CCS Signal Processing and AI Engines
Axis Leader
Mark Coates
McGill University
The research challenge is to develop signal processing and AI algorithms with bounded complexity, calibrated uncertainty, and resource awareness. Our methodology is to design efficient and energy aware algorithms that are robust to distribution shift and model/data imperfections induced by real-world operation. By employing distillation, quantization, compression, and sparse computation, we will reduce compute and memory footprints to enable distributed sensing and deployment across edge clouds. The intra-axis collaborations bring together signal processing researchers specializing in wireless and optical communications(e.g., Champagne, Falk, Psaromiligkos), and machine learning and AI experts(e.g., Aïvodji, Bouguila, Coates). The research infrastructure relies upon computational clusters and software at McGill.
