- Tensorflow 1.0
- sklearn
- jupyter notebook
- Use data_explore.ipynb to explore the data and preform split.
- Use build_data.ipynb to build training/val/test set.
- Use main.py to train the mlp model
- Use train_svm.ipynb to train the SVM model
- Use vis_result.ipynb to monitor network training results.
- Use eval.py to evaluate the model
- I used voting among SVM, softmax MLP and sigmoid MLP to produce the final results, see build_eval.ipynb for details.
To accesss my data split, trained models and test predictions, see this link.