@misc{laurent2023featlearning,
title={Long-Tailed Learning Requires Feature Learning},
author={Thomas Laurent and James von Brecht and Xavier Bresson},
year={2023},
booktitle={International Conference on Learning Representations (ICLR)}
}
conda env create -f environment.yml
conda activate pytorch_env
The folder eval_formula_from_theorem7 contains a code that evaluates the combinatorial formula between square bracket appearing in Theorem 7. This code will display inequality (105) for every
The folder neural_net_and_svm_experiments contains codes that will run the neural network in Figure 2 on our data model, as well as a plain SVM or a SVM with Gaussian kernel on the top of features extracted by
The folder nearest_neighbor_experiments contains codes to run a nearest neighbor classification rule on the top of features extracted by