Daniel Bates

About

Daniel Bates is a researcher in the Department of Computer Science and Technology at the University of Cambridge. His research focuses on computer architecture, particularly co-optimizing architecture with software or algorithms. He has worked on projects like Loki, a manycore processor with low-latency communication, and Muntjac, a minimal 64-bit RISC-V multicore processor. He is also interested in optimizing machine learning workloads through techniques such as hardware acceleration, pruning, and dynamic gating.

Selected publications

• Lane Compression: A Lightweight Lossless Compression Method for Machine Learning on Embedded Systems
• Sponge examples: Energy-latency attacks on neural networks
• A communication-centric low-power manycore processor with a configurable memory system
• Exploiting Tightly-Coupled Cores
• Focused Quantization for Sparse CNNs