AI/Machine Learning Winter School 2022

December 12, 2022 - December 23, 2022
Loading Events
  • This event has passed.

Local organizer and Chief Sponsor: Infosys Limited

Instructional Organisers: Sanjeev Arora and Subhashis Banerjee
Application process and outreach: Ashoka University
(December 12-23, 2022, Infosys Campus, Mysore)


Note: Accommodation and travel costs for the selected candidates will be borne by Infosys Ltd.  Interested college teachers should send an application as soon as possible. Selected candidates will be expected to attend the full two weeks. The first round of invitations will be sent in early November.

1. Summary:

Rapid advances in AI and Machine Learning (AI/ML) promise to not only revolutionize computing sciences and the economy, but also to greatly impact many STEM disciplines such as environmental and climate sciences, ecology, economics and other social sciences. India needs a scientific infrastructure for cutting-edge research and training in AI and Machine Learning. This is separate from (and in addition to) the need for a well-trained AI/ML workforce in industry.

Development of this research and scientific infrastructure is hindered by the fact that many existing college faculty were not trained in AI/ML, and also by the fact that the exciting developments around deep learning in the past decade have made old courses obsolete. Furthermore, there is a nation-wide shortage of faculty in computing, and even more so in AI/ML. However, the deep learning revolution of the past decade has the huge plus of making AI/ML very accessible to newcomers with some minimal technical background. The primary goal of the Winter School is to give faculty from leading Indian universities —- whether or not they are from computer science or machine learning—-the knowledge base and tools to start using AI/ML techniques in their research as well as teaching. We will try to continue these schools over next few years. By focusing on training the teachers (including those currently not engaged in AI/ML teaching and research), the Winter School could have a big multiplier effect on the AI/ML ecosystem within a few years.

The 2022 Winter School offers two courses, Introductory Machine Learning, and advanced undergraduate level course on Computer Vision + Natural Language Processing. It will provide sufficient theoretical knowledge and hands-on experience to faculty members to enable them to develop and teach similar courses at their universities, and also to initiate research. Participants will get a certificate of completion upon completing all assignments at the winter school.

Infosys is the primary sponsor and host of this Winter School. The lead organizers are Sanjeev Arora (Princeton University) and Subhashis Banerjee (Ashoka University). The course instructors are faculty at leading universities in the US and India. Graduate preceptors are chosen from leading universities in the US and India. Full list of Faculty (alphabetical order): Pulkit Agrawal (MIT), Chetan Arora (IIT Delhi), Sanjeev Arora (Princeton), Subhashis Banerjee (Ashoka University), CV Jawahar (IIIT Hyderabad), Dinesh Jayaraman (U. Penn), Deepak Pathak (CMU), Sudeshna Sarkar (IIT Kharagpur), Sameer Singh (UC Irvine).

2. Target groups
2.1. Basics Machine Learning

Will teach faculty members at leading Indian universities who wish to become active in teaching and research in AI and Machine Learning. These will often include faculty from fields adjacent to computer science such as engineering, OR, climate science, ecology, economics, management science, etc. The curriculum would assume prior knowledge of:

1. multivariate calculus, linear algebra, and probability

2. computer programming via a modern language, preferably Python

This course will have room for approximately 60 people.

2.2. Computer vision and natural language processing

Computer vision and natural language processing have been hugely impacted by the new developments in AI/ML, and core techniques developed here are adaptable to host of practical applications. The course will give an introduction to the basic techniques in these fields. This course will be targeted  at faculty members from Computer science or Electrical engineering (primarily but not exclusively) who wish to initiate teaching and research in these topics. The curriculum will assume some familiarity with basic machine learning, including deep learning.

This course will have room for approximately 20 people.

3. Course structures

There will be 2.5 hours of lectures per day (by visiting instructors and staff), 2 hours of precept, 4 hours of self-study, discussion, and programming assignments supervised by preceptors. 5 days/week for 2 weeks.

3.1. Introduction to Machine Learning and AI

Tentative course outline: Linear models: regression, logistic, SVMs. Unsupervised learning: K-Means clustering, SVD, PCA, and Feature Selection. Recommender systems. Language Models. Deep Learning: fully connected and convolutional nets, backpropagation, and gradient-based training. Reinforcement Learning. Applications of machine learning and societal/ethical discussions will be worked into the course.

Faculty: Sanjeev Arora (Princeton) and Dinesh Jayaraman (U. Penn).

3.2. Advanced Machine Learning (Computer vision and Natural Language Processing )

All participants will attend one week of lectures on modern deep learning techniques as applied to Computer Vision and NLP. These will be supplemented with hands-on experience in various programming environments. During the second week, the advanced course may divide into separate tracks on Computer Vision + Robotics and NLP, to allow deeper immersion in research topics.

Computer vision: Convolutional neural networks and automatic feature extraction. End-to-end models. Transfer learning and domain adaptation. Object detection frameworks. Semantic segmentation. Video processing. Generative Adversarial Networks (GAN). Reliability, adversarial attacks, and robustness to out-of-distribution (OOD) examples. Recurrent networks. Elements of RL and robotics.

Natural language processing: Basic concepts including language modeling, word embeddings, representation learning, text classification, sequence tagging, syntactic parsing, machine translation, and question answering. Deep language models, and transformer-based architectures.

Faculty: Pulkit Agrawal (MIT), Chetan Arora (IIT Delhi), Subhashis Banerjee (Ashoka University), C V Jawahar (IIIT H), Deepak Pathak (CMU), Sudeshna Sarkar (IIT Kharagpur), Sameer Singh (UC Irvine).

4. Instructors

The instructors for the courses will be faculty member or industry experts from India and US. The courses will be coordinated by

1. Professor Sanjeev Arora, Princeton University

2. Professor Subhashis Banerjee, Ashoka University

5. Application process

Kindly use this link to apply for the Winter School 2022:



December 12, 2022
December 23, 2022


Infosys Campus, Mysore
We are using cookies to give you the best experience. You can find out more about which cookies we are using or switch them off in privacy settings.
AcceptPrivacy Settings