Short Bio [In third person]

Aadirupa Saha is an Assistant Professor in the Department of Computer Science at the University of Illinois Chicago (UIC), where she belongs to the UIC CS Theory group, and a member of the IDEAL Institute. Prior to this, she was a Research Scientist at Apple MLR, working on Machine Learning theory, and a short-term visiting faculty at the Toyota Technological Institute at Chicago (TTIC). She completed her postdoctoral research at Microsoft Research (NYC) and earned her PhD from the Indian Institute of Science (IISc), Bangalore.

Her primary research focuses on AI alignment through Reinforcement Learning with Human Feedback (RLHF), with applications in language models, assistive robotics, autonomous systems, and personalized AI. More broadly, she works on Machine Learning theory, including online learning, multi-armed bandits, reinforcement learning, optimization, federated learning, differential privacy, and mechanism design. Her research aims to develop robust and scalable AI models for sequential decision-making under uncertain and partial feedback.

Aadirupa has organized several workshops and tutorials in recent years, including a [NeurIPS, 2023] tutorial on Preference Learning, a [UAI, 2023] ] tutorial on Federated Optimization, two tutorials at [ECML, 2022] , [ACML, 2021], two ICML workshops [ICML, 2023] and [ICML, 2022], and two TTIC workshops [TTIC, 2023] and [TTIC, 2022]. In addition, Aadirupa has also served in several panel discussions and senior reviewing committees for major ML conferences.

Tutorials:


Do you Prefer Learning with Preferences? [Tutorial Website] [NeurIPS Website]
With Aditya Gopalan. Our (amazing) panel: Yoshua Bengio · Craig Boutilier · Elad Hazan · Robert Nowak · Tobias Schnabel
37th Conference on Neural Information Processing Systems (NeurIPS), New Orleans. Dec 11th, 2023.
Online Optimization meets Federated Learning. [UAI Website] [Tutorial Recording]
With Kshitij Kumar Patel, TTIC.
39th Conference on Uncertainty in Artificial Intelligence (UAI), Pittsburgh. July 31st, 2023.
ML Beyond Rewards: Online Learning with Preference Feedback. [Tutorial Website]
European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD). September, 2022.
Battle of Bandits: Online learning from Preference Feedback. [Tutorial Website]
Asian Conference of Machine Learning (ACML). November, 2021.
Bandits for Beginners. [Video link]
Microsoft Reactor: Data Science and Machine Learning Track. November, 2021.
Preference based RL. [Video link]
RL Track, Microsoft Research Summit. October, 2021.
Short Tutorial: (1). Support Vector Machines, (2). Winnow and Perceptron Algorithms.
M.S. Ramaiah Institute of Technology, Bangalore. May, 2018.
Let's Tame the Bandits! [Video link]
Undergraduate Summer School, CSA department, IISc Bangalore. July 2018.

Panel:


Preference based Learning through a Critical Lens [Tutorial Website]
Do you Prefer Learning with Preferences? (at NeurIPS'24 Tutorial). December, 2023.
Next decade of Federated Learning and role of Theory [Workshop Website]
New Frontiers in Federated Learning, September, 2023.
Trusted and Trustworthy AI [Summit Website]
The Summit on AI in Society at the University of Chicago, October, 2022.

Invited Talks:


Principled Methods for Leveraging Human Feedback towards AI Alignment
IDEAL Annual Meeting and Industry Day, UIC Chicago. June 2024
Online Federated Learning
Federated and Collaborative Learning Workshop, Simons Institute, UC Berkeley. July 2023 [Talk Recording]
Dueling-Opt: Convex Optimization with Relative Feedback
IFDS Seminar, University of Wisconsin–Madison. October, 2022
Fall OSL Seminar, Northwestern University. October, 2022
Research at TTIC Series. Toyota Technological Institute at Chicago (TTIC), October, 2022
Theory Seminar, CS, Purdue University. November, 2022
Samueli CS dept Seminar, UCLA. June, 2023
Personalized Prediction Models with Federated Human Preferences
TILOS Seminar, University of California, San Diego. November, 2023
CATS Seminar, University of Maryland. November, 2023
ML-Opt Seminar, University of Washington. October, 2023
Efficient and Optimal Algorithms for Contextual Dueling Bandits under Realizability
UMich AI Symposium. October, 2023
Theory-ML Seminar, CS dept, Carnegie Mellon University (CMU). August, 2023
RL Theory Seminar. May, 2022 [Talk Recording]
Talks at TTIC Series. Toyota Technological Institute at Chicago (TTIC), August, 2022
CS Seminar, Northwestern University. October, 2022
ML Seminar, University of Illinois, Chicago (UIC). October, 2022
University of Illinois Computer Science Speaker Series, UIUC. October, 2022
ML Seminar, UChicago. October, 2022
Adversarial Dueling Bandits
NASSCOM AI Gamechangers. April, 2022
Data Science in India, KDD Conference, India, August 2021.
Information Aggregation from Unconventional Feedback
Oracle Research, November, 2021.
Chalmers University of Technology, November, 2021.
Battle for Better: When and How Can We Learn Faster with Subsetwise Preferences?
Bangalore Theory Seminar, IISc Bangalore. October, 2023
Spring Seminar, UT Austin. March 2023
ISyE Seminar, Goergia Tech. March 2023
The Institute for Data, Econometrics, Algorithms, and Learning (IDEAL) Talk Series. October, 2022
Preference based Reinforcement Learning (PbRL)
Microsoft Research Tri-Lab Offsite. November, 2021
RL Track, Microsoft Research Summit. October, 2021
Battling Bandits: Exploiting Subsetwise Preferences
Sabarmati Seminar Series, IIT Gandhinagar. July 2021.
SIERRA-Seminar, Inria, Paris. January 2020.
Microsoft Research, Bangalore, India. October 2019.
EECS department, University of Michigan, Ann Arbor. September, 2019
Computer Science department, Stanford University, Serra Mall, Stanford. August, 2019
EECS Symposium, IISc Bangalore. April, 2019.
Carnegie Mellon University (CMU), Pittsburgh. March, 2019
Qualcomm Research, Bangalore. May, 2018
Bandits, Experts and Rank Aggregation
TCS Research Lab, Bangalore. June, 2018
Indian Institute of Technology (IIT) Madras. November, 2018
Amazon, Bangalore. October, 2018
IBM-IRL, Bangalore. July 2018.
Online Learning with Structured Losses
Conduent Labs, Bangalore. October, 2017.