Closed
Milestone
SP9-item4 : scikit-learn Models Integration
Milestone ID: 2407
As a researcher i want to be able to run all the scikit-learn models that were identified to run with partial fit (so that i have a more complete support for scikit-learn)
- for each model identify parameters that must be federated
- benchmark the results with respect to local training
Currently fedbiomed offers the possibility to apply SGDRegressor and perceptron sklearn models. As a researcher I want to extend this capability to :
- sklearn.naive_bayes.MultinomialNB
- sklearn.naive_bayes.BernoulliNB
- sklearn.linear_model.SGDClassifier
- sklearn.linear_model.PassiveAggressiveClassifier
- sklearn.linear_model.PassiveAggressiveRegressor
- sklearn.cluster.MiniBatchKMeans
- sklearn.decomposition.MiniBatchDictionaryLearning
Next Sprint.
- LatentDirichletAllocation¶
Model assignment :
Yannick
- MiniBatchKMeans
- DictionnaryLearning
- PassiveAggressiveRegressor
- PassiveAggressiveClassifier
José:
Complement.