y4R6ORIdqgbr3WcrXxilrEtHNwBgCOpC7/atuazYocPEubPHzWkuxh0F4JzGnwgfZ3X9bCBEowbu9DI17r0Jl0jMvMwQ3U71RM+p0Lg/oKgbcUD1atxaLbioi7BEauMIP/ezet63JccuqCO8WwbUUOCG0ubylCM09nGmtaY4cBf/8nE+4zJn0/LQaTV0KbMTWEMmNe56ZtKkSeVWShs8eDCDBw+uMP2IESMYMWJEpffr2rWreXqYEA3RV7+uZVHO06Q5+hA05n/8fFrDrrhL7IpL51GuPgDsVEoWoXmxoIG23W/HQW9Dgm0AwYY4jh85QHCTNgCknztGAJDrFIxOqzE3lfsbkwHLKWHbz6TTVnOaAE0a2DqpQVtrC6fWqQnsnMGrSfUK6RoImQnMs52LPsegbuXZdhSn96u/nJj7n2uCb0s4+ltxjXtIaVN58S8fm4tr220bueFiL+NuQGrcQghRqZNn47jj4DOEaS8QZTqI/3eD6Ger9gHvjruEco1Fkvbs3oGnJocCjZ7AyK4AFLmqrV0X44+a0xUVjyjXeRUvXFJc4/YsSkKLyaKpPDHjMkNKmsmbDlSbm5vcVjqK3L8aA9NKFA9Q02sMJOkC4dG15Pu0NXcNRFR1itf1KFPjbumci68mAyNa8Fd/qSlpJpf526UkcAshRAWUossYvh1LmCaZVBt/NSDmpdFs9QQm2K4lNafQvEd1ZXJPbgIg1a0N6NTaolNAUwCKLp7GZFIDv0O2OorbPaiZeqFbI3U9b6UIf9JJK9NUnnTpMrdriwN38TaZaLXQ63n1uFnFrWtXFRQFwJ/GTjysfxv8WnE2LRdFARd7G7yd7ap/z6oqOyXMcAKAU6ZACrX2FBpMrD+u9rl3byL92yUkcAshxJUUhcT/PUyLwkNkKY4UjV4CD8dAh/FoUHjR5jt0GNl1lX5ug9GER+puAGzDe5jPewerC4n4GZM4nJTF5UIj3gZ1ulOjiFZqIq3OPNAsVHvBosZtd+kY4doLGLV20HRQ6QM73A/TjkD3p6pf3m5Pcn7iDiYXTeV0tg5FUTh9MRdQa9s1utyxV2O1qb8wB/f41QDEKuHEpeWy7mgKl/KK8HXR0zlM+rdLSOAWQogrnFz9KUHnfqdI0bGq9X8JaNoBbO3hzvfB3h0nJY/WmjPsiqs8cMeez6S9ojaHe7XsYz6v82oMqAPP3l97gpe+24SXRl2X3CekeekNipvLgzUppOaU1rjDstS525cb9QL9FXOrXQPVoF9dGg1ejZoAGvKLTGTlGzh9UR2YVqP926C2RHirLQ2aw+pyrLGmcE6m5PDDbnVe+t0dgxr0bmBXksFpFbhWv5WoGvlzFLe6QoOJBxZsJynzMp3CPOkQ4sGO2CPMPvcqaGCB3f1MvGtM6QVaHYT1hKO/0UN7iDVx7Su99/4jR5mkTcGEFl1wmd29PEtHjMccTqaVJg70kK1zx8Xe1TLdKQjVXOCvXLXGnVdowNWQBjZg61vNAWjXYG+rw83BlszLRaRk5Ztr3FVewvRG+LWElENQqP6ycNAUhnI6jb+Oqc3k90U17N3AriSBu4ySlcLy8vJkK0wryMtT+/9kBTZxqzqclMWOM+qKYQnp51m+5zwf287FTZfHeYfm3Dv5TRzsrqjBhveBo7/RTXuITy7kkHm5CDeH8v/Gs4+r+29fcmmGV9mA7BaMotHiSAEjm9kxzDYTToGTf2PLGxTXuEM1KaTnFmIyKSRm5OONuuiK3s3PSn8Kpfxc9WReLuJCVgGnUoubym/GTlwlA9QABQ2HlTAO7EjAaFJoH+xOE98aWLWtDpPAXYZOp8Pd3Z2UFHXupKOjo2xleR0URSEvL4+UlBTc3d3R6WpgxSUhrOB4stpEHRngyoAWvmiP/caw9B0oGh1BE74AtwqCVoTa7N1Zdxy7oiL2xF+iX3NfiySFBhPuF3eDFjSh3Syvt7FD49oIMuP5b9d8WPEpANqWIyzTuQUB4Ku5hNGkkJVfRGLG5dLV0pwsn2kNfq72HL+QQ3JWvrmpvEZHlJcoGaAG5LpEkJdvD0Z1ZbqRUtsuRwL3FUp2wSoJ3uL6ubu7y65i4pZ27IIauLtGePJ8u0I4MB8ATc+p6nrfFfFuBs5+6HMu0EFzkt1nI8sF7v3nMuiA2r/t3rx3+Xt4hkFmPPz2LORnqFOfuj5umcZZrVH7azMAdS53UuZlWpYEbmfrB25fF3sADp7PJDvfgEYDoV43vtnRtR9cpsYd0A6KF2+zs9FyZ9ta2lb5FiaB+woajYaAgAB8fX0pKiqq7ezUWba2tlLTFjfNezHHiUvLZUznELqEe1a5pex4ceAeXLQOvpgNhnzwiYTe/6z8Io0GwntD7DK66w6xNa5HuSS7j8XxqEZdxUwb2rX8PTzC4cwGyEtVN9MY/pF5uphZceD2KW4aT8spIDEjn96a4q0unay/kqGfq7oE67bT6hrpwR6O2NvehP/HbiHqYjJFuTiGRWN7SEORUWFQSz/cHKWr7UoSuCuh0+kk8AhRB5xJzeX9ter835/3JRIZ4MozA5pye+trt/acTU7jdZsFdD2wVj3R5Da453N1BPnVlARu7UHmJWRQZDRhW2bUc/qxLeg0CtkOQbi4VlBjLB6gBqjTtypaW7w4cDuQjxOXScstJPFSHl7UXI3bz1Ut99HiLoSImh5RXkKrVbsgTqxG17gfrQIvsS8hgzGdQm7O8+sYCdxCiDot5rC6LKivi56s/CKOJGXx5Ld72PvywKsukZmVdJKPCmbQzuY0Cho0fadD7xeqtupYuNr83V57Cm1+HkeSsmjbyB2A/CIjrqm7QAtKcAW1bTCvCoZnY+j7UsVp9M7q8qWFOfhoMkjLKSAzIxU7TfFufY7WX0mspMZdIsL7JvRvl7j7U7UFwjOCD8fmEZeWR8+mslpaRWRinBCiTos5fAGAJ/s3Ydv0AXg52WEwlS4gUqGTa3BcNIB22tNk4ozmgR+g74tVXyrUIwzcQ7DFSCftMYuFWPbEXaJj8fxtl2a9Kr6+8QAY9RU8uFJdsrQyxbVqXzJIzSmkIEP9JcVg63LtVoHr4Otqec+bVuMGsHc1L9sa7OkoQfsqJHALIeqstJwCdhcvgnJbpB/ujnY09lVriadTcyq+KD8Lvp+ITUEG+0wRvBY0X20ir67iWnc37SFzHgA2HE2ig/YkAJqQbhVeikYDLe8Cl2s05zur3/toMknLLcCUrf6SotRA/zaUNpWXuKmBW1SZBG4hxC0t9lwmS3fGsz8hg/wio8V3a4+mYFKgVaArge5qzbVkpa9Ka9yZCVCYw2WdC6MKZxavGHYdwtVpYRN0MQw6NRvl7GbYu5g79z6Cg6aQQjt38Gl+9XtcS3GN20eTwamUXFyNGQDoXKw/hxvAx9myqbzJzZgKJqpN+riFELes/CIj477YRna+AQCdVkOfZj58OLYDTnobczP5wJalgaxkpa9KA3euuttUmsaDQmxpdr2LezQdhMknEseLR7jLtBYWqQPcWgEGRYux82S1Zn0jimvkvpoMfkzM5J7iqWBaF+sPTAN1+pWXkx1puYU4623wcdFf+yJx00mNWwhxy9p0IpXsfAOOdjq8nOwwmhTWHU3hhR/2c7nQyMYT6oTfsoG7pHn31MVKmsrz1MB9waCma+5/nYHbwR3tlK1Md/sPPxp7YdA5kOUYyn+KxjDZ52scbpt+ffctq6TGTQbZ+Qa8zIuv1ExTOZQ2l0f4OMkCVLcoqXELIW5Zfx5SB2ONig5m5p0t2X4mnfELtrMyNpns/F3kF5kIcnegZUDpkqIlI6HPpuViMilotVcEn+Ia9wWjCxoNNPG9geZgjQaHJr15bnMw+6JCScrMZ036BZ5v2ez671mWc2mNGzAvd1oTq6aV8HPVczjpJi11Kq6L1LiFELckg9HE2iNqU/igVn5oNBq6RnjxynB168uNJ9QAPLCln0XNsJGHA7Y6dZerxMzLABQZTYyav5X7v9iGIVtdFTFNcSXU88YXGIkO8wBg6+k0tpxS89S3uZUCa8kiLMU1bZ+SxVeca67GHeqlBuwWZX4ZErcWqXELIW5JO86mcymvCA9HW4u9mMd1DuFAQiZLd6lbPpZtJgew0WkJ9XLiZEoOpy7m0sjDkQPnMthxVt1M5GDRadoD6bjSzO/GN6+IDlUD98kUtWnex0VPq0ArBb3iQWi+GnXUek2uU17iiX5NaOzjxN0dZY3wW5XUuIUQt6RVB9Vm8tsi/Sz2YtZoNLx6Vytui/SjV1NvOod7lru2pJm3ZKOMbafTzd8lJ50DIE1xuf7+7TJ8Xe0J9iydi923mY/1+oaLa9yemmx0GEubymtg1bQSPi56xncLw1kv9bpblfzNCCFuOYqisLp4xPjgVuXnOtvb6vhiYnSl1zf2dYbDF8wjy0vW3vZwtMXDoDY3pyludLJCjRsgOtSThPTzgBWbyQEcvUCjQ6sY8SILL/M65bI4SUMmNW4hxC3nwLlMkjLzcbTTXdcKWuYad2oORUaTeWWzT+6PwlerrsOdjotVmsoBOhY3l+u0Guuu+KXVmUeQh2mScdQUqOdrsKlc3PokcAshbjklo8n7Nfet3uCx/Cz4djQD90/FjiJOX8zlwLlMLhcZ8XC0pUu4J0F2ai08V+dOuJVGTg9o4YuLvQ13tg3AzcHKu1kV93NHauMBUGwd1XXMRYMlTeVCiFvO6uLAPahVNVYIMxrghwfh5BrcgSdsXHgv8z7WHVWb3LuEe6FVjNgVZgDwr1G9sbOxTt0l0N2BfS8PokZmPRf3c7cs3iZUU4NzuEXdIDVuIcQtJSUrn1MXc9FortJfbDLC1k9gzSuQchQUBf54AU6uAa1aH5lis4IWmniW7VIHo3WN8ITLJYPUNHRt1dSq+dZpNeXnjFuDc0mNO674szSTN3RS4xZC3FJKpm1F+ruqzc7ndkPaCWh2Ozi4Q24aLH8UThXvob3pPfBuDqnHAA3c9z/Y/x22R3/jTdvPuCd7FqClS4QX5KpBHAcPtf+4LigO3M01xXmXGneDJ4FbCFFr5vxxhIPnM/lsfDROxdOPdpxRA3fncE8ozIOvR0BBFtjYQ+RwiN+qbhRi4wBhPeDUX8VBGxj8OkTeAUFRXD7+N+05zYO6P/hRP4Lmfi5wVl0itU4Fv+LArdcUqZ/rUt5Fjaj1pvJPPvmE8PBw7O3tiYqKYuPGjVdN/80339CuXTscHR0JCAjgwQcfJC0t7SblVghhLfFpeXy6/jSbT6axqrhPG64I3KfWqkEbDRjyIfZ7NWh7RsCja+GBH2HaERj8Bgz/ELpOUW/iGsCOps8C8KzNj3QPc1GbsYvXKa9T06mu3AlMmsobvFoN3EuXLmXq1KnMmDGDvXv30qtXL4YMGUJ8fHyF6Tdt2sSECRN4+OGHOXToEMuWLWPnzp088sgjNznnQogb9e2O0v/nfxQvtpKZV8SxC+p0rU5hnnB4hZqg6xR4ZB1EP6y+Hvsb/NSlT3Hxg25PQMcJFrtxFbZ9gDxFj7Mmn/5++erJ3OJf8h29arRsVuV8ReCWqWANXq0G7nfffZeHH36YRx55hMjISObOnUtwcDDz5s2rMP22bdsICwvj6aefJjw8nJ49e/KPf/yDXbt23eScCyFKfL31LGM+28ql3MIqX1NgMLKseMlSgA3HL5JbYGBXXDqKos7D9nHQwPE/1QQth0OjKLjjXfVl73bNZ0T4OhOnqEGvs1uGejK37jaVl36uQ3kXNaLWAndhYSG7d+9m0KBBFucHDRrEli1bKryme/funDt3jpUrV6IoChcuXOCHH35g2LBhlT6noKCArKwsi5cQwjryi4y89ecxtp1O59cDiVW+btWhC6TlFuLnqifE05ECg4m/jqWYB6Z1CvOEM+vVZnJnP2jUudp5C/V0JMMhGIBGpuK81cWm8nI1bgncDV2tBe7U1FSMRiN+fpb/KP38/EhOTq7wmu7du/PNN98wevRo7Ozs8Pf3x93dnQ8//LDS58yZMwc3NzfzKzg42KrlEKIh+/tYCtkFBqB0WdGq+Ha7OrVpTKcQhrRRlzT982CyZf/24V/UxC3uAG31f1TZ6LR066QGfO2l0+rJ4i09caxDgdvOEfRlNi2RpvIGr9YHp125GL+iKJUu0H/48GGefvppXn75ZXbv3s2ff/7JmTNnmDx5cqX3nz59OpmZmeZXQkJCpWmFENXzy77SWva20+mYTMo1rzmZksO20+loNTCmczBDWgcAsO5oCrHn1E00Ooe6wrGV6gUth19/Bj0bq+/pVwTuulTjBssBadJU3uDV2nQwb29vdDpdudp1SkpKuVp4iTlz5tCjRw9eeOEFANq2bYuTkxO9evVi9uzZBAQElLtGr9ej1+utXwAhGris/CLWHlX3ttZpNaTnFnIiJafCHbd2nEnnz4PJKCgcOq92V/Vv4UeAmwP+rvYEuNmTlKkOIAtws6dR1l7IS1PnW4f2uP5Mekao72mn1Pe62FQO4OwPaSdBZwf27rWdG1HLaq3GbWdnR1RUFDExMRbnY2Ji6N69e4XX5OXlob2iyUynUxdRUJRr/6YvhLCe1YcuUGgw0cTXme6N1VHaW0+llkuXV2jgsa938eXmMyzcfNbcj31/1xBAbXUruwNYpzBPNEd+VT80Hwa6G1j726u4xp2ZAIbCutlUDqU1bicfi5HzomGq1QVYpk2bxvjx44mOjqZbt2589tlnxMfHm5u+p0+fzvnz5/nqq68AuPPOO3n00UeZN28egwcPJikpialTp9K5c2cCAwNrsyhCNDi/7FO3sRzeLhCdVsPGE6lsO53OpB7hFum+35lARl4RgW723N0xCBSFUDcb+jYrbfId0tqfRVvOAsX921v+UL+4kWZyUAd22TpBUS6knypd8rSuDfByKf7Fpq61FIgaUauBe/To0aSlpTFr1iySkpJo3bo1K1euJDQ0FICkpCSLOd2TJk0iOzubjz76iOeeew53d3f69+/Pf/7zn9oqghAN0sXsAjafVGuvw9sFklY8FWzbmTRMJsW8ZrfBaOKLTWcAeLxfE8Z3DYWlD8C+DdBmt7m/NjrMkyB3By5k5dMrSANZxct7hlbc+lZlGo3aXH4hFs6VTBvVgKPnjd33ZjPXuGVgmrgFljydMmUKU6ZMqfC7RYsWlTv31FNP8dRTT9VwroQQV/P7gURMCrQLdifM24kgDwcc7XRkFC+gEhmgjoJeeTCZc5cu4+Vkx31RjcBkghNrwHAZkvZB04GA2ke+5LGupOcWEmo8qD7EPQT0Vtgv26s4cCdsUz/XpXXKSzQdBLu+hFZ313ZOxC2g1keVCyHqnpWx6qDS4e3ULipbnVadew1sPaVOC1MUhU/Xq4PCJnQLU/fVzk5Sgzao/c5lBHs60i7YHVKOqCd8W1knsyUjyxN2qO91rZkc1FXipsZCh/trOyfiFiCBWwhRLVn5ReyOvwTAoJalM0C6RqgD1Ermc285lcahxCwcbHVM6KZ2f5F+qvRGmecqfkDKYfXdN9I6GS4ZoJZ6XH2XfmJRx9V6U7kQom7ZcjINo0kh3NuJYE9H8/lujUsD92u/HTbP8R4V3QgPJzs1Ucl8arhK4C6pcbe0ToZLpoSVqEvrlAtRAQncQohq2XhCXe+7d1PLmmvrQFec9TZk5RtYUDwgLdDNnn/0aVyaKK1MjTujgsWQFAUuWLnG7dnY8nNdbCoXogwJ3EKIKlMUhQ3FgbtXU8sAaKPT8kDXUL7flUCvpt7c0TaQXk291b7tEteqcWclQkEmaHTg3dQ6mXb2BTtnKMxRP0tTuajjJHALISq1PyGDmMMXeLxvY5z0NsSl5ZGQfhlbncbcNF7WS0Na8NKQFpXfsGyNO+s8mIyWI7xLmsm9moCNlVY81GjAMxySY9XPdW3xFSGuIIFbCFGh1JwCHlq0k7TcQnIKDLwyvJW5th0V6oGTvpo/PkwmuHSm9LNihOxkcAsqPWftgWklPBuXBm6pcYs6TkaVCyHKURSF6ctjzQurfL0tjqPJWWw4ri66cmUzeZVkJ4IhH7Q24FocrK9sLrf2wLQSXmX6uSVwizpOArcQopwfdp8j5vAFbHUaokM9MJoUXv75kHkt8j7NriNwlzSTu4eCR5h6fMVc7hqtcZeQpnJRx0ngFkJYOHcpj1d/VQPotIHNmTumPfa2WnacTSe30IiXkx0tA1yvcZcKlAxM82oMbo3U47I1bpMRLh5Tj/2stPhKibJTwmRUuajjJHALIcwy8gp57Kvd5BQYiAr14LHeETTycGRK3ybmND2bepvXIq+WksVXPCMqDtyXzqqrqtnYl9bIrcW7mTpS3dZJXfJUiDpMArcQAoDMvCIeWLCdw0lZeDvb8d6o9uiKA/RjvSMI9nQAoF9zXyjIgdSTV79hfhaklxmMllZc4/aspMZd0r/t09z6a4k7ecHY79SXTsbkirpN/gULIcjKL2LCl9s5eD4LLyc7vn20KyFepaui2dvqWPxwFzafTFPXJ182AY6sgPsWVb7xxbKJcGYDPLQKGkWX1ri9ItSFVqDiwG3tgWklmg2umfsKcZNJjVuIBi4tp4D7P9/O/nOZeDja8s2jXWjmV35XrlAvJ8Z1CUFblAvH/1RP/v485KWXv2lBNpz+G0wG2PqROhWspPZtUeMuMzitpgamCVHPSOAWogE7dymP++ZvJfZ8Jp5Odix+pAst/K8x8OzMBjCq08TIS4U/p1dw412gmNTjwyvg3E4wFoDWFtyCS6eD5WeoQR7KBO4aqnELUU9I4BaigTp1MYeR87ZyOjWXIHcHfpjcjVaBbte+8GSM+h7SHdDAgSXqHttlxW8rPVaMsHqGeuwRqvYx27uCffGzMs+rr4tH1fsFtLvRoglRr0ngFqIBUhSFF5btJzkrn6a+zvzweDcifJyrciGcKA7cPZ+Fro+rx79NVQeslYjfqr437q++n9upvpedT+0WrL5nnoOjv6nHwV3UtcWFEJWSwC1EPWc0KcSn5VmcW3/8InviM7C31bL4kS4EuDlU7WYXj6n90jb2ENYT+v8fuIeo5w4sKX5gkdpUDnDbq+AWUnp92RXMyvZzH16hHrccfh0lFKJhkcAtRD03+/fD9H77L+auOQ6ote331pwA4IEuofi52lf9ZidWq+9hPcHOEeycoMtk9dz+4sCdHAtFuWpTuF9r6PxI6fVlF0IpCdyJeyF+i3oceWd1iydEgyOBW4h67MSFbP635SwAc9ec4LcDifx97CL7E9TatsVe2VVR0r/dZGDpudYj1cVNzu1U53aX9G8HdwWtFjqMB5viGr1X6UIu5sB94Ht1IFtAe7X2LoS4KpnHLUQ99sbKI5gUcHe0JSOviOeX7SewuFl8YrcwfFyqsXVmQTbEFfddNy0TuF38oMkAtTa+/ztILV62NKSr+u7oCSM+hoSdEN679LqSPm7DZfVdmsmFqBKpcQtRT206kcpfxy5io9Xww+Tu9G3uQ36RidOpuTja6Xisd8S1b1LW6fVgKlKbu72uqKm3G6u+H1haWuMO6Vb6fet7YcibliuilQTuEpF3VS8/QjRQEriFqIeMJoXZv6vzoh/oGkoTX2c+GNuBJr7qyPEJ3cLwcq5GbRtK+7ebDir/XfMhoHdTB5rlXgSdHQR2uPr9SprKAXwiwbtJ5WmFEGbSVC5EPfTT3vMcTc7G1d6GZwY0BcDV3pZvH+3CuiMp3N0xqHo3VBQ4WTxXu2z/dglbB2g1Avb8T/0c2BFsrzHozcVf7RtXjNJMLkQ1SI1biHrox93qGuD/6NMYDyc783lfF3vGdA5Bb1PNTTxSDkPWeXWQWViPitO0H1d6HNqt4jRlaXUQ0FZdTa3VPdXLjxANmNS4hahnsvOL2BWnrh8+tE2AdW5asuhKeC+1dl2R4C7qAivppywHoV3N2CXqWue+LayTTyEaAAncQtQzm0+mUWRUCPNyJNzbyTo3LQncFfVvl9BoYNxSSNoPEf2qdl8Xf/UlhKgyCdxC1DPrj6cA0Ld5NZcOLcxTF1W5Un4mJBSPFG9y29Xv4d1UfQkhaoz0cQtRjyiKwl9HLwLQt7lP1S/86w14Mxj2fVv+u5LtOb2agGe4dTIqhLhuEriFqOPyCg3m42MXsknOykdvo6VrhFfVb3LwRzU4r3hKna9dVlWayYUQN40EbiHqgPwiI78fSGLLyVQSMy6TkVfIV1vPMuyDjbR8eRWfrj8FwN/H1Np298Ze2NtWceR4zkVIO6kemwywdLy6mQhcMQ3sGs3kQoibQvq4hbjFmUwKzyzZy6pDFypNM+ePo4R7O/HXUbV/u1+LavRvl/RfezcDB0/18zcjYeh/1S02s5PA1hFCK5kGJoS4qSRwC3GLm7f+FKsOXcBWp6GRhyMJ6XkYTArN/VwY3SmYEyk5fLcjnqlL91FoMAHQt1k1AnfJEqVhvaDfDFhwG6Sfhm9Hga54dbXw3tdeUEUIcVNI4BbiFvb3sRT+u1pttp51V2vGdg6hyGgiI68Ib2c7NBoNRUYT8em5bD6ZBkCEjxMhXhWMDq9MfPHGISHdwMkLJq2ELR+o23ReVueDS/+2ELcO6eMW4hYVn5bH09/tRVFgbOdgxnZWt7y01WnxcdGj0WjMnz8e15Gw4mDdrzrTwApz1XnXULqbl2sA3D4HnjsK9y2Cga9BxwnWKpYQ4gZJjVuIW5DBaGLq0r1k5RtoH+zOK8NbXTW9u6Mdix/pwg+7zzGhW1jVH3RulzogzbURuF+xW5eNHlrdXf3MCyFqlARuIW5Bn244zZ74DFz0Nnw0rkOV1hZv5OHI1NuaVe9B5i04u15HLoUQtUGayoW4xRxKzGTumuMAzBzeikYe1eivri5z/7YEbiHqCgncQtxC8ouMTFu6nyKjwqCWftxb3e03ryVhByx7EBL3gtEA53aq50OqsJuXEOKWIE3lQtxC5v19imMXsvF2tuONe9qYB6BZzd9vwqm1cPQ36PQIFOaA3g18W1r3OUKIGiM1biFuEeczLjO/eAW0V4a3wttZb90HmIxqjRvAWAjbPlGPQ7qAVn4UCFFXVPt/a1hYGLNmzSI+Pr4m8iNEg3DiQjb3f7GNpTtL/x/954+jFBhMdA7zZJi19tEu68IhKMwGOxfo/2/QFP/3l/5tIeqUagfu5557jl9++YWIiAgGDhzIkiVLKCgoqIm8CVEvGU0Kzy3bz+aTabz4Yyxz1xxnd1w6K/YnotHAy3e2tH4TOZSOIA/uDL2fh4m/QdcpEPWg9Z8lhKgx1Q7cTz31FLt372b37t20bNmSp59+moCAAJ588kn27NlTE3kUol75bkc8B85lYqdT//vNXXOCSV+qg8RGRQXTOsitZh5cdoU0gLAe6kIrjp418zwhRI247o6tdu3a8f7773P+/HlmzpzJF198QadOnWjXrh1ffvkliqJYM59C1AupOQW89edRAGYMi2TWXa3QaCC7wICz3obnBzevmQcrikz9EqKeuO5R5UVFRfz0008sXLiQmJgYunbtysMPP0xiYiIzZsxgzZo1fPvtt9bMqxB13pt/HCUr30CrQFce6BqKTqvB3dGOuTHHeWpAE3xcrDwgrURGvLrLl9YGgqJq5hlCiJui2oF7z549LFy4kO+++w6dTsf48eN57733aNGihTnNoEGD6N27t1UzKkRdtzsunR92n0OjgdkjWqPTqv3Yw9sFMrxdYM0+vKR/O6A92NXggi5CiBpX7cDdqVMnBg4cyLx58xgxYgS2trbl0rRs2ZIxY8ZYJYNC1Bf/XaWuhjY6OpgOIR7WuWluGhz8EdqMtOyrzr4Ax36H1veCvZs0kwtRj1Q7cJ8+fZrQ0NCrpnFycmLhwoXXnSkh6putp9LYejoNO52Wpwc0td6NN70LWz+Cfd/AgyvBzgny0mHRMEg7AfuXwoRfyqxJLiukCVHXVXtwWkpKCtu3by93fvv27ezatcsqmRKiPlEUhfeK1x4f3SmYQHcH6938fPFMjqR98OOjUHQZlo5XgzZAwjb44UG4eET9LDVuIeq8agfuJ554goSEhHLnz58/zxNPPGGVTAlRn2w9lcaOM+nY6bRM6dfYejdWFHVRFQA0atP4x50hbpO6yMrQ/6qD0Y6tVJN4NwMnb+s9XwhRK6oduA8fPkzHjh3Lne/QoQOHDx+2SqaEqC/K1rbHdg4mwM2Kte2MeCjIBK0t3P1p6TmNDkYtgs6Pwp3vl6aX2rYQ9UK1+7j1ej0XLlwgIiLC4nxSUhI2NrJniRAJ6XnM/v0wJ1JyuJhVQHaBATsbLVP6NbHugy4cVN99WkC70ZCXBhvfgYGvQpPb1O86PADZybDlA2g72rrPF0LUimrXuAcOHMj06dPJzMw0n8vIyOBf//oXAwcOtGrmhKhrLmYXMH7BdlYdusDpi7lkFxgAeLxPY/xc7a//xqkn4YeH4EKZVq3kWPXdv7X63m0KvHBSDdZl9X4eXoyDsJ7X/3whxC2j2lXkd955h969exMaGkqHDh0A2LdvH35+fnz99ddWz6AQdUVWfhGTFu7gbFoejTwcmHNPGwLcHPBz1eNiX37aZJUpCqx4CuK3gLEIRhf/PysJ3H6tS9NWtsZ5Tax9LoSoFdWucQcFBXHgwAHeeustWrZsSVRUFO+//z6xsbEEBwdXOwOffPIJ4eHh2NvbExUVxcaNG6+avqCggBkzZhAaGoper6dx48Z8+eWX1X6uENaUX2Tk0f/t4lBiFt7Odnz9cBd6NfWhia/zjQVtgNN/qUEb4PTfavCG0qZy/zY3dn8hRJ1yXZ3STk5OPPbYYzf88KVLlzJ16lQ++eQTevTowaeffsqQIUM4fPgwISEhFV4zatQoLly4wIIFC2jSpAkpKSkYDIYbzosQ18tgNPH0d3vZfiYdZ70Nix7sTLi3k3Vuriiwbnbp54IsSNgO/m3h0ln1nARuIRqU6x5NdvjwYeLj4yksLLQ4P3z48Crf49133+Xhhx/mkUceAWDu3LmsWrWKefPmMWfOnHLp//zzT9avX8/p06fx9FRXiQoLC7veIghxwxRFYcZPB1l9+AJ2Nlo+nxBt3d29jv8J53eDraM6KvzUOjgRo44cB3AJlN29hGhgrmvltLvvvpvY2Fg0Go15F7CS/YONRmOV7lNYWMju3bt56aWXLM4PGjSILVu2VHjNihUriI6O5q233uLrr7/GycmJ4cOH89prr+HgUPE0m4KCAov9wrOysqqUPyGupcho4r+rj7F0VwJaDXw4tgPdGntZ7wEmE6x7XT3u8g/wbaUG7pNrwK2Ret6/deXXCyHqpWoH7meeeYbw8HDWrFlDREQEO3bsIC0tjeeee47//ve/Vb5PamoqRqMRPz8/i/N+fn4kJydXeM3p06fZtGkT9vb2/PTTT6SmpjJlyhTS09Mr7eeeM2cOr776atULKMRVKIrCbweS+ONgEhuPp5pHjc+5pw2DW/lb80Hq1K4LsaB3he5PF3+hUfu2T6xWP0ozuRANTrUD99atW1m3bh0+Pj5otVq0Wi09e/Zkzpw5PP300+zdu7da99NcMdpVUZRy50qYTCY0Gg3ffPMNbm5qc+S7777LyJEj+fjjjyusdU+fPp1p06aZP2dlZV3XIDohABZsOsPs34+YP3s52TF1YDNGd6p4TEaVpJ+B354F35bQcTy4BsEvT8CRFer3vZ4rbQ5vFA3ndpYGbj+pcQvR0FQ7cBuNRpydnQHw9vYmMTGR5s2bExoayrFjx6p8H29vb3Q6XbnadUpKSrlaeImAgACCgoLMQRsgMjISRVE4d+4cTZuW37xBr9ej19fQHseiQTmUmMlbf6r/xid0C+Wejo1oG+SGVnuDU622fqyOHD/9F2z7GPRupSui3T4HOj1SmrbpIDVwl5AatxANTrWng7Vu3ZoDBw4A0KVLF9566y02b97MrFmzyq2mdjV2dnZERUURExNjcT4mJobu3btXeE2PHj1ITEwkJyfHfO748eNotVoaNWpU3aIIUWWXC408s2QfhUYTA1v68erwVrQPdr/xoK0ocLL4/0BgR3Vt8YJMtdb90J/qsqVlW6BKVkQDsHEAz6r/nxNC1A/VDtz/93//h8lkAmD27NnExcXRq1cvVq5cyQcffFCte02bNo0vvviCL7/8kiNHjvDss88SHx/P5MmTAbWZe8KECeb048aNw8vLiwcffJDDhw+zYcMGXnjhBR566KFKB6cJYQ1vrDzCyZQcfFz0/OfetpV251Rb2kl1WpfODib+CtOOwL0LYPImtVn8SgHtwclHPfZrBVqddfIhhKgzqt1UPnjwYPNxREQEhw8fJj09HQ8Pj2r/MBs9ejRpaWnMmjWLpKQkWrduzcqVK837fSclJREfH29O7+zsTExMDE899RTR0dF4eXkxatQoZs+eXdkjhLhh20+n8fW2OADeHdUOTyc76938RHFtO7Q76J3VV5uRlafXatVa9/7vpJlciAZKo5TM56oCg8GAvb09+/bto3XrujkoJisrCzc3NzIzM3F1da3t7IhbnKIojP5sGzvOpDO2cwhz7rFysPxqhNq3Peh16P5k1a7JiIf1/4HeL4BHmHXzI4SoFdWJTdWqcdvY2BAaGlrludpC1HVl99J+eoCVd/cqzIW4zepx00FVv849BO762Lp5EULUGdfVxz19+nTS09NrIj9C3DJqdC9tgDMbwVioBmLv8jMihBCiItXu4/7ggw84efIkgYGBhIaG4uRkuSbznj17rJY5IWrT5pNp7Dx7qWb20obSudhNB8nuXUKIKqt24B4xYkQNZEOIW0vZ2va4ziE3tpd2xQ8onQbWRPaxF0JUXbUD98yZM2siH0LcUv4+dpHdcZfQ22iZ0rex9R+QtE8dZKbTQ3gv699fCFFvXffuYELUVwajiddXqsuaTuwehq+1atuKAsdWwu5F6kYhAGE9wc5KW4AKIRqEagdurVZ71fnaMuJc3OpyCgxMXbKPdo3ceGpA+UFhS3YmcDIlBw9HW56wZt/22ldh03uln0O6w9C3rXd/IUSDUO3A/dNPP1l8LioqYu/evfzvf/+TXbhEnfDV1rOsOXKBdUcvMKJDEMGejubvsvOLeC9G7dt+ZkBT3BxsrfPQ3YtKg3bXKRD9MHjXwIA3IUS9V+3Afdddd5U7N3LkSFq1asXSpUt5+OGHrZIxIWrC5UIjCzaeAcCkwNfb4vjX0Ejz9/P+PkVabiER3k7c3zXUOg89tQ5+K96hrs+L0O9f1rmvEKJBqvY87sp06dKFNWvWWOt2QtSIJTvjScstxN5W/ae/ZEc8eYXqntpnU3NZsEkN6i8NaYGtroL/HgU5sPQBiP2hag+8cBi+nwiKEdqOhr7TrVIOIUTDZZXAffnyZT788EPZoUvc0goNJj7bcBqAGcNaEurlSFa+geV7zlNoMPHUd3spMJjoFuHFwJYVby3L8T/hyK8Q87I62OxqspPh21FQkAWhPWD4hzJfWwhxw6rdVH7lZiKKopCdnY2joyOLFy+2auaEsKble86RlJmPn6ueUdGNKDSYeO23wyzacpbTF3OJPZ+Ju6Mt74xqV/kAzMwE9T3rPKSfBq9KpooV5sJ3Y9T0Xk1g9GKwkX3hhRA3rtqB+7333rP4oabVavHx8aFLly54eHhYNXNCWIvBaGLe+lMAPNorAr2NjvuiG/Hu6mOcTMnhZIq6x/vbI9sR6H6VpU0zz5Uen1lfceA2GWH5Y5C4Fxy9YNz34OhpzeIIIRqwagfuSZMm1UA2hKhZ8/4+RVxaHp5OdozrEgKAq70tI6Ma8b+t6padD/YIq7yJvIRF4N4A0Q+VT7P1Izj6m7q4yphvK6+VCyHEdah2H/fChQtZtmxZufPLli3jf//7n1UyJYQ1HTyfyftrTwDw7zsicbQr/X31wR7hONnpiAr14KUhLa59M4vAvRFMJsvv007BX2+ox0PfhpCuN5p9IYSwUO3A/eabb+Lt7V3uvK+vL2+88YZVMiWEteQXGXl26T4MJoWhbfwZ0T7I4vswbye2z7iNJY91RW+ju/YNS/q4AfJSIeVw6WdFgV+fAUM+RPSFjhOsUwghhCij2oE7Li6O8PDwcudDQ0OJj4+3SqaEsJb/rjrGiZQcvJ31zB7RpsJBZ856m4qnfl0pPwvyM9Xj4OKa9JkNpd/v+R+c3Qi2jnDHXBlBLoSoEdUO3L6+vhw4cKDc+f379+Pl5WWVTAlhDacu5rBgszov+62RbfB0sruxG2adV98dPKD5EPW4JHBnnoPVL6vH/f8PPMv/ciuEENZQ7cA9ZswYnn76af766y+MRiNGo5F169bxzDPPMGbMmJrIoxDX5ftdCSgK9G/hS/8W1xh0VhUl/dtujSCij3octxkuX4JvR0NBJgRFQZfJN/4sIYSoRLVHlc+ePZu4uDgGDBiAjY16uclkYsKECdLHLW4ZRUYTP+5Wa8ijOwVb56Yl/dtuweDfFuzd1KbzL4fAxSPg5Av3LQJtFfrKhRDiOlU7cNvZ2bF06VJmz57Nvn37cHBwoE2bNoSGWmldZyGsYP2xi6TmFODtbEf/Fr7WuWlGSeBupAbnsF7qtK+LR8DGAcYtAfcQ6zxLCCEqcd37cTdt2pSmTctviSjEreD7XWqQvbtDUNUGnlVF2aZygPDeauBGA/d+oTaTCyFEDav2T7SRI0fy5ptvljv/9ttvc99991klU0LciIvZBaw7mgLAfdFWaiaH8oG77WiIvBPung+Rd1jvOUIIcRXVDtzr169n2LBh5c7ffvvtbNiwoYIrhLi5ft57HoNJoX2wO838XKx3Y3PgLv5lwMFdXYO8nQzKFELcPNUO3Dk5OdjZlZ9WY2trS1ZWllUyJcT1upRbyJKd6noCo26ktn3hMHx9N+z4XP1sMpZOB3OTXfCEELWn2oG7devWLF26tNz5JUuW0LJlS6tkSojqOnUxh+nLY+n25lpOXczFwVbHHe0Cru9mB5bBFwPg1DpY9xoYDeoWnYoRtDbgbIWpZUIIcZ2qPTjt3//+N/feey+nTp2if//+AKxdu5Zvv/2WH374weoZFOJa9idkMOrTrRQY1HXDWwW6Mn1IJK72ttW7kckIf74EOz4rPZefCed2gqb4d1zXQJnuJYSoVdUO3MOHD+fnn3/mjTfe4IcffsDBwYF27dqxbt06XF1dayKPQgBgMin83y8HURT419AWuNjbkppTwOTFuykwmOgU5sELg1vQKcyj8v20r2bVv0qDdu8X1A1DDi2HkzHgW9ya5GbFwW5CCHEdrms62LBhw8wD1DIyMvjmm2+YOnUq+/fvx2g0WjWDQpTYFXeJb7er/dfbT6fx4bgOzPr1MEmZ+UT4OPHlpE64VLeWXWLbfNg+Xz2+dwG0GQn7l6qB+8Rq0Bf/Uir920KIWnbd87jXrVvHl19+yfLlywkNDeXee+9lwYIF1sybEBZ+2lu6pebp1FyGfbAJUDcJ+Wx89PUH7WN/wKrp6vFtr6hBG6DJAEADybHgXrzAkARuIUQtq1bgPnfuHIsWLeLLL78kNzeXUaNGUVRUxI8//igD00SNyi8y8tuBJAA+GteBJTsS2HQyFYB3RrWjia/z9d04KxF+eAgUE3ScCD2mln7n5A1BHeH8bjj+p3pOArcQopZVOXAPHTqUTZs2cccdd/Dhhx9y++23o9PpmD9/fk3mTwgA1h1NITvfQICbPUNbBzCkdQDf70rAw9GWwa38r//GB5dDUR4EtIdh75TfirPJQDVwmwzqZ+njFkLUsioH7tWrV/P000/z+OOPy1Kn4qb7aa86h/qu9kFotWpwHdvZCuuCH1mhvre/H3QVNLU3HQjry6wUKDVuIUQtq/I87o0bN5KdnU10dDRdunTho48+4uLFizWZNyEAdVGVv4+pS5je0zHIejfOToaE7epxZUuWBnYAxzL7zLta8flCCHEdqhy4u3Xrxueff05SUhL/+Mc/WLJkCUFBQZhMJmJiYsjOzq7JfIoGSFEUAH47kEiRUaFVoKt1lzA98qv63qiTOj+7IlodNB6gHtu7gb1MeRRC1K5qr5zm6OjIQw89xKZNm4iNjeW5557jzTffxNfXl+HDh9dEHkUDk19k5K6PNhH58p/cPncDH6w7Cag7fV03RYHEfZB9ofRcSTN55DX+3TYbrL57hF//84UQwkquezoYQPPmzXnrrbeYM2cOv/76K19++aW18iUaIkMBoGHBpnj2n8sE4Giy2pJjq9MwvH0lteIrFeapo8UBTEXqiPC9iyHtJDj5wCNrwc4Zzm5W00TeefX7tbobMhMgtOd1FEoIIaxLo5S0RzYQWVlZuLm5kZmZKSu93UqMBvi0N4aCXKIvvUZGoY5/39GSCG8nzqbl0tzPhe5NvK99n+RY+OouyEurPI13c+g4AVbPAP82MHmT9cohhBDXoTqx6YZq3EJYzfldkHIIGyCi6BTG4M482D3MPIK8SrIS4ZtRatC2cQCb4l3sfCKhwwMQ0hX+NxxSj6lBGyDyLqsXRQghapIEbnFrOBFjPuygPcmwOydVL2gX5MC3oyA7EXxawEOr1P2yrzRuKSwcAoU56ueWMi5DCFG3VHtwmhA1QTmx2nx8h+d5OoZ4VP1io0Fd/Sw5Vu3DHvd9xUEbIKAtjFyobs8ZFAU+zW8s40IIcZNJjVvUOiUrCU3yAfPnNpoT1bhYUbfiPLEKbOxh7FLwCL36Nc0GwTP7SzcOEUKIOkRq3KLWrfx5MQAnlUAUNNhknYOclKpdvH0+7Pwc0MA9n0OjqKpd59ZI5mQLIeokqXGLm+5kSjY7z14it8DA6dRcup+IAR0UNBuOJuNvuHgUzu2CFkOvfqOjv8Ofxbt6DXpN+quFEA2CBG5xU+UVGhg5fysZeUUA6DDykj4WgFa9R8LuHDVwn79G4D6/B358BFAg+iHo9uRNyL0QQtQ+aSoXN9WKfYlk5BXh7WzHXe0DebFVFq6aPBQHT3ULzZKm7nO7Kr9JRjx8N0bd1avJbTDk7fK7egkhRD0lNW5xU327PY7/2HxGP4cUfEPGQfppADRNBqjrgjfqpCZM3Asmk7ry2fcTIS8V2o6GZrfDt6Mh5wL4tVZHiOvkn7EQouGQn3ji+uxdDOd2wu3/AVv7Kl1y4FwGNom7GK3/G7KB1f9X+mWTgeq7TyTYOkJBFqQeh0M/wfE/1O/O7YSVz6vHzv7qnGwZYCaEaGCkqVxUX2EerHwBdi+Cfd9U+bJvt8fzoM2f6oeQbhDYUT3Wu6pN3qDWngM7qMd7v4aN76jHUQ+Cb0v12NZJDdqyN7YQogGSGreovlNr1f5lgO2fqoPDKuhjLjSYKDAYcbG3JSu/iG37Ypmt3aF+OfRtdZ3wi8dBZwtOZfa8DoqCuM2w9SP1c/OhcMd76vGFg6B3AY+wmiufEELcwiRwi+or2cca1HW/T/8NjfuZTxlNCkt2xvPO6uNcyiukd1Mf/Fz13KusxkZjQgntgca/jZrYp1n5+zeKLj3Wu8Kwd0p/MSi5TgghGigJ3KJ6DIVwrLi5Oyhanba1/VNz4I49l8mLPx7gcFKW+ZL1xy+ip5At+rUAaLpMvvozgsoE7oGvgmsVt/MUQogGQAK3qJ4z66EgE5z9YMQ8+LiTut91+mkKXcN4cNFOUnMKcLW34dmBzejV1Jvle86Tv/MrvIzZmFwboW1+jYVV3IKg73QozIWOk25KsYQQoq6QwC2q58gK9b3FHWozd5OBcDIGdnzOukbPkJpTgK+Lnj+e7onXykcg5lf+WeZybedHqzZ9q+9LNZJ9IYSo62RUuag6o0FdZhRKlxftWtzsvXcxv+08BsA9HRvhlRFr2RcO4BIIHSfcpMwKIUT9JDVuUXXxWyEvDRw8ILSnei6iP3g3g9TjRJz6GribkVFBsGGq+n3b0TD4DfXY3k0dQS6EEOK61XqN+5NPPiE8PBx7e3uioqLYuHFjla7bvHkzNjY2tG/fvmYzKEqVNJM3H1ba3K3VQp8XAXhE9zs9gnQ0sc+Bwz+r33d7Apy81ZcEbSGEuGG1GriXLl3K1KlTmTFjBnv37qVXr14MGTKE+Pj4q16XmZnJhAkTGDBgwE3KqQDUQWgAkXdanFZa3c0ZbSiumjxecouBXV+CyQAh3SGgXS1kVAgh6q9aDdzvvvsuDz/8MI888giRkZHMnTuX4OBg5s2bd9Xr/vGPfzBu3Di6det2k3IqyDyvbu6h0UFYD4uvDiXlMCf/HgBaJ3wHuxaoX3T5x83OpRBC1Hu1FrgLCwvZvXs3gwYNsjg/aNAgtmzZUul1Cxcu5NSpU8ycObNKzykoKCArK8viJa5Dwjb13b+NunJZGd/vSmC1KZo4fXM0RblqP7hrkDryXAghhFXVWuBOTU3FaDTi5+dncd7Pz4/k5OQKrzlx4gQvvfQS33zzDTY2VRtXN2fOHNzc3Myv4ODgG857gxS3VX0PKW3lKDAYmfXrYb7aGgdoyOj2Ymn6To/Irl1CCFEDan1wmuaKNa4VRSl3DsBoNDJu3DheffVVmjWrYJnMSkyfPp3MzEzzKyEh4Ybz3CDFF9e4Q7oCcDIlh7s/3sKXm88A8EjPcNr2vhta36tuBhI1qZYyKoQQ9VutVYm8vb3R6XTlatcpKSnlauEA2dnZ7Nq1i7179/Lkk08CYDKZUBQFGxsbVq9eTf/+/ctdp9fr0ev1NVOIhiI/U93cAygM7Mz8tSf46K+TFBpMeDrZ8fbItgyILP47G/llLWZUCCHqv1oL3HZ2dkRFRRETE8Pdd99tPh8TE8Ndd91VLr2rqyuxsbEW5z755BPWrVvHDz/8QHh4eI3nucFK2AkoFLiGMmzhCU6m5ADQp5kPb41si59r1fbjFkIIceNqtRNy2rRpjB8/nujoaLp168Znn31GfHw8kyerq3FNnz6d8+fP89VXX6HVamndurXF9b6+vtjb25c7L6wsXu3fXp0dzsnLOXg72/HvO1oyvF1ghd0aQgghak6tBu7Ro0eTlpbGrFmzSEpKonXr1qxcuZLQ0FAAkpKSrjmnW9yA/Ey4FAcBba+errh/e1NhUxr7OPHj491xd7S7CRkUQghxJY2iKEptZ+JmysrKws3NjczMTFxdXWs7O7Xr2zFw/A8Y9XXp2uNXMhRimtMIrbGAAQVv88pDd9Orqc/NzacQQtRz1YlNtT6qXNSS3DQ4sVo9XjtL3UCk2G8HEpm2dB8bjl/ElLgPrbGANMWFppEdJWgLIUQtk4m2DdWx30ExqsdpJyB2GbQfS+y5TKYu2YfBpLB873med/qTJ4E9Sgtm3NGyVrMshBBCAnfDVbLlpkcYXDoLf88hr/ldTPtuJ89qv2O4/W6MRhOehgzQgD6iO8GejrWYYSGEECCBu2HKz4RTf6nH9y2Cb0ZBRhxbFs3gjexNdLI5DiageMC4CS1dBo2qrdwKIYQoQwJ3Q3R8FZiKwLs5BHaAXs/Bny9