GitHub

Rethinking Temporal Fusion with a Unified Gradient Descent View for 3D Semantic Occupancy Prediction

In autonomous driving, understanding the 3D world over time is critical. Yet, most vision-based 3D Occupancy (VisionOcc) methods only scratch the surface of temporal fusion, focusing on simple ...
IEEE

Vision-Language Gradient Descent-driven All-in-One Deep Unfolding Networks

Abstract: Dynamic image degradations, including noise, blur and lighting inconsistencies, pose significant challenges in image restoration, often due to sensor limitations or adverse environmental ...
IEEE

AlphaGrad: Normalized Gradient Descent for Adaptive Multi-Loss Functions in EEG-Based Motor Imagery Classification

Abstract: In this study, we propose AlphaGrad, a novel adaptive loss blending strategy for optimizing multi-task learning (MTL) models in motor imagery (MI)-based electroencephalography (EEG) ...