CSE 8803 Deep Learning for Text Data project
Embedding attention visualizations in CNN-biLSTMs designed for protein subcellular localization based on primary sequence of amino acids
Dataset
DeepLoc https://academic.oup.com/bioinformatics/article/33/21/3387/3931857
Models
CNN-biLSTM-Attention https://github.com/JJAlmagro/subcellular_localization
LSTM-Attention https://www.aclweb.org/anthology/D19-1002.pdf
LSTM https://github.com/ThanhTunggggg/DeepLoc
Interpretability Framework
SHAP https://github.com/slundberg/shap
Attention visualization + deconvolution if needed
FRESH https://arxiv.org/abs/2005.00115
Deliverables
Presentation https://docs.google.com/presentation/d/1-qk5zt8Ci-8k-OinnM0WFir8EePpYL7E1R5Sh2ktiT4/edit?usp=sharing