In the latest in our series of interviews meeting the AAAI/SIGAI Doctoral Consortium participants, we caught up with Aniket Roy to find out more about his research on generative models for computer vision tasks. Tell us a bit about your PhD - where did you study, and what was the topic of your research? I recently completed my PhD in Computer Science at Johns Hopkins University, where I worked under the supervision of Bloomberg Distinguished Professor Rama Chellappa. My research primarily focused on developing methods for resource-constrained image generation and visual understanding. In particular, I explored how modern generative models can be adapted to operate efficiently while maintaining strong performance.

Mamba is an effective state space model with linear computation complexity. It has recently shown impressive efficiency in dealing with high-resolution inputs across various vision tasks. In this paper, we reveal that the powerful Mamba model shares surprising similarities with linear attention Transformer, which typically underperform conventional Transformer in practice. By exploring the similarities and disparities between the effective Mamba and subpar linear attention Transformer, we provide comprehensive analyses to demystify the key factors behind Mamba's success. Specifically, we reformulate the selective state space model and linear attention within a unified formulation, rephrasing Mamba as a variant of linear attention Transformer with six major distinctions: input gate, forget gate, shortcut, no attention normalization, single-head, and modified block design.

Weight decay is a broadly used technique for training state-of-the-art deep networks from image classification to large language models. Despite its widespread usage and being extensively studied in the classical literature, its role remains poorly understood for deep learning. In this work, we highlight that the role of weight decay in modern deep learning is different from its regularization effect studied in classical learning theory. For deep networks on vision tasks trained with multipass SGD, we show how weight decay modifies the optimization dynamics enhancing the ever-present implicit regularization of SGD via the . In contrast, for large language models trained with nearly one-epoch training, we describe how weight decay balances the in stochastic optimization leading to lower training loss and improved training stability. Overall, we present a unifying perspective from ResNets on vision tasks to LLMs: weight decay is never useful as an explicit regularizer but instead changes the training dynamics in a desirable way.

Our experiments demonstrate that MSVMamba is highly competitive, with the MSVMamba-Tinymodel achieving83.0%

As aresult, there is aproliferation of distinct architectures and loss functions for different vision tasks.