Mathematician and physicist, working as a data scientist.
I maintain a python library called sktools, it has tools to add extra functionalities to scikit-learn.
Some of my relevant answers:
- Saddle points in deep learning.
- Embedding layers are linear layers and embeddings backpropagation is easy.
- Root finding in logistic regression.
- Choosing a learning rate.
Some (so far, and in my opinion) unanswered questions: