How Microsoft is applying artificial intelligence concepts and technologies to deliver high impact and measurable outcomes across core sectors like agriculture, healthcare and education for India and the rest of the world.
This panel discussion aims to provide a fertile ground to foster technological innovation and create an impact on society. The panel brings together key stakeholders (Microsoft Leadership, Faculty, Industry Experts and students) to discuss the importance of collaborative research, consulting and informal relationships for academiaâ€“industry knowledge transfer. The focus will be on learning and education, and how to engage in advocacy and thought leadership in areas of innovation that impact society and the economy.
The term, related languages, refers to languages that exhibit lexical and structural similarities on account of sharing a common ancestry or being in contact for a long period of time. People speaking these languages generally reside in contiguous geographical areas and communicate heavily amongst themselves for administrative, business and social needs. Hence, machine translation involving related languages is an important requirement. In this talk, I will discuss our work on Statistical Machine Translation (SMT) for related languages, and briefly touch upon ongoing work on Neural Machine Translation.
Limited parallel corpora hinders the ability of Statistical Machine Translation systems to acquire bilingual mappings, and many related languages have very limited parallel corpora available. We show that lexical similarity between related languages (i.e., sharing of words with similar form and meaning) can help improve SMT systems when limited parallel corpus is available. We utilize lexical similarity via subword-level SMT i.e., SMT with units smaller than the word. We propose the usage of two well-motivated subword translation units: linguistically motivated orthographic syllables and statistically motivated byte pair encoded units. Multilingual translation, which uses data from multiple language pairs, is another method for increased acquisition of bilingual mappings by pooling together data from multiple language pairs. We show that when multilingual translation is complemented with subword-level representation, the resulting translation quality in a resource-constrained scenario can match a resource-rich scenario.
Deep Reinforcement Learning methods have achieved significant successes recently by marrying the representation learning power of deep networks and the control learning abilities of RL. This has resulted in some of the most significant recent breakthroughs in AI such as the Atari game player and the Alpha Go engine from Deepmind. This success has opened up new lines of research and revived old ones in the RL community. In this talk, I will introduce the reinforcement learning paradigm and briefly review the progress made in deep RL.
An exciting, interactive session that will examine the importance of having a growth mindset to be able to experiment, innovate and do amazing stuff. The audience will be invited on stage for a hands-on activity that will involve code changes to a drone, deploy the changes using VSTS and make it fly around and do some cool stuff.
Akash Saxena is a software engineer with Visual Studio Team Services at Microsoft India R&D Pvt. Ltd. He takes pride in working on full stack development and is passionate about computer networks. Being a Zumba instructor is on his mind but for now he is busy flying drones and doing other cool stuff in the Microsoft garage.
Anil Bhansali is the Corporate Vice President of Microsoftâ€™s Cloud + AI Platform & Managing Director of Microsoft India (R&D) Pvt. Ltd. Anil holds a post graduate degree in Computer Science from SUNY, Stony Brook and has been with Microsoft from 1991. In a career spanning more than two decades he has worked in multiple technologies in Office, Search, Windows and Azure.
In the India Development Center Anil was responsible for starting and growing several engineering efforts including Search, MSN, Windows Server and Client and Microsoft Azure. In his current role Anil is leading efforts in Microsoftâ€™s Cloud + AI Platform division to drive Azure growth with primary focus on AI to drive digital transformation in key verticals like Healthcare, Agriculture and Education.
He is on the Board of Advisors for the Srini Raju Center for IT and Networked Economy, a research centre at Indian School of Business which aims to foster research in (ICT) to create value for business and society. He is also on the Board of Directors of Hyderabad Eye Care Institute. Anil is committed to the cause of improving the quality of education and health care for the under-served in India.
He is on the board of several non-profits which support initiatives in education and healthcare When not at work, Anil spends time with his wife Shobha and two teenage boys Aditya and Eshaan. He is also an avid trekker he has been to the Annapurna and Everest Base camp. He is also a marathon runner and has completed several national and international marathons.
