IBM Research, Japan
Beyond von-Neumann computing: Analog memory-based AI hardware
Modern AI has witnessed tremendous growth in recent years. However, AI faces a severe computing efficiency problem since traditional von-Neumann computing systems involve separate processing and memory units. In-memory computing using analog non-volatile memory could play a key role to overcome the issue and have a significant impact on improving the energy and area efficiency, where certain tasks are performed in place in the memory itself. This presentation introduces our current research activities on such analog memory-based AI hardware accelerators, including our recent research progress on a spiking neural network chip which consists of large-scale phase change memory synaptic arrays and stochastic leaky integrate-and-fire neuron circuits.
Masatoshi Ishii is a Research Staff Member at IBM Research – Tokyo. He joined IBM Japan in 1998, where he has been engaged in broad electrical engineering fields including printed circuit board development for ThinkPad, signal and power integrity simulation for memory interfaces and full custom circuit design for SRAM and CAM macros. His current research interests include neuromorphic chip design and hardware-aware neural network simulator development.
If you wish to modify any information or update your photo, please contact the Publicity Chair at the following address: