Group Member: Mu Cai, Yunyu(Bella) Bai, Xuechun Yang
Source code and dataset available at: https://github.com/mu-cai/cs766_21spring
Toward Robust Visual Recognition System under Group Shifts
Recently, people are making more and more efforts to strive for fairness. While human beings can be a main source of bias, machines can also be biased when generating prediction results. As the frequency of using computers to make decisions, fairness-enhancing techniques have received a lot of attention. Our project focuses on reducing bias generated by image classification algorithms, and thus improve the worst-case performance in image classification.
Featured Work
Dataset
- Natural Images crawled from the online sources
- Contains group-shifted data
Our Proposal
This proposal explained what problem we are trying to solve, why we want to solve it, and what are the possible steps to the solution. [LINK]
Our Mid-term report
Our mid-term report includes a brief summary of progress before mid-term, our results, the difficulties that appeared during the implementation, and how our proposal may have changed in light of current progress.​ [LINK]