Artificial Intelligence At Scale

19 December, 2019 - 10:00 am - 12:00 noon

Abstract

The field of Artificial Intelligence (AI) has emerged as the most key technological revolutions of this century. AI has evolved from being a niche research area to one with vast and profound impact on the day to day lives of billions of people all over the world. Today, AI has applications in the fields of Agriculture, Aviation, Education, Entertainment, Finance, Healthcare, Industrial Manufacturing and Transportation among other things. Application of AI as a method to solve previously intractable problems presents a huge opportunity as well as many concerns. One systems level concern in particular is scalability. In this Industry Research User Symposium session on Artificial Intelligence at scale, we want to explore some of these issues, get leaders from academia and industry to share their thoughts, experiences, and best practices on how they are thinking about addressing the problem of scalability in the field of AI, and also discuss how HPC can help power the field of AI to address some of these challenges.

 

Key Questions

  1. What are the biggest challenges facing the field of AI as it scales to more applications and users?
  2. How many of these challenges are due to the limitations of today’s AI infrastructure and/or H/W platforms?
  3. How is the field of AI trying to solve these problems?
  4. The field of HPC has solved similar problems facing the field of AI today but for other applications (like Weather Forecasting, Bio-molecular analysis, etc.,) that share very similar characteristics to typical AI workloads (i.e., data intensive, concentrated kernel computations on the data such as fully connected layers, convolutions, sequence models, embeddings, etc). How can HPC help the field of AI scale?
  5. Can HPC help AI break new ground by tapping into new applications and/or use cases? If so, what are some examples of new AI applications and/or use cases that will be enabled by HPC infrastructure?
  6. Conversely how can AI techniques help the HPC community? What are the similarities? What are the differences?

 

Academia:

       Dr. P J Narayanan, Professor and Director, IIIT Hyderabad

       Dr. Chakravarthy Bhagvati, Professor, Central University, Hyderabad

      Tridib Roy Chowdhury,  Chief Programme Officer of the Machine Intelligence and Robotics Center (MINRO Center), IIIT Bangalore

 

 

Industry:

       Dr. Ashish Tendulkar, AI/ML Specialist, Google

       Sandeep Kohli, VP of Engineering, Flipkart

       Biswajit Mishra, Senior Deep Learning R&D Engineer, Intel

 

 

 

Moderator: Sundar Dev, Senior Software Engineer, Google Platforms Engineering, Sunnyvale

 

Format

The IRUS session will start with each speaker sharing their thoughts on the topic (focusing on the key questions listed above). This will be followed by a Q&A session with questions from the moderator and the audience. This session is currently scheduled on 19th December, 2019 from 10:00 am to 12:00 noon.

 

Speaker Bios

 

Dr. P J Narayanan

P J Narayanan is the Director of IIIT, Hyderabad and a researcher in the areas of 3D vision, computational cameras, and parallel computing. He built the Virtualized Reality system in mid 1990s at the Carnegie Mellon University to capture 3D geometry and appearance of dynamic events. He also was an early adopter of GPUs for several Computer Vision and general computing tasks. He got his bachelors from IIT Kharagpur, masters and PhD from the University of Maryland, all in Computer Science. He was a research faculty member at the Robotics Institute of CMU from 1992 to 1996 and headed the Vision and Virtual Reality groups of the Centre for Artificial Intelligence and Robotics, Bangalore from 1996 to 2000. He has been with IIIT Hyderabad from 2000 and has been its PG Coordinator, Dean of Research, and, from 2013, the Director. He was the President of ACM India from 2009 to 2014 and is currently the Chair of its Research Board.

 

Biswajit Mishra

Biswajit Mishra is currently working as Senior Deep Learning R&D Engineer in Intel Corporation, Bangalore where his main focus is research, develop and optimize deep learning algorithm for Intel’s  Nervana Neural Network Processor. Prior to Intel he was in Shell technology Center Bangalore for 6 years where he worked High Performance Computing and Deep Learning applications for subsurface data. Prior to that he was working in Tata’s Computational Research Laboratories where he was involved in research and development of optimized parallel library for EKA supercomputer. Biswajit Mishra had received his M.Tech from IISc Bangalore and B.Tech from IIT, Dhanbad.

 

Sandeep Kohli

Sandeep Kohli works as VP of Engineering in Flipkart and is responsible for building platforms for Flipkart. Over the last 5 years in Flipkart he has been responsible for building a complex event processing platform, User Graph Services, Machine Learning Platform, Streaming platform and Big Data Platform.  Sandeep Kohli has 22+ years of industry experience. He holds a masters degree from Indian Institute of Science.  Prior to Flipkart he has worked for C-DAC, Rational , IBM and Numerify and also tried hands at his own startup.

 

Dr. Ashish Tendulkar

Ashish is a machine learning architect at Google Pay.  Ashish is interested in solving real world problems with Machine Learning.  He has 15+ years of experience in developing ML systems in different domains and academic research.  Ashish was an Assistant Professor in the Department of Computer Science and Engineering at IIT Madras prior to Google.  He has also worked with Persistent Systems Ltd., Reliance Industries, Media.net and Tata Institute of Fundamental Research in his career spanning 20 years. Ashish holds M. Tech and PhD from IIT Bombay.

 

Dr. Chakravarthy Bhagvati

Dr. Chakravarthy Bhagvati is currently a professor in the School of Computer and Information Sciences, University of Hyderabad. His research interests are Image processing, Colour image processing, Pattern recognition and, of late like everyone else in image processing and pattern recognition, deep learning. Chakravarthy received his PhD in Computer Science from Rensselaer Polytechnic Institute, Troy, New York (USA) and has been working at the University of Hyderabad since 1998. Chakravarthy worked as Professor-in-Charge of the Centre for Modelling, Simulation and Design (CMSD) at the University of Hyderabad which houses several high-performance computers; and was a Member, Advisory board for establishing a high-performance computing centre by the Govt. of Rajasthan. He is also a life member of the Indian Unit of Pattern Recognition and Artificial Intelligence (IUPRAI).

Outside of research, Chakravarthy is keenly interested in enthusing young students to explore the joys of Computer Science and is also an avid amateur astronomer armed with an 8-inch telescope!

 

Tridib Roy Chowdhury

Tridib Roy Chowdhury has over 30 years of experience in starting and building software businesses for startups and large MNCs in India and USA with a history of 11 new global product launches and 80+ product releases and starting 6 successful product businesses and multiple patents.  He is a B-Tech (Computer Science) from IIT-Kharagpur and has completed M.S (Computer Science- Intelligent Databases) from University of Maryland, College Park.  He is currently enrolled at IIIT Bangalore as a full time research student pursuing another post-graduate degree and heads the MINRO COE @ IIITB, working on fundamental research in AI and robotics.
He is a keen follower of the advances in machine learning, databases, digital marketing and design thinking and passionate about advancing technology in education and mentoring and building high performance teams.

 

 

 

 

 

 

 

 

 

Scroll Up