feat: TabNetMultiTaskClassifier

.circleci/config.yml

@@ -151,6 +151,20 @@ jobs:
  shell: bash -leo pipefail
  command: |
  make test-nb-regression
+ test-nb-multi-task:
+ executor: python-executor
+ steps:
+ - checkout
+ # Download and cache dependencies
+ - restore_cache:
+ keys:
+ - v1-dependencies-{{ checksum "poetry.lock" }}
+ - install_poetry
+ - run:
+ name: run test-nb-multi-task
+ shell: bash -leo pipefail
+ command: |
+ make test-nb-multi-task
 
 workflows:
  version: 2
@@ -171,6 +185,9 @@ workflows:
  - test-nb-forest:
  requires:
  - install
+ - test-nb-multi-task:
+ requires:
+ - install
  - lint-code:
  requires:
  - install

Makefile

@@ -89,6 +89,10 @@ test-nb-multi-regression: ## run multi regression example tests using notebooks
  $(MAKE) _run_notebook NB_FILE="./multi_regression_example.ipynb"
 .PHONY: test-obfuscator
 
+test-nb-multi-task: ## run multi task classification example tests using notebooks
+ $(MAKE) _run_notebook NB_FILE="./multi_task_example.ipynb"
+.PHONY: test-obfuscator
+
 help: ## Display help
  @grep -E '^[a-zA-Z_-]+:.*?## .*$$' $(MAKEFILE_LIST) | awk 'BEGIN {FS = ":.*?## "}; {printf "\033[36m%-30s\033[0m %s\n", $$1, $$2}'
 .PHONY: help

README.md

@@ -10,7 +10,7 @@ This is a pyTorch implementation of Tabnet (Arik, S. O., & Pfister, T. (2019). T
 
 ![PyPI - Downloads](https://img.shields.io/pypi/dm/pytorch-tabnet)
 
-Any questions ? Want to contribute ? To talk with us ? You can join us on [Slack](https://join.slack.com/t/mltooling/shared_invite/zt-e4y14rbp-IReRxmjoNy27XQFQgh~4ZQ)
+Any questions ? Want to contribute ? To talk with us ? You can join us on [Slack](https://join.slack.com/t/mltooling/shared_invite/zt-fxaj0qk7-SWy2_~EWyhj4x9SD6gbRvg)
 
 # Installation
 
@@ -37,6 +37,11 @@ If you wan to use it locally within a docker container:
 
 - `make notebook` inside the same terminal. You can then follow the link to a jupyter notebook with tabnet installed.
 
+# What problems does pytorch-tabnet handles?
+
+- TabNetClassifier : binary classification and multi-class classification problems
+- TabNetRegressor : simple and multi-task regression problems
+- TabNetMultiTaskClassifier: multi-task multi-classification problems
 
 # How to use it?
 
@@ -50,7 +55,23 @@ clf.fit(X_train, Y_train, X_valid, y_valid)
 preds = clf.predict(X_test)
 ```
 
-You can also get comfortable with how the code works by playing with the **notebooks tutorials** for adult census income dataset and forest cover type dataset.
+or for TabNetMultiTaskClassifier :
+
+```
+from pytorch_tabnet.multitask import TabNetMultiTaskClassifier
+clf = TabNetMultiTaskClassifier()
+clf.fit(X_train, Y_train, X_valid, y_valid)
+preds = clf.predict(X_test)
+```
+
+# Useful links
+
+- explanatory video : https://youtu.be/ysBaZO8YmX8
+- binary classification examples : https://github.com/dreamquark-ai/tabnet/blob/develop/census_example.ipynb
+- multi-class classification examples : https://github.com/dreamquark-ai/tabnet/blob/develop/forest_example.ipynb
+- regression examples : https://github.com/dreamquark-ai/tabnet/blob/develop/regression_example.ipynb
+- multi-task regression examples : https://github.com/dreamquark-ai/tabnet/blob/develop/multi_regression_example.ipynb
+- multi-task multi-class classification examples : https://www.kaggle.com/optimo/tabnetmultitaskclassifier
 
 ## Model parameters
 
@@ -176,9 +197,11 @@ You can also get comfortable with how the code works by playing with the **noteb
  1 : automated sampling with inverse class occurences
  dict : keys are classes, values are weights for each class
 
-- loss_fn : torch.loss
+- loss_fn : torch.loss or list of torch.loss
 
  Loss function for training (default to mse for regression and cross entropy for classification)
+ When using TabNetMultiTaskClassifier you can set a list of same length as number of tasks,
+ each task will be assigned its own loss function
 
 - batch_size : int (default=1024)