Ali Rahimi-Kalahroudi


My name is Ali Rahimi Kalahroudi (Written as "_علی رحیمی‌کلهرودی_" in my native language—Persian). I am currently a research assistant at Mila - Quebec AI under the supervision of Prof. Sarath Chandar.
My ultimate research goal is to develop AI agents capable of solving diverse tasks given minimal supervision.

I am currently interested in designing reliable machine learning (ML) algorithms for sequential decision-making in open-ended interaction settings. In these contexts, the algorithms must learn and continually adapt their skills and knowledge. To tackle these challenges, my approach centers on leveraging model-based learning mechanisms. This involves empowering AI agents with the capability to construct and utilize world models and employing planning to enhance their decision-making processes.

Bio: Perviously I received my M.Sc. in computer science at the University of Montreal and Mila.

Hobbies: While I am not doing science, I enjoy reading books, watching movies, playing football and volleyball, and hiking.

CV Mail Scholar Github LinkedIn

Profile picture

News

Mar 15, 2024  I am honored to share that my master's thesis has been recognized as the best master's thesis
by the Canadian Artificial Intelligence Association (CAIAC).
Sep 15, 2023  My M.Sc. thesis was accepted with an Exceptional Grade (extremely rare at Mila/UdeM).
May 15, 2023  Our work on introducing the LoFo Buffer for adaptive deep MBRL methods got accepted (Oral Presentation) at CoLLAs 2023.
(Also presented at the Neurips 2022 Workshop on Deep Reinforcement Learning.)

Publications

* denotes equal contribution.

Project image
Partial Models for Building Locally Adaptive Deep Model-Based Reinforcement Learning Agents
Safa Alver, Ali Rahimi-Kalahroudi, Doina Precup
Proceedings of the Conference on Lifelong Learning Agents (CoLLAs), 2024
Paper /
@InProceedings{safa2023partialloca, 
	author = {Safa Alver and Ali Rahimi-Kalahroudi and Doina Precup}, 
	title = {Partial Models for Building Locally Adaptive Deep Model-Based Reinforcement Learning Agents}, 
	booktitle = {Proceedings of the Conference on Lifelong Learning Agents (CoLLAs)}, 
	year = {2024}, 
}
Project image
Replay Buffer with Local Forgetting for Adapting to Local Environment Changes in Deep Model-Based Reinforcement Learning (Oral Presentation)
Ali Rahimi-Kalahroudi, Janarthanan Rajendran, Ida Momennejad, Harm Van Seijen, Sarath Chandar
Proceedings of the Conference on Lifelong Learning Agents (CoLLAs), 2023
Also at: Neurips 2022 Deep Reinforcement Learning Workshop
Paper / Video / Poster / Code /
@InProceedings{rahimi2023replay, 
	author = {Ali Rahimi-Kalahroudi and Janarthanan Rajendran and Ida Momennejad and Harm Van Seijen and Sarath Chandar}, 
	title = {Replay Buffer with Local Forgetting for Adapting to Local Environment Changes in Deep Model-Based Reinforcement Learning}, 
	booktitle = {Proceedings of the Conference on Lifelong Learning Agents (CoLLAs)}, 
	year = {2023}, 
}
Project image
Towards Evaluating Adaptivity of Model-Based Reinforcement Learning Methods
Yi Wan*, Ali Rahimi-Kalahroudi*, Janarthanan Rajendran, Ida Momennejad, Sarath Chandar, Harm Van Seijen
Proceedings of the International Conference on Machine Learning (ICML), 2022
Also at: ICLR 2022 workshop on Agent Learning in Open-Endedness (Spotlight), and RLDM 2022
Paper / Video / Poster / Code /
@InProceedings{pmlr-v162-wan22d, 
	author = {Yi Wan and Ali Rahimi-Kalahroudi and Janarthanan Rajendran and Ida Momennejad and Sarath Chandar and Harm Van Seijen}, 
	title = {Towards Evaluating Adaptivity of Model-Based Reinforcement Learning Methods}, 
	booktitle = {Proceedings of the International Conference on Machine Learning (ICML)}, 
	year = {2022}, 
}

Talks

Project image
Replay Buffer with Local Forgetting for Adapting to Local Environment Changes in Deep MBRL
Conference on Lifelong Learning Agents (CoLLAs), McGill University, Montreal, 2023
Slides
Website template adopted from Michael Niemeyer.