tengjunliu-bmi / schizophrenia-classification Goto Github PK

• Python==3.6.13
• pytorch==1.8.1
• matplotlib==3.2.2
• numpy==1.19.5
• h5py==2.10.0
• scikit-learn==0.24.2
• scipy==1.5.3

1. Data Segmentation & Feature Extraction: In the filefolder "1. Data Segmentation & Feature Extraction", excute the file "dataSegmentationAndFeatureExtraction.m" to segment the data and extract the features, and obtain the samples to be classified. The ready-to-classify samples at different levels (5s-sample and individual) were already stored in the file "Segmented Data".
2. Baseline Model Training: In the filefolder "2. Baseline Model Training & Direct Transferring Prediction & Fine Tuning\*** Level", where "***" indicates different levels, execute the file "cnnClassification_***-PolandTrained.py" to train the baseline models.
3. Direct Transferring Prediction: In the filefolder "2. Baseline Model Training & Direct Transferring Prediction & Fine Tuning\*** Level", where "***" indicates different levels, execute the file "cnnClassification_***-RussiaTested.py" to obtain the baseline transferring accuracy.
4. Fine Tuning the Trained Models: In the filefolder "2. Baseline Model Training & Direct Transferring Prediction & Fine Tuning\*** Level", where "***" indicates different levels, execute the file "cnnClassification_individual-RussiaTested-FineTuning.py" to obtain the fine-tuning transferring accuracy.
5. Applying Transfer Component Analysis (TCA): TCA is implemented in the file "3. TCA & TCA_FT\cnnClassification_Individual-TCA-TCA_FT.py".
6. Train the Models on the Transformed Samples and Fine Tuning the Trained Models: Execute the file "3. TCA & TCA_FT\cnnClassification_Individual-TCA-TCA_FT.py" to obtain the direct transferring predictions of model trained on the TCA-transformed samples, and the corresponding fine-tuning transferring accuracy.
7. Figure Plotting: Execute the file "4. Figure Plotting\resultAnalysis.m". The plotted results were already in the filefolder "4. Figure Plotting".

More details please refer to the file "EEG精神分裂症分类任务的迁移学习.pdf" (Requested from [email protected], in mandarin btw).

1. Singh, K., S. Singh, and J. Malhotra, Spectral features based convolutional neural network for accurate and prompt identification of schizophrenic patients. Proc Inst Mech Eng H, 2021. 235(2): p. 167-184.
2. Pan, S.J., et al., Domain Adaptation via Transfer Component Analysis. IEEE Transactions on Neural Networks, 2011. 22(2): p. 199-210.