Abstract of the paper “Competitive Gradient Descent”
Published:
This blog explains method proposed in the paper Competitive gradient descent (Schäfer et al., 2019). This has been written as a supplimentary to the reproducibility report for reproducibility challenge of NeurIPS’19. The report has been published on arXiv. It can be seen at [https://arxiv.org/abs/2001.10820] with this github repository as its source.
The aim of this blog is to give a gentle introduction to the original paper. Follow this link to read the blog. It can also be seen in slides format.
The blog also contains the implementation details of the conjugate gradient algorithm in python3 which is a component of the proposed algorithm, which itself can be seen in this jupyter notebook. Code in python3 (based on PyTorch)