Shruti Joshi

I am a graduate student at Mila with Dhanya Sridhar, and jointly mentored by Devon Hjelm.

The areas of my scientific research interests are causal representation learning and reinforcement learning. I work on making the representations learnt by agents operating in an interaction-based setting identify latent variables which may be statically unobservable, and on learning how to intervene towards learning such representations. I also work on standardized procedures and metrics for evaluating representation and reinforcement learning methods, for example, a current direction I find particularly interesting is evaluating mathematical reasoning abilities of these methods.

In addition to my research interests, I like to learn and incorporate better software design in my research towards well-maintained (!) iterative codebases. I also like to spend my time studying mathematics and some topics from physics such as thermodynamics and statistical mechanics (for some interesting interpretations of causality!).

I co-organise a Causality Study Group at Mila, check it out!

Email / Github / Google Scholar / Twitter

Experience and new stuff

(Winter '23) Very grateful to receive the Women in AI Excellence Scholarship from Mila, the international students' excellence scholarship from the Government of Québec, and an exemption from additional tuition fees for international students.
Happy to serve as a reviewer for ICML'23 and Neurips '23! (and previously for ICLR'22, Neurips'22, and a bunch of workshops)
(Fall '22) I was a visiting researcher at ServiceNow Research with Alexandre Piché.
Grateful to have spent ~ 2 wonderful years as a research programmer at the Max Planck Institute for Intelligent System in Tübingen in the amazing group of Bernhard Schölkopf and Stefan Bauer.
Graduated from the Department of Electrical Engineering at the Indian Institute of Technology Kanpur (IIT Kanpur) in 2019.

Publications

Learning Robust Dynamics through Variational Sparse Gating
Arnav Kumar Jain, Shiva Kanth Sujit, Shruti Joshi, Vincent Michalski, Danijar Hafner, Samira Ebrahimi Kahou.
NeurIPS '22, DRLW '21 Paper

Dynamic Inference with Neural Interpreters
Muhammad Waleed Gondal*, Nasim Rahaman*, Shruti Joshi, Peter Gehler, Yoshua Bengio, Francesco Locatello, Benrhard Schölkopf
NeurIPS '21 Paper Twitter Videos

TriFinger Simulation
Shruti Joshi, Felix Widmaier, Vaibhav Agrawal, Manuel Wüthrich.
2020 Code

Function Contrastive Learning of Transferable Representations
Muhammad Waleed Gondal, Shruti Joshi, Nasim Rahaman, Stefan Bauer, Manuel Wüthrich, Bernhard Schölkopf.
ICML'21 Paper Twitter

TriFinger: An Open-Source Robot for Learning Dexterity
Manuel Wüthrich, Felix Widmaier, Felix Grimminger, Joel Akpo, Shruti Joshi, Vaibhav Agrawal, Bilal Hammoud, Majid Khadiv, Miroslav Bogdanovic, Vincent Berenz, Julian Viereck, Maximilien Naveau, Ludovic Righetti, Bernhard Schölkopf, Stefan Bauer.
CoRL'20 Paper Twitter Project Page Real Robot Challenge

Online Utility-Optimal Trajectory Design for Time-Varying Ocean Environments
N Mohan Krishna, Shruti Joshi, Ketan Rajawat.
ICRA'19 Paper Code

I also contribute to the Open Dynamic Robot Initiative. Some of my contributions include the development of a realtime logging interface for the TriFinger robots, and drivers for image acquisition from the cameras on the TriFinger platforms.

I love his website design.