The Machine Learning Lunch is a weekly 20-minute talk series for folks in the Vector Institute and University of Toronto ecosystem who are interested in machine learning. The talks are intended for a broad ML audience, with the goal of appealing to anyone who is interested in how to learn from data. Presenters are graduate students from multiple groups and research areas, covering the spectrum between theory and applications. Lunch is provided.
When: Fridays from 12pm - 1pm (with a few exceptions, see below schedule)
Where: Schwartz Reisman Institute, 11th Floor.
Who: You, if you are affiliated with the Vector Institute or the University of Toronto and interested in machine learning!
Food: Free, provided by our sponsors.
Slack: Join #ml-lunch-vector on the Vector Institute slack for updates!
Mailing list: Join our google group to receive updates via email
Contact: ml-lunch-committee@cs.toronto.edu
Attendance Form (optional): let us know when you're coming!
Upcoming Speakers
Date | Presenter | Title |
---|---|---|
Sep. 5, 2025 | Daniel Eftekhari | On the Importance of Gaussianizing Representations |
Sep. 11, 2025 (THURSDAY) | Jonas Guan | Adaptive Computer Viruses |
Sep. 19, 2025 | Jackie Zou | El Agente: An Autonomous Agent for Quantum Chemistry |
Sep. 26, 2025 | Haonan Duan | Measuring Scientific Capabilities of Language Models with a Systems Biology Dry LabD |
Oct. 3, 2025 | TBD | TBD |
Oct. 10, 2025 | TBD | TBD |
Oct. 16, 2025 (THURSDAY) | TBD | TBD |
Oct. 24, 2025 | TBD | TBD |
Oct. 31, 2025 | TBD | TBD |
Nov. 7, 2025 | TBD | TBD |
Nov. 13, 2025 (THURSDAY) | TBD | TBD |
Nov. 21, 2025 | TBD | TBD |
Nov. 28, 2025 | TBD | TBD |
Dec. 12, 2025 | TBD | TBD |
Format and Theme
The goal of the lunch talks is to introduce, educate, and discuss new research problems. As such our suggested theme for the talks is Mysteries & Discoveries, with the following overall format:.
-
Student talk (~20 minutes):
- Mysteries (~10 minutes): What is the research question, and why should the rest of the ML community be interested? Think of the background and broader impact sections in a paper. Consider explaining what is counterintuitive about the problem.
- Discoveries (~10 minutes): What do people know about this problem? The results may be derived from your own or previous work. Take a critical eye towards the presented results and discuss their limitations as well.
- Audience questions (~10 minutes)
- Socializing, unstructured discussion, and finishing lunch (30 minutes)
Student talks may vary from the above theme, so long as the spirit of educating others of a research problem is adhered to.
Policy on Speakers: While anyone can sign-up to give a lunch talk, we will prioritize giving slots to:
- local researchers
- students and post-docs
How do I get involved?
You can find out about new talks as they happen by joining the #ml-lunch-vector channel on the Vector Institute slack or our google group.
If you are a grad student and would like to present, please email us at ml-lunch-committee@cs.toronto.edu, or ping one of the organizing committee members on the Vector Institute slack. We will follow up with you to schedule your talk!
We highly encourage all grad student attendees of the ML Lunch talks to sign up as presenters, to ensure that the talks represent the full machine learning community at the Vector Institute and the University of Toronto. Don't worry if you don't have a polished presentation yet, we are happy to accept talks on in-progress work and open questions, and can work with you to ensure your talk appeals to a broad audience!
Organizing Committee
- Anvith Thudi (Executive Chair)
- Viet Nguyen (Master of Ceremonies)
- Ella Rajaonson (Communications Chair)
- Quentin Clark (Chief Secretary)
- Jay (Gyung Hyun) Je
- Chris J. Maddison
Faculty Sponsors
- Rahul Krishnan
- Igor Gilitschenski
- Chris J. Maddison
- Sheila McIlraith
- Colin Raffel
Institute Sponsors
- Vector Institute
- Acceleration Consortium
Former Committee Members
- Daniel Johnson
- Mert Vural
Past Speakers
Date | Presenter | Title |
---|---|---|
April 14, 2025 | Yangjun Ruan | Reasoning to Learn from Latent Thoughts |
April 7, 2025 | Rahul Krishnan | Making Sense of a Decade of Deep Learning |
March 31, 2025 | KC Tsiolis | Sample Complexity Lower Bounds for Neural Networks |
March 28, 2025 | Dirk Englund | Engineering Intelligence into (Quantum-)Physical Systems |
March 17, 2025 | Austin Cheng | Can you predict a molecule's structure from its shape? |
March 10, 2025 | Olive Franzese | Making ML Auditing Possible with Cryptographic Verification |
March 3, 2025 | Nick (Hengrui) Jia | What is unlearning and why it's often hard |
Feb. 24, 2025 | Chuning Li | Explaining Scaling Laws |
Feb. 10, 2025 | Marta Skreta & Lazar Atanackovic | The Superposition of Diffusion Models |
Feb. 3, 2025 | Claas Voelker | RL: what can we do, and why is it so hard? |
Jan. 20, 2025 | Vahid Balazadeh | Data-Driven Algorithms for Causal Decision-Making with Hidden Contexts |
Jan. 13, 2025 | Sebastian Aegidius | Learning Where to Walk in the Wild |
Nov. 29, 2024 | Sierra Wyllie | Models Cause Feedback Loops in Data, What's Happening? |
Nov. 22, 2024 | Michael Cooper | Survival Analysis: Frontiers and Application to Liver Transplant Prioritization |
Nov. 15, 2024 | Nikita Dhawan | Scaling Causal Inference for Real-World Healthcare Challenges |
Nov. 1, 2024 | Alireza Mousavi | Towards a Theory of Feature Learning Under Adversarial Robustness |
Oct. 25, 2024 | Andreas Burger | Smaller, Faster, Better Forces on Molecules with Deep Equilibrium Models |
Oct. 18, 2024 | Jonas Guan | Deep Learning Has Issues, Can Biology Help Fix Them? |
Oct. 10, 2024 | Ayoub El Hanchi | Does Deep Learning Require Rethinking Generalization? |