Lab Representatives

About the Lab Reps

Lab Reps are a point of contact to facilitate communication between/among students, interns, postdocs, faculty staff, as well as to enhance student involvement in decision-making processes.

The Lab Reps are elected every year by the students at Mila. While the specific initiatives undertaken by Lab Reps vary each year, depending on current context and interest of the members, they are always working to improve the Mila community.

Some of the Lab Reps roles and responsibilities include:

  • Supporting students (channeling feedback / helping solve problems)
  • Providing feedback on Information Technology infrastructure (clusters, wifi, etc.)
  • Managing social activities (reading groups, tea talks, socials)
  • Managing physical spaces (relaxation room, student lounge)
  • Discussing issues around Equity, Diversity and Inclusion
  • Developing conflict resolution procedures
  • Voicing opinion about prospective partnerships
  • Voicing opinion about AI for Humanity research projects

Students in need of advice or resources are welcome and encouraged to reach out to the Lab Reps at any time: reps@mila.quebec

Arna Gosh

Indian / PhD / ghosharn@mila.quebec

My research entails building brain-inspired AI systems, specifically self-supervised learning systems. I am currently working on biologically plausible self-supervised learning for training deep networks that can better mimic the brain’s visual processing pathway. Additionally, my work aims to explore biologically-plausible credit assignment algorithms in deep neural networks with the hope of demystifying the computational framework of learning in the brain. I am also interested in exploring reinforcement learning strategies inspired by insights and theories about human learning, eventually leading to biologically motivated artificial learning systems.

Supervisor : Blake Richards

Fun Fact : I was more into theater and cricket before choosing science.

Arushi Jain

Indian / PhD / jainarus@mila.quebec 

I am working in the area of Reinforcement Learning (RL) particularly risk-averse RL. I have been looking at variance based techniques for risk reduction. Recently, I started looking in the area of Constrained MDP for extending RL to real-world usage where constraints are useful for limiting the undesirable behavior of the RL agent.

Supervisors : Doina Precup, Pierre-Luc Bacon

Fun Fact : I am a plant enthusiast.

Jose Gallego

Colombian / PhD / jgalle29@gmail.com

I am working towards generalizing standard information theory to geometric domains and the application of these ideas in machine learning problems.

Supervisor : Simon Lacoste-Julien

Fun Fact : When I was young, I used to think that Harper Lee’s book was titled “Tequila Mockingbird”.

Manuela Girotti

Italian / Postdoc / girