Download or git clone this project onto local machine into folder on local machine.
git clone https://github.com/asaraog/msdsthesis.git
cd msdsthesis
pip install zenodo_get
zenodo_get 6387085
unzip Kato_et_al.zip
ls
| File | Description |
|---|---|
eeg_decoding_citcyc.ipynb |
Primary analysis notebook: model training, temporal/spatial/spatiotemporal occlusion analysis, class-specific occlusion, decision-threshold sweep, and paper comparison. |
eeg_decoding_citcyc.html |
Rendered HTML export of the analysis notebook (collapsible sections). |
citcyc_cnn_lstm_model.keras |
Final saved Keras model used for Cit vs Cyc evaluation and thesis figure generation. Includes the trained network architecture and weights. |
compute_kato_baseline.py |
Standalone script that reproduces the 54% Kato et al. (2022) cross-subject ridge-regression baseline by loading 4-2vs9-2.mat and averaging MVPA.accs over the 22 LOSO subject folds and 31 post-odor decoding time points (t ≥ 0 ms). |
kato_cit_cyc_grid2d_events.pkl |
Pre-processed Cit vs Cyc trial data in 2-D grid format (28 time bins, 10×11 grid, 2 channels). Direct input for model training and evaluation.Tracked via Git LFS. |
The full upstream dataset is available at
https://zenodo.org/records/6387085 under Kato_et_al/.