Research team receives best paper award

At the 2024 IEEE International Workshop on Machine Learning for Signal Processing (MLSP), a team of researchers from Tufts University, East Tennessee State University, Boston University, and Worcester Polytechnic Institute received the best paper award for their paper titled "On neural collapse in contrastive learning with imbalanced datasets". 

The team was composed of first author Thuan Nguyen of East Tennessee State University (previously a postdoctoral researcher at Tufts), PhD student Ruijie Jiang and Associate Professor Shuchin Aeron of the Tufts Department of Electrical and Computer Engineering, Professor Prakash Ishwar of Boston University, and Professor and Department Head D. Richard Brown III of Worcester Polytechnic Institute.

Learn more from the abstract: Neural collapse is a phenomenon in neural networks where all the samples from the same class collapse to their class mean and the class means have a specific geometric structure called Equiangular Tight Frame (ETF). Recent studies have empirically and theoretically confirmed that neural collapse solutions attain a global minimum of Contrastive Learning (CL) losses if the class distribution is uniform (balanced classes). This paper investigates the neural collapse phenomenon in Contrastive Learning for imbalanced datasets. We show that even if the classes are imbalanced, global minima of Contrastive Learning losses include solutions in which all samples from the same class collapse to their class mean. However, the geometric structure of the class means is, in general, different from an ETF. In addition, we show that under certain conditions the optimal geometry of the class means can be found by solving a convex optimization problem. We provide a program based on the CVX package to find these optimal class means.

The IEEE MLSP is the oldest and most respected technical event organized by IEEE Signal Processing Society’s MLSP Technical Committee. The workshop convenes top researchers, experts, and professionals from across the globe to discuss the latest developments in machine learning for signal processing. The 2024 event was hosted at Imperial College in London and included a focus on applications for machine learning-driven signal processing techniques.