Microsoft Research (MSR) is organising the fourth edition of the Academic Research Summit, in partnership with the Association for Computing Machinery (ACM) India and the Robert Bosch Centre for Data Science and AI at IIT Madras. The summit will be held on the 24th and 25th of January 2019 at the ICSR Auditorium in IIT Madras.
AI researchers are striving to create intelligent machines that complement human reasoning and enrich human experiences and capabilities. At the core is the ability to harness the explosion of digital data and computational power with advanced algorithms that extend the ability for machines to learn, reason, sense, and understand—enabling collaborative and natural interactions between machines and humans. The Academic Research Summit 2019 is focused on the theme of Data Science and AI.
The agenda of the summit will include keynotes, talks and other tracks from distinguished researchers from India and across the world. Among the speakers expected at the event are , Raghu Ramakrishnan, Sunita Sarawagi, Sriram Rajamani, Amit Sharma, Chandu Thekkath, Varun Aggarwal etc.
More details on the list of speakers, sessions and the complete agenda will be posted here soon.
Participation at the summit is via invitation and prior registration only.
For any queries about the summit, please email: firstname.lastname@example.org
Broad agenda as below (subject to changes). More details will be updated soon.
Thursday, 24th January 2019
09:00 – 09:30 Registration
09:30 – 10:15 Welcome session
10:15 – 11:15 Plenary Talk: Raghu Ramakrishnan , CTO for Data & Technical Fellow, Microsoft
Title: Cloud + Data + ML: Opportunities and Challenges
Abstract: Breakthroughs in machine learning, scale-out storage and analytics, and cost-effective cloud services are creating a fundamental shift in computing. Data is now a core asset. AI is becoming a foundational tool in realizing value from data. These changes offer us an opportunity to rethink business strategies and scientific enterprises in radically different, data-driven ways.
In this talk, I will discuss the current technology landscape, the opportunities to bring ML to bear on a range of data-driven tasks, and the challenges of responsible data governance when doing so on high-value and sensitive data.
11:15 – 11:45 Break
11:45- 12:45 Plenary Talk: Sriram Rajamani, Managing Director, Microsoft Research India
Title: Program Synthesis meets Machine Learning
Abstract: We give a tutorial overview of program synthesis, from its first formulation by Church in 1957, through its pragmatic evolution through sketching and programing-by-examples, and compare program synthesis with supervised machine learning. We then present our recent efforts in combining program synthesis and machine learning techniques to solve the problem of synthesizing extractors from heterogeneous data. Finally, we explore several opportunities at the intersection of program synthesis (and more broadly the PL community) and machine learning, such as pruning and ranking programs during synthesis, neural program synthesis and automatic differentiation.
12:45 – 14:00 Lunch
14:00 – 15:45 Track: Systems Support for AI, Moderated by Chandu Thekkath, Microsoft
15:45: 16:05 Break
16:05 – 16: 50 Panel Discussion
16:50 – 18:00 Poster Session
18:00 – 19:00 Travel to Dinner
19:00 – 22:00 Dinner
Friday, 25th January
09:30 – 10:30 Plenary Talk: Sunita Sarawagi, Professor, IIT Bombay
Title: Redesigning Neural Architectures for Sequence to Sequence Learning
Abstract: The Encoder-Decoder model with soft-attention is now the defacto standard for sequence to sequence learning, having enjoyed early success in tasks like translation, error correction, and speech recognition. In this talk, I will present a critique of various aspect of this popular model, including its soft attention mechanism, local loss function, and sequential decoding. I will present a new Posterior Attention Network for a more transparent joint attention that provides easy gains on several translation and morphological inflection tasks. Next, I will expose a little known problem of mis-calibration in state of the art neural machine translation (NMT) systems. For structured outputs like in NMT, calibration is important not just for reliable confidence with predictions, but also for proper functioning of beam-search inference. I will discuss reasons for mis-calibration and some fixes. Finally, I will summarize recent research efforts towards parallel decoding of long sequences.
10:30 – 11:20 Industry talk: Varun Aggarwal, Co-Founder, Aspiring Minds
Title: Using AI in the industry, sprucing up academia
Abstract: I will discuss how we have used AI over the last 7 years to build tools for grading skills and providing feedback. I will discuss examples of grading programs, video interviews and simulating chats. Through these, I will demonstrate the opportunities in using AI in the industry, the challenges and pitfalls. In the latter part of the talk, I will talk about the gaps in science policy in India based on my recent book. These seriously impede India’s AI efforts, which need to be immediately addressed.
11:20 – 11:45 Break
11:45 – 13:45 Track: AI for Societal Impact , Moderated by Amit Sharma, Microsoft Research
13:45 – 14:30 Lunch
14:30 – 15:45 Track: Technologies for India, Moderated by ACM India
15:45 – 16:15 Wrap-up & Closing