(CSE Colloquium Lecture Series)
Speaker: Hadi Esmaeilzadeh, Georgia Institute of Technology
Date: Friday, April 7, 2017
Location: Room 1202, CSE Building
Project PHI: System Design for Pervasive Hierarchical Intelligence
Abstract: This talk presents, Project PHI (Pervasive Hierarchical Intelligence) a holistic effort to provide a comprehensive solution for making immersive machine intelligence a reality. Our guiding principle is to retain as much generality and automation while delivering large performance and efficiency gains through specialization and acceleration for a wide range of learning and intelligence workloads. As the first milestones of Project PHI, we have developed Tabla and DnnWeaver, which are open source and publically available (http://act-lab.org/artifacts/tabla/ and http://act-lab.org/artifacts/dnnweaver/). DnnWeaver is the very first open-source hardware acceleration framework for deep neural networks. Tabla is a cross-stack solution—spanning from programming language to the hardware—that rethinks the hardware/software abstraction by delving into the theory of machine learning. It leverages the insight that many learning algorithms can be solved using stochastic gradient descent that minimizes an objective function. The solver is fixed while the objective function changes with the learning algorithm. Therefore, Tabla uses stochastic optimization as the abstraction between hardware and software. Consequently, programmers specify the learning algorithm by merely expressing the gradient of the objective function in our domain specific language. Tabla then automatically generates the synthesizable implementation of the accelerator for scale-out FPGA realization using a set of template designs. Real hardware measurements show orders of magnitude higher performance and power efficiency while the programmer only writes 60 lines of code. These encouraging results show that rethinking the hardware/software abstractions from an algorithmic perspective can open new dimensions in system design for Pervasive Hierarchical Intelligence.
Bio: Hadi Esmaeilzadeh joined the Georgia Tech’s School of Computer Science as the inaugural holder of the Allchin Family Early Career Professorship in 2013. He founded the Alternative Computing Technologies (ACT) Lab to develop cross-stack solutions for building the next generation of computing systems. He obtained his PhD in Computer Science and Engineering from the University of Washington, where he received the 2013 William Chan Memorial Dissertation Award. His work has been recognized by four Communications of the ACM (CACM) Research Highlights, four IEEE Micro Top Picks, one honorable mention in IEEE Micro Top Picks, one nomination for CACM Research Highlights, and a Distinguished Paper Award in HPCA 2016. He has received the Air Force Young Investigator Award (2017), College of Computing Outstanding Junior Faculty Research Award (2017), Google Research Faculty Award (2016 and 2014), Microsoft Research Award (2016), Qualcomm Research Award (2016), and Lockheed Inspirational Young Faculty Award (2016). His students have received the Microsoft Research Fellowship (2016), Qualcomm Innovation Fellowship (2014), and the Georgia Tech President’s Undergraduate Research Award (2015, 2016). His work on dark silicon has been profiled in New York Times. For more information visit: http://www.cc.gatech.edu/~hadi/.
Related Research Publications
TABLA: A Unified Template-Based Framework for Accelerating Statistical Machine Learning, (HPCA 2016 Distinguished Paper Award)
AxGames: Towards Crowdsourcing Quality Target Determination in Approximate Computing, (ASPLOS 2016)
From High-Level Deep Neural Models to FPGAs, (MICRO 2016)
Faculty host: Dean Tullsen (email@example.com)