Realtime Machine Learning on Edge AI Accelerators

Explore the cutting-edge developments in real-time machine learning for edge AI accelerators in this informative 55-minute talk by Dr. Ramtin Zand. Delve into the growing demand for specialized hardware in AI applications, focusing on edge AI accelerators like NVIDIA Jetson Nano, Intel NCS2, and Google TPU. Discover optimization techniques such as neural architecture search (NAS) and system-level innovations for deploying advanced models, including transformers, on these platforms. Learn about recent projects addressing computer vision and natural language processing tasks on edge accelerators through computational graph modifications, partitioning, and operation refactoring. Gain insights from Dr. Zand's extensive research experience in neuromorphic computing, edge computing, processing-in-memory, and AI/ML hardware acceleration, backed by collaborations with major tech companies and recognition from ACM/IEEE.