Deep Learning Trend and how NVDLA empowers it



Speaker:Yilin Zhang (张益林)
Yilin is a hardware architect at NVIDIA who responsible for NVDLA architecture define. He holds his MS degree at SJTU in 2009, his research interests lie in deep learning, image processing and hardware architecting.

Topic:Deep Learning Trend and how NVDLA empowers it

Deep learning has proved itself as a break-through technology for different problem domains (e.g.: NLP, computer vision, recommendation, etc.), however, due to tremendous computation workloads required, deploy deep learning on embedded system (e.g.: Mobile phones, Drones, etc.) become a challenge for industry.
NVDLA is designed to overcome this problem by providing a high performance, low power solution. In this talk, we’re going to cover topics below:
* Basic concepts in deep learning;
* Research trend for deep learning landing;
* Challenges for design a deep learning accelerator and how NVDLA achieved that;