y24UvQQsGW2dshr0NnhEAaF380XuE1W6ehRBCABK4G6bDv6jvxSPJTR0nkrrqLXyVNC44NMHvke/Bq3EtZlAIIURlZFR5Q1OYCyfXqseRauBefyabsfkv8RqP4PLkXxK0hRDiFiaBu6E5uQYMl8E9VN2iE1i8NY5TShBK1MM4OjXwue1CCHGLk8DdkOSmwub31eOWw0GjISE9j3XHUgC4v2tILWZOCCFEVUgfd0ORsBOWTYSs8xh0DhS1HocD8N2OeBQFejTxorGPc23nUgghxDVIjbshiP0BFg6BrPPEawIZkvcqd3yXwv6EDJbuVPcnH981tJYzKYQQoiqkxl3fZcTDiqfBVMQJr/7cfX4cOTjCxVzu+ngzAH6uem6LLL8HuhBCiFuP1LjrM0WBX6dCUS75gV2448Ij5ODIayNaM6CFrznZ2M4h2Ojkn4IQQtQFUuOuzw4shVNrUXR6Zir/oMAAvZp680CXEB7oEsKSnQkcOJfJwz3DazunQgghqkgCd31iMkH6aTAZoCgP/nwJgOMtprB0tz12Oi2z7mqNRqOuZTq2cwhjO9dmhoUQQlSXBO6bQVGgOFhW6fz1+uMF2PmFxak8z5aMio0G4B99Igj3drLe84QQQtx00rFZ085sgLcbw6oZYCxSz5mM8Ncb8J8wOPRz5dceWAb/CYfT66/9nNSTsGuheuzoBY5eFLqGMjH9QTILNfRo4sWT/ZvcaGmEEELUMgncNe34KshLg60fwaI7IPkgfDMS1v8H8jNg+6cVX2c0wNpX4XI6bHr32s9Z/6a6v3az2+GfpznzUCw9Lr/Lzvwg2ge789n4aPQ2OqsWTQghxM0nTeU1Let86XHCNpjfQz22cVCXHo3fCtkXwOWK6VjHVkKmOsea0+sh8zy4BQGQmVfEX8dSCPZ0JDLABcdLx9W52gD9ZvDbgUSmL48lO99AC38XFj3YCSe9/FULIUR9ID/Na1pmceC+7RU1uF44CJ6NYfTX8MuTkLgHjv0O0Q9ZXmeuiWsARR0h3msaAC8tP8AfB5MB0Grgf04f0AuF074Dmb9Z4ftdewHoEOLOp+OjcHe0q/lyCiGEuCkkcNe0khp3WG/o/A84uxFCu4PeRV0vPHEPHF5hGbiTYyFuE2h00OdF+PsN2L8Eej7LqdRcNh46w93aXfg4mNDkZ9HLsA2jouHRhEGcij+HRgNP9G3CM7c1xVbmZwshRL0igbsmGQ2QnaQeuwaCnSM0G1z6feRwWPOKGszz0sHRUz1fUttuORy6Pg6b3oPUY5C4h6+2FLHE9jVaa8+CEbBVk57wH0ozlygCCww83rcx3Rt736RCCiGEuJkkcNeknGRQTKC1AWff8t97NQbfVpByCI7/Ce3HQW4axC5Tv+8yGexdIfIOiF1G3o6v6B0bS2vtWQx6d2zCivvL9S60GPgq81z8b17ZhBBC1AoJ3DUpK1F9dwkEbSUjulsOVwP34RXQ6m51LrYhHwLaoTTqzOmLOYS3GYM2dhmO+xcxQAuF2GJ7//cQ0uXmlUUIIcQtQTpAa1LmOfW9eDR4hSLvVN9PrYMFA+Hgj6DRovR5ied/iGXAO+u57WfI1fuYLznU9W00ErSFEKJBksBdk0oGprkGVp7Gt6U6ytxYoA5Kc/SG8T+xKC2SH/eogf90egEf5QwAYL7dJNoOmlTDGRdCCHGrksBdk0qmgrlepcat0UDb0epxo87wjw1sow2zfz8CwAuDmzN7RGtWeYyhbf5nBA57EZ3WisukCiGEqFOkj7smldS43RpdNdn3DiNZUqgn9WJLmv+SxJ64SxhNCne1D2RK38ZoNBrGdQ4hO9+Am6PtTci4EEKIW5UE7pqUde0ad2LGZV79/QS5pqaQWUR85gUAIgNcefOetuadvLRajQRtIYQQErhrVOa1+7hfWXGI3EIjUaEevDC4OYcSs0jMuMwjvcJxsJO1xYUQQliSwF1TDIWQo9aeK2sqX3UomdWHL2Cj1fDG3W1o7u9C1wivm5hJIYQQdY0MTqspOcmAAjo7daT4lV8XGJj5yyEAHusdQXN/l5ucQSGEEHWRBO6aUraZXFv+j/nLTWdIzsonxNORp/o3vcmZE0IIUVdJ4K4p1xiYtjJWXcP8qf5NpC9bCCFElUngriklq6ZVELjj0/I4mpyNTqvhtki/ct8LIYQQlZHAXVNK1imvYLnT1YfVvbQ7h3ni4SR7ZQshhKg6Cdw15SpN5asPq6PNB7WS2rYQQojqkcBdU8wbjFhOBUvLKWDX2XQABraUwC2EEKJ6JHDXlEo2GFl7JAWTAq2DXGnk4VgLGRNCCFGXSeCuCYYCyL2oHrta1rhL+rcHtfS/2bkSQghRD0jgrgklA9Ns7MHR03w6t8DAhhOpgPRvCyGEuD6y5GlNKDswTaPBYDSxNyGDH3ado9BgIsTTkeZ+slKaEEKI6pPAXRPKrJq2Oy6dR7/aTXpuofnrO9oGmHf9EkIIIapDAndNOLlGfXcPZcmOBNJzC3FzsKVPMx8GRPoypHVA7eZPCCFEnSWB29pOrIHY7wENStQktnyTBsCHYzvQu5lP7eZNCCFEnSeD06ypIAd+m6oed5lMvGNLzmdcxlanITrMo1azJoQQon6QwG1N616DzARwD4H+/8eWU2ptu0OwB4520rghhBDixkngtpbze2D7p+rxHXNB72wO3N0ae9VevoQQQtQrEritZe9iQIHW90KTASiKwtZT6pzt7hK4hRBCWIkEbmtQFDgZox63GQXAiZQcUnMKsbfV0iFE+reFEEJYhwRua0g9DhnxoLOD8F4AbDmp1rY7hXliZyN/zEIIIaxDIoo1nCiubYf2ADsnADYX9293b+xdW7kSQghRD0ngtoYTq9X3poMAMJoUtp0uCdzSvy2EEMJ6JHDfqIIciN+qHjcdCMChxEyy8w242NvQKtC1FjMnhBCivpHAfaPObABjIXiEgVcTADYW7wDWJdwTG538EQshhLCeWo8qn3zyCeHh4djb2xMVFcXGjRsrTbt8+XIGDhyIj48Prq6udOvWjVWrVt3E3FagpJm8yUAo3jjkz4PqntsDImXrTiGEENZVq4F76dKlTJ06lRkzZrB371569erFkCFDiI+PrzD9hg0bGDhwICtXrmT37t3069ePO++8k717997knBdTlNINRYr7txPS84g9n4lWA4NaSuAWQghhXRpFUZTaeniXLl3o2LEj8+bNM5+LjIxkxIgRzJkzp0r3aNWqFaNHj+bll1+uUvqsrCzc3NzIzMzE1fUG+59TjsAnXUGnhxfPgp0jn204xRsrj9K9sRffPtr1xu4vhBCiQahObKq1GndhYSG7d+9m0KBBFucHDRrEli1bqnQPk8lEdnY2np6elaYpKCggKyvL4mU1JdPAwnuBnSMAK2PVZvIhbWTrTiGEENZXa4E7NTUVo9GIn59lc7Kfnx/JyclVusc777xDbm4uo0aNqjTNnDlzcHNzM7+Cg4NvKN8WGkVDxwnQeiQA5zMusy8hA40GBreSZnIhhBDWV+tbVmmKB3SVUBSl3LmKfPfdd7zyyiv88ssv+Pr6Vppu+vTpTJs2zfw5KyvLesE7tLv6KlYyKK1TmCe+LvbWeYYQQghRRq0Fbm9vb3Q6XbnadUpKSrla+JWWLl3Kww8/zLJly7jtttuumlav16PX6284v1XxR2wSAENb+9+U5wkhhGh4aq2p3M7OjqioKGJiYizOx8TE0L1790quUmvakyZN4ttvv2XYsGE1nc0qS87MZ1fcJQBuby3920IIIWpGrTaVT5s2jfHjxxMdHU23bt347LPPiI+PZ/LkyYDazH3+/Hm++uorQA3aEyZM4P3336dr167m2rqDgwNubm61Vg6ANUcuABAV6oG/mzSTCyGEqBm1GrhHjx5NWloas2bNIikpidatW7Ny5UpCQ0MBSEpKspjT/emnn2IwGHjiiSd44oknzOcnTpzIokWLbnb2Lew4kw5A76Y+tZoPIYQQ9VutzuOuDVadx11GjzfXcT7jMt880oUeTWRHMCGEEFVXJ+Zx1ydJmZc5n3EZnVZD+2D32s6OEEKIekwCtxXsLh6UFhnggpO+1mfYCSGEqMckcFvBrrNq4I4K8ajlnAghhKjvJHBbQUmNOyqs8qVXhRBCCGuQwH2D8goNHE5S1z+PDpUatxBCiJolgfsG7UvIwGhSCHSzJ9DdobazI4QQop6TwH2Ddp+VZnIhhBA3jwTuG1SyzGlUiHvtZkQIIUSDIIH7BphMCnvi1cAdLTVuIYQQN4EE7htwIiWH7HwDjnY6Wvi71HZ2hBBCNAASuG/Arjh1ffIOIe7Y6OSPUgghRM2TaHMD3Bxs6RTmQffGsja5EEKIm0PW57wBd7QN5I62gbWdDSGEEA2I1LiFEEKIOkQCtxBCCFGHSOAWQggh6hAJ3EIIIUQdIoFbCCGEqEMkcAshhBB1iARuIYQQog6RwC2EEELUIRK4hRBCiDpEArcQQghRh0jgFkIIIeqQBrdWuaIoAGRlZdVyToQQQghVSUwqiVFX0+ACd3Z2NgDBwcG1nBMhhBDCUnZ2Nm5ubldNo1GqEt7rEZPJRGJiIi4uLmg0mhu+X1ZWFsHBwSQkJODq6mqFHN66GlJZoWGVtyGVFaS89VldLauiKGRnZxMYGIhWe/Ve7AZX49ZqtTRq1Mjq93V1da1T/0huREMqKzSs8jaksoKUtz6ri2W9Vk27hAxOE0IIIeoQCdxCCCFEHSKB+wbp9XpmzpyJXq+v7azUuIZUVmhY5W1IZQUpb33WEMra4AanCSGEEHWZ1LiFEEKIOkQCtxBCCFGHSOAWQggh6hAJ3EIIIUQdIoH7BnzyySeEh4djb29PVFQUGzdurO0s3bA5c+bQqVMnXFxc8PX1ZcSIERw7dswijaIovPLKKwQGBuLg4EDfvn05dOhQLeXYuubMmYNGo2Hq1Knmc/WpvOfPn+eBBx7Ay8sLR0dH2rdvz+7du83f16eyGgwG/u///o/w8HAcHByIiIhg1qxZmEwmc5q6XN4NGzZw5513EhgYiEaj4eeff7b4viplKygo4KmnnsLb2xsnJyeGDx/OuXPnbmIpquZqZS0qKuLFF1+kTZs2ODk5ERgYyIQJE0hMTLS4R10pa5Uo4rosWbJEsbW1VT7//HPl8OHDyjPPPKM4OTkpcXFxtZ21GzJ48GBl4cKFysGDB5V9+/Ypw4YNU0JCQpScnBxzmjfffFNxcXFRfvzxRyU2NlYZPXq0EhAQoGRlZdVizm/cjh07lLCwMKVt27bKM888Yz5fX8qbnp6uhIaGKpMmTVK2b9+unDlzRlmzZo1y8uRJc5r6UlZFUZTZs2crXl5eym+//aacOXNGWbZsmeLs7KzMnTvXnKYul3flypXKjBkzlB9//FEBlJ9++sni+6qUbfLkyUpQUJASExOj7NmzR+nXr5/Srl07xWAw3OTSXN3VypqRkaHcdtttytKlS5WjR48qW7duVbp06aJERUVZ3KOulLUqJHBfp86dOyuTJ0+2ONeiRQvlpZdeqqUc1YyUlBQFUNavX68oiqKYTCbF399fefPNN81p8vPzFTc3N2X+/Pm1lc0blp2drTRt2lSJiYlR+vTpYw7c9am8L774otKzZ89Kv69PZVUURRk2bJjy0EMPWZy75557lAceeEBRlPpV3iuDWVXKlpGRodja2ipLliwxpzl//ryi1WqVP//886blvboq+iXlSjt27FAAc0Wqrpa1MtJUfh0KCwvZvXs3gwYNsjg/aNAgtmzZUku5qhmZmZkAeHp6AnDmzBmSk5Mtyq7X6+nTp0+dLvsTTzzBsGHDuO222yzO16fyrlixgujoaO677z58fX3p0KEDn3/+ufn7+lRWgJ49e7J27VqOHz8OwP79+9m0aRNDhw4F6l95y6pK2Xbv3k1RUZFFmsDAQFq3bl3ny5+ZmYlGo8Hd3R2of2VtcJuMWENqaipGoxE/Pz+L835+fiQnJ9dSrqxPURSmTZtGz549ad26NYC5fBWVPS4u7qbn0RqWLFnCnj172LlzZ7nv6lN5T58+zbx585g2bRr/+te/2LFjB08//TR6vZ4JEybUq7ICvPjii2RmZtKiRQt0Oh1Go5HXX3+dsWPHAvXr7/ZKVSlbcnIydnZ2eHh4lEtTl3+O5efn89JLLzFu3DjzJiP1rawSuG/AlduCKopila1CbxVPPvkkBw4cYNOmTeW+qy9lT0hI4JlnnmH16tXY29tXmq4+lNdkMhEdHc0bb7wBQIcOHTh06BDz5s1jwoQJ5nT1oawAS5cuZfHixXz77be0atWKffv2MXXqVAIDA5k4caI5XX0pb0Wup2x1ufxFRUWMGTMGk8nEJ598cs30dbWs0lR+Hby9vdHpdOV+U0tJSSn3G25d9dRTT7FixQr++usvi21Q/f39AepN2Xfv3k1KSgpRUVHY2NhgY2PD+vXr+eCDD7CxsTGXqT6UNyAggJYtW1qci4yMJD4+Hqh/f7cvvPACL730EmPGjKFNmzaMHz+eZ599ljlz5gD1r7xlVaVs/v7+FBYWcunSpUrT1CVFRUWMGjWKM2fOEBMTY7GlZ30rqwTu62BnZ0dUVBQxMTEW52NiYujevXst5co6FEXhySefZPny5axbt47w8HCL78PDw/H397coe2FhIevXr6+TZR8wYACxsbHs27fP/IqOjub+++9n3759RERE1Jvy9ujRo9zUvuPHjxMaGgrUv7/bvLw8tFrLH3E6nc48Hay+lbesqpQtKioKW1tbizRJSUkcPHiwzpW/JGifOHGCNWvW4OXlZfF9fSorINPBrlfJdLAFCxYohw8fVqZOnao4OTkpZ8+ere2s3ZDHH39ccXNzU/7++28lKSnJ/MrLyzOnefPNNxU3Nzdl+fLlSmxsrDJ27Ng6M4WmKsqOKleU+lPeHTt2KDY2Nsrrr7+unDhxQvnmm28UR0dHZfHixeY09aWsiqIoEydOVIKCgszTwZYvX654e3sr//znP81p6nJ5s7Ozlb179yp79+5VAOXdd99V9u7dax5JXZWyTZ48WWnUqJGyZs0aZc+ePUr//v1vySlSVytrUVGRMnz4cKVRo0bKvn37LH5uFRQUmO9RV8paFRK4b8DHH3+shIaGKnZ2dkrHjh3NU6bqMqDC18KFC81pTCaTMnPmTMXf31/R6/VK7969ldjY2NrLtJVdGbjrU3l//fVXpXXr1oper1datGihfPbZZxbf16eyZmVlKc8884wSEhKi2NvbKxEREcqMGTMsfpjX5fL+9ddfFf5fnThxoqIoVSvb5cuXlSeffFLx9PRUHBwclDvuuEOJj4+vhdJc3dXKeubMmUp/bv3111/me9SVslaFbOsphBBC1CHSxy2EEELUIRK4hRBCiDpEArcQQghRh0jgFkIIIeoQCdxCCCFEHSKBWwghhKhDJHALIYQQdYgEbiGEEKIOkcAthLjpNBoNP//8c21nQ4g6SQK3EA3MpEmT0Gg05V633357bWdNCFEFsh+3EA3Q7bffzsKFCy3O6fX6WsqNEKI6pMYtRAOk1+vx9/e3eHl4eABqM/a8efMYMmQIDg4OhIeHs2zZMovrY2Nj6d+/Pw4ODnh5efHYY4+Rk5NjkebLL7+kVatW6PV6AgICePLJJy2+T01N5e6778bR0ZGmTZuyYsWKmi20EPWEBG4hRDn//ve/uffee9m/fz8PPPAAY8eO5ciRI4C6z/Xtt9+Oh4cHO3fuZNmyZaxZs8YiMM+bN48nnniCxx57jNjYWFasWEGTJk0snvHqq68yatQoDhw4wNChQ7n//vtJT0+/qeUUok6q7e3JhBA318SJExWdTqc4OTlZvGbNmqUoirq16+TJky2u6dKli/L4448riqIon332meLh4aHk5OSYv//9998VrVarJCcnK4qiKIGBgcqMGTMqzQOg/N///Z/5c05OjqLRaJQ//vjDauUUor6SPm4hGqB+/foxb948i3Oenp7m427dull8161bN/bt2wfAkSNHaNeuHU5OTubve/Togclk4tixY2g0GhITExkwYMBV89C2bVvzsZOTEy4uLqSkpFxvkYRoMCRwC9EAOTk5lWu6vhaNRgOAoijm44rSODg4VOl+tra25a41mUzVypMQDZH0cQshytm2bVu5zy1atACgZcuW7Nu3j9zcXPP3mzdvRqvV0qxZM1xcXAgLC2Pt2rU3Nc9CNBRS4xaiASooKCA5OdninI2NDd7e3gAsW7aM6OhoevbsyTfffMOOHTtYsGABAPfffz8zZ85k4sSJvPLKK1y8eJGnnnqK8ePH4+fnB8Arr7zC5MmT8fX1ZciQIWRnZ7N582aeeuqpm1tQIeohCdxCNEB//vknAQEBFueaN2/O0aNHAXXE95IlS5gyZQr+/v588803tGzZEgBHR0dWrVrFM888Q6dOnXB0dOTee+/l3XffNd9r4sSJ5Ofn89577/H888/j7e3NyJEjb14BhajHNIqiKLWdCSHErUOj0fDTTz8xYsSI2s6KEKIC0scthBBC1CESuIUQQog6RPq4hRAWpPdMiFub1LiFEEKIOkQCtxBCCFGHSOAWQggh6hAJ3EIIIUQdIoFbCCGEqEMkcAshhBB1iARuIYQQog6RwC2EEELUIf8P4X2pDRHhPH4AAAAASUVORK5CYII=",
