Presentation
UniCoS: A Unified Neural and Accelerator Co-Search Framework for CNNs and ViTs
DescriptionCurrent algorithm-hardware co-search works often suffer from lengthy training times and inadequate exploration of hardware design spaces, leading to suboptimal performance. This work introduces UniCoS, a unified framework for co-optimizing neural networks and accelerators for CNNs and Vision Transformers (ViTs). By introducing a novel training-free proxy that evaluates accuracy within seconds and a clustering-based algorithm for exploring heterogeneous dataflows, UniCoS efficiently navigates the design spaces of both architectures. Experimental results demonstrate that the solutions generated by UniCoS consistently surpass state-of-the-art (SOTA) methods (e.g., 3.54x energy-delay product (EDP) improvement with a 1.76% higher accuracy on ImageNet) while requiring notably reduced search time (up to 48x, ~3 hours). The code will be open-sourced.
Event Type
Research Manuscript
TimeWednesday, June 254:00pm - 4:15pm PDT
Location3000, Level 3
AI
AI3: AI/ML Architecture Design