Events Calendar

EE 598 Cyber-Physical Systems Seminar Series

Speaker: Yanzhi Wang, Syracuse University

Talk Title: Deep Neural Network and Deep Reinforcement Learning: Ultra-Low Energy Implementation and Broad Applications

Abstract: Recently, deep convolutional neural networks (DCNNs) have made unprecedented progress, achieving the accuracy close to, or even better than human-level perception in a variety of tasks. There is a timely need to map the latest software-based DCNNs to application-specific hardware, in order to achieve orders of magnitude improvement in performance, energy efficiency and compactness. Stochastic computing (SC), as a low-cost alternative to the conventional binary computing paradigm, has the potential to enable massive parallel and highly scalable hardware implementation of DCNNs. The first part of my presentation is a holistic design and optimization framework of SC-based DCNN systems from key arithmetic operations, function blocks, feature extraction blocks, to the overall LeNet5 structure, achieving ultra-low hardware footprint and energy consumption.

Deep reinforcement learning (DRL) has been recently invented and has been successfully utilized in AlphaGo, game playing, etc. Deep reinforcement learning has the potential of control of complicated systems with high state and action spaces (which cannot be achieved by traditional reinforcement learning techniques), thereby resulting in very wide application domains. The second part of my presentation first provides a formal statement of the DRL framework. Effective hardware implementation of the DRL framework, which is critical in the embedded control systems and IoTs, will be investigated. The more broad applications of the emerging technique will be discussed with sample examples on cloud computing and smart grid applications. Open questions and future directions will be finally presented.

Finally I will briefly present the recent work on Luminescent Solar Concentrator-based PV cells and application on electric vehicles, which is transparent and flexible and fits the streamlined surface and aesthetic requirement of modern vehicles. The proposed system can help propel the vehicle or charge the vehicle whenever solar energy is available.

Biography: Yanzhi Wang is currently an Assistant Professor at Syracuse University, starting from August 2015. He received B.S. degree from Tsinghua University in 2009 and Ph.D. degree from University of Southern California in 2014, under supervision of Prof. Massoud Pedram. His research interests include low-power circuit and systems design, neuromorphic computing, embedded systems and wearable devices, etc. He has received best paper awards from International Symposium on Low Power Electronics Design 2014, International Symposium on VLSI Designs 2014, top paper award from IEEE Cloud Computing Conference 2014. He has two popular papers in IEEE Trans. on CAD. He has received multiple best paper nominations from ACM Great Lakes Symposium on VLSI, IEEE Trans. on CAD, and Asia and South Pacific Design Automation Conference.

Host: Paul Bogdan

Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248

Audiences: Everyone Is Invited

Contact: Estela Lopez