https://docs.google.com/document/d/1Kb_J5WfAwnJBapz6ZQFczAWgAVzlgKiWKP9tk-ZRwig/edit?usp=sharing
[1] S. Popov, A. Babenko, S. Morozov - Neural Oblivious Decision Ensembles for Deep Learning on Tabular Data. arxiv/1909.06312
[2] Mechanisms of Action (MoA) Prediction - kaggle.com/c/lish-moa
There's currently 2 models available.
notebooks/MoA_DNN.ipynb
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
flatten_8 (Flatten) (None, 876) 0
_________________________________________________________________
dense_32 (Dense) (None, 480) 420960
_________________________________________________________________
dense_33 (Dense) (None, 480) 230880
_________________________________________________________________
dense_34 (Dense) (None, 480) 230880
_________________________________________________________________
dense_35 (Dense) (None, 480) 230880
_________________________________________________________________
last_frozen (Dense) (None, 480) 230880
_________________________________________________________________
dense_36 (Dense) (None, 206) 99086
_________________________________________________________________
dense_37 (Dense) (None, 206) 42642
=================================================================
Total params: 1,486,208
Trainable params: 1,486,208
Non-trainable params: 0
notebooks/MoA_NODE.ipynb
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
lambda (Lambda) multiple 0
_________________________________________________________________
batch_normalization (BatchNo multiple 3140
_________________________________________________________________
batch_normalization_1 (Batch multiple 3652
_________________________________________________________________
batch_normalization_2 (Batch multiple 4164
_________________________________________________________________
batch_normalization_3 (Batch multiple 4676
_________________________________________________________________
dropout (Dropout) multiple 0
_________________________________________________________________
dropout_1 (Dropout) multiple 0
_________________________________________________________________
dropout_2 (Dropout) multiple 0
_________________________________________________________________
dropout_3 (Dropout) multiple 0
_________________________________________________________________
odst (ODST) multiple 39510
_________________________________________________________________
odst_1 (ODST) multiple 41814
_________________________________________________________________
odst_2 (ODST) multiple 44118
_________________________________________________________________
dense (Dense) multiple 241020
=================================================================
Total params: 382,094
Trainable params: 371,974
Non-trainable params: 10,120