Socionext Inc., Japan
Low Power Edge AI SoC for Quantized Deep Neural Network
Quantized Deep Neural Network is very important for edge devices that enable low power inference. In this session, we present a new Edge AI SoC with custom hardware implementations for quantized Deep Neural Networks and post-training quantization approach to improve accuracy and performance. Deep Neural Networks require massive memory accesses and computations. These hardware and software techniques can realize low power and high performance of edge devices.
Takanori Isono is a Director of System Development Department of Automotive and Industrial Business Group at Socionext Inc., which was established in 2015 by the integration of System LSI divisions of Panasonic and Fujitsu. Recently he has been working on development of Vision Processor for Automotive SoC, and is currently focused on research and development of edge AI accelerator architecture of SoCs for automotive and industrial market.
If you wish to modify any information or update your photo, please contact the Publicity Chair at the following address: