On Friday, March 5th, MATRIX is pleased to host Dr. Amey Kulkarni from NVIDIA as part of our seminar series.  All spring seminars will be hosted via Zoom. Full details regarding the seminar may be found below –

Demystifying Deep Learning Inference on Embedded Platforms

Dr. Amey Kulkarni

NVIDIA

Friday, March 5, 2021

11 AM -12 PM CST

tinyurl.com/MATRIXSpringSeminar

Implementing deep learning inference on resource constraint embedded platforms is challenging. In recent years, we are witnessing artificial intelligence being deployed on embedded platforms in our everyday life, such as keywords detection on voice assistant devices, voice and image recognition on cell phones, robotics platforms performing food delivery and smart security systems etc. However, successful deployment of highly complex deep learning models require optimization at both the hardware and the software levels. In this talk, I will share my experience on enabling efficient inference on embedded systems using two widely adopted software techniques i.e. quantization and pruning. Performing both quantization and pruning is challenging due to accuracy vs complexity trade-off. Additionally, we will review some of the NVIDIA tools to improve end-to-end deep learning workflow with examples and their performance evaluations on NVIDIA’s Jetson embedded platform.