Anoop Kunchukuttan is an applied researcher at the Search Technology Center - India, Microsoft. He has recently submitted his Ph.D thesis at IIT Bombay. He works on natural language processing and machine learning. His areas of interest include machine translation, machine transliteration and multilingual learning. He has also dabbled with problems in information extraction and automated grammar correction. Home Page: http://www.cse.iitb.ac.in/~anoopk.
Chitra Sood heads the Biz Management function at MSIDC and the Cloud & Enterprise division. Chitra assists the organizationâ€™s leadership team to deliver against its operational and organizational priorities. She has been part of MSIDC since its inception in 1998 and has headed a variety of support functions across the organization. Prior to joining Microsoft, Chitra was the Head of the Hutchison Max Paging operations in Hyderabad. Chitra is a CA/CS and also holds an MBA from the Manchester Business School.
Professor Balaraman Ravindran
Prof. Ravindran is the head of the Robert Bosch Centre for Data Science and AI at IIT Madras and a professor in the Department of Computer Science and Engineering. He is also the co-director of the reconfigurable and intelligent systems engineering (RISE) group at IIT Madras. He received his PhD from the University of Massachusetts, Amherst. He has nearly two decades of research experience in machine learning and specifically reinforcement learning. He has held visiting positions at the Indian Institute of Science, Bangalore, India and University of Technology, Sydney, Australia. Currently, his research interests are centered on learning from and through interactions and span the areas of network analysis and reinforcement learning.
Professor Sandeep Shukla
Professor Sandeep Kumar Shukla is currently Poonam and Prabhu Goel Chair Professor and Head of Computer Science and Engineering Department, Indian Institute of Technology, Kanpur, India. He is currently the Editor-in-Chief of ACM Transactions on Embedded Systems, and associate editor for ACM transactions on Cyber Physical Systems.
Professor Sandeep K. Shukla is an IEEE fellow, an ACM Distinguished Scientist, and served as an IEEE Computer Society Distinguished Visitor during 2008-2012, and as an ACM Distinguished Speaker during 2007-2014. In the past, he has been associate editors for IEEE Transactions on Computers, IEEE Transactions on Industrial Informatics, IEEE Design & Test, IEEE Embedded Systems Letters, and many other journals.
He was a faculty at the Virginia Tech, Arlington, Virginia between 2002 and 2015. In 2014, he was named Fellow of the Institute of Electrical and Electronics Engineers (IEEE) for contributions to applied probablistic model checking for system design. He has authored several books on systems. Professor Sandeep K. Shukla also has been a visiting faculty at INRIA, France and University of Kaiserslautern, Germany.
Sandeep Chadda is a practicing program manager with Visual Studio Team Services at Microsoft India R&D Pvt. Ltd. When he is not playing pranks on his family members, he would like to believe that he is playing a small part in changing the world by creating products that make the world more productive.
Professor Sathya Prasad
Professor Sathya Prasad has spent 25 years in the tech industry and 20 yrs at Intel, spread across US & India and has donned roles across Corporate Innovation, Strategic Planning, Product Management, Marketing and R&D. A significant achievement was leading the Product Management and being a member of Intel Server R&D leadership team that conceived, designed and developed a brand-new product line (Intel Xeon Processor D) which was widely recognized by the industry as an outstanding new product.
As an intrapreneur and Director of New Products/Business, Sathya led incubation efforts at Intel Indiaâ€™s Idea-to-Reality (I2R). He has helped other entrepreneurs under â€˜Intel India Maker Labâ€™ program in mentoring early stage tech startups.
Outside Intel, as the founding President of SEMI India, he built the India subsidiary of global firm SEMI and launched the B2B platform SOLARCON India. Sathya has several publications in Technology Management (co-authored with IIM-B) and is regularly invited for Guest Lectures for Exec Education at IIM-B and other leading business schools. After leaving Intel in 2017 to follow his passion, Sathya joined PES University as the Founding Director for Centre for Innovation & Entrepreneurship (CIE).
