Scanning all new published packages on PyPI I know that the quality is often quite bad. I try to filter out the worst ones and list here the ones which might be worth a look, being followed or inspire you in some way.

An end-to-end PyTorch framework for image and video classification.

A set of utility function which help build convolutional neural networks

MICO: Mutual Information and Conic Optimization for feature selection. MICO is a Python package that implements a conic optimization based feature selection method with mutual information (MI) measure. The idea behind the approach is to measure the features’relevance and redundancy using MI, and then formulate a feature selection problem as a pure-binary quadratic optimization problem, which can be heuristically solved by an efficient randomization algorithm via semidefinite programming. Optimization software Colin is used for solving the underlying conic optimization problems.

Helper modules for doing data cleaning, training, exploration, etc. A set of modules that contain helper methods, e.g. the old fastai ‘’ file which contains functions that split a date into parts and automatically convert dataframe object columns to categorical variables.

A Deep Learning GUI-based Framework

A networking protocol for agent-environment communication.

Python supercharged for fastai development

A toolkit for applying machine learning to large source code corpora

A data science pipeline tool to speed up data science life cycle. Using this library, you can:
• Test new experiments easily and keep track of their results.
• Keep details of each preprocessing/FE step easily accessible in collapsibles.
• Do hyperparameter search. (Bayesian search, quick linear search)
• Create a pipeline that consists of useful steps and save/load it.
• Automatically try different processing steps and use useful ones. (imputations, binning, one-hot encoding, …)
• Make your predictions more reliable by averaging results obtained from different CV splits and random seeds.

Neyman-Pearson (NP) Classification Algorithms and NP Receiver Operating Characteristic (NP-ROC) Curves

Ontogram is an OWL ontology diagram generator.

Optimize hyperparameters using the Paddy field algorithm