Generative Adversarial Network (GAN) research

During my Quantitative Analytics Internship in an AI modeling team, I had the opportunity to research and implement GANs to solve the problem of imbalanced datasets, which are popular in ML problems. Imbalanced datasets can cause difficulties in models’ convergence and undesirable biases towards the major class of the datasets. My research emphasized fixing this problem by using GANs to generate necessary data of the minor class for better classification and observation. An overly simplified version of my project can be found at github, which has been adjusted to be public.