"text/plain": [
"<Figure size 1200x400 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plt.figure(figsize=(12, 4))\n",
"plt.subplot(1, 2, 1)\n",
"plt.plot(history.history['accuracy'])\n",
"plt.plot(history.history['val_accuracy'])\n",
"plt.title('Model accuracy')\n",
"plt.ylabel('Accuracy')\n",
"plt.xlabel('Epoch')\n",
"plt.legend(['Train', 'Test'], loc='upper left')"
]
},
{
"cell_type": "markdown",
"id": "617d7de2-cf38-4b87-a94e-8a1f6c406c37",
"metadata": {},
"source": [
"## Description\n",
"\n",
"- **X-axis (Epoch)**: Represents the number of epochs, ranging from 0 to 130\n",
"- **Y-axis (Accuracy)**: Represents accuracy, ranging from 0 to 1.0.\n",
"\n",
"## Observations\n",
"\n",
"- **Training Accuracy (Blue Line)**: The 'Train' accuracy line shows a consistent increase over epochs, indicating continuous improvement in performance on the training data.\n",
"- **Test Accuracy (Orange Line)**: The 'Test' accuracy line fluctuates but generally follows the upward trend of the 'Train' line. This suggests that the model's performance on unseen test data also improves over epochs, with some variability."
]
},
{
"cell_type": "markdown",
"id": "476b3c27-8d4a-4e89-846f-513824e987cb",
"metadata": {},
"source": [
"# Confusion Matrix\n",
"\n",
"The following plot shows the confusion matrix for the Conv1D model's predictions on the test dataset.\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 91,
"id": "1a149853-2e28-4a36-92a6-ca4f1312bd2c",
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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",
"text/plain": [
"<Figure size 640x480 with 2 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"\n",
"# Compute confusion matrix\n",
"conf_matrix = confusion_matrix(true_labels, predicted_labels)\n",
"sns.heatmap(conf_matrix, annot=True, fmt=\"d\", cmap=\"Blues\", xticklabels=label_encoder.classes_, yticklabels=label_encoder.classes_)\n",
"plt.ylabel('Actual')\n",
"plt.xlabel('Predicted')\n",
"plt.show()\n"
]
},
{
"cell_type": "markdown",
"id": "7042688c-5e80-412b-a2f1-9cb5c62b9281",
"metadata": {},
"source": [
"\n",
"\n",
"### Description\n",
"\n",
"- **X-axis (Predicted)**: Represents the predicted emotion labels by the model.\n",
"- **Y-axis (Actual)**: Represents the actual emotion labels.\n",
"- **Color Scale**: The intensity of the color represents the number of instances classified in each category, with darker colors indicating a higher number of instances.\n",
"\n",
"### Emotion Labels\n",
"The numerical labels correspond to the following emotions:\n",
"- 0: Neutral\n",
"- 1: Calm\n",
"- 2: Happy\n",
"- 3: Sad\n",
"- 4: Angry\n",
"- 5: Fearful\n",
"- 6: Disgust\n",
"- 7: Surprised\n",
"\n",
"### Observations\r\n",
"\r\n",
"1. **Class 0:**\r\n",
" - Perfect prediction: 36 instances correctly predicted.\r\n",
" - No misclassifications.\r\n",
"\r\n",
"2. **Class 1 to 7:**\r\n",
" - The majority of predictions are correct for each class.\r\n",
" - There is a slight off-diagonal spread indicating some misclassifications to neighboring classes.\r\n",
"\r\n",
"3. **General Trend:**\r\n",
" - High accuracy for each class.\r\n",
" - Misclassifications mostly occur in adjacent classes, indicating that errors are minor and close to the actual class.\r\n",
"\r\n",
"### Conclusions\r\n",
"\r\n",
"- **High Model Accuracy:** The LSTM model performs exceptionally well, particularly for class 0 with no misclassifications.\r\n",
"- **Adjacent Class Misclassifications:** The few errors made are predominantly in predicting adjacent classes, suggesting the model's confusion is limited to closely related classes.\r\n",
"- **Strong Performance Across All Classes:** The confusion matrix demonstrates that the model has robust predictive capabilities for a range of classes.\r\n"
]
},
{
"cell_type": "code",
"execution_count": 57,
"id": "989966bb-a2b4-48b9-b2b6-c6f4e9a6fb61",
"metadata": {},
"outputs": [],
"source": [
"model.save('emotion_recognition_model_LSTM.keras')"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "f4596afd-1c4a-4dfe-8dd5-7d8c907e9737",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.9"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
|