Sathya is an alumnus of MIT Sloan School of Management (Executive Program in General Mgmt) & Arizona State University (MS, Electrical Engineering). More details @ https://in.linkedin.com/in/sathyaprasad1.
Vishwa Kumbalimutt, Partner Software Engineer Manager, Bing Sponsored Search Engineering Team.
Vishwa leads the Bings Ads Marketplace team, his team is responsible for Bing monetization for INTL markets where they extensively use Machine learning algorithms for matching user queries to the relevant ads. He started his career as a Software developer working on telephony switches in Siemens and Ericsson. He joined Microsoft in April 1999 and has previously worked on various products including Windows Networking, Web Services protocols, Server Virtualization etc.
His qualification includes a Bachelorâ€™s in Electronics and Communications from University of Mysore. Vishwa lives in Bangalore with his wife and daughters. His passions outside of work includes reading, running and playing squash.
An automated AR navigation system which uses virtual traffic signals and objects to assist users navigate.
A crowdsourced risk analysis platform for sending nice messages to people who may be having a bad day or show signs of depression. Uses Azure Text Analytics' Sentiment analysis service for classification on public tweets to classify tweets as happy/ sad, then a Markov chain on encouraging tweets to generate a happy message to tweet at the person who tweeted something sad.
A smart video player which enables video analytics, actor recognition and content moderation as per userâ€™s preference.
Personalizes movie watching experience on a laptop with simultaneous screening of media content with multiple language support. Ideal for a group of people streaming the same content simultaneously across multiple languages.
Smart Healthcare Companion
The system detects pulse rate of user using regular web cam stream. This helps analyze userâ€™s health during long sessions on computer.
A web app for speech to sign language conversion useful for people with hearing disability.
Smart dustbins that work on micro controllers which collect waste and store data for further analysis. Automatically opens up when someone wants to dump waste and notifies municipality once the dustbin is almost full.
Smart Surveillance System for real time tracking and analysis of CCTV footage in private spaces. By using Object Detection and Face Detection APIs, unauthorized people in an area can be detected and tagged.
The chrome extension allows you to summarise any content. This summarised content is stored and also will be shown on the app of the respective user. The app also allows the user to view his history of summarizations where he can share and delete his summarizations.
Enjoy multiplayer karaoke in virtual reality. Synchron enables users to create/join karaoke rooms. Room members see each otherâ€™s models in the VR environment as they hear each other sing in real time while performing karaoke to the same music, streamed via a synced music player. Karaoke rooms have 3D sound with volume level determined by userâ€™s distance from speakers in the room, navigation for which is empowered via voice commands.
Virtual reality application integrated with Microsoft Azure Machine Learning Studio and Azure cloud for visualising datasets & results of complex machine learning experiments.
Scripty is a Python-based universal app; it can be used to read scanned documents and reflow its content according to device screen size. We present a technique that reflows text in scanned document images in a manner that is agnostic to the script used to compose them. It works mostly for scripts which use white space to delimit words and lines (like English, Hindi, Bengali etc.).
We are building a next gen platform using the latest advancements in ai and ml. We will be integrating computer vision, speech and NLP in a one of it's kind platform.
In today's day and age we are constantly bombarded with news from various sources and our project looks to make some 'new sense' out of this 'nuisance'. Using named-entity recognition and sentiment-analysis on world news involving two nations, we predict the recent developments in the relations between these countries and depict the same in an interactive, graphical form.
Travelio is an an android app developed to solve real world solutions among commuters and travelers, an unfocused sector in today's technology. It uses robust latest technology integrated with a rock solid UI. It consists of variety of solutions, primary features include using Augmented reality embedded with Google places to augment places and show them in realtime. Others include using Geo Fencing to Monitor geofence transitions whenever a device enters or exits a geofence. Predict current popular tracks based on a person's mood, using computer vision to embibe into different cultures and know more about it and realtime speech translations to prevent language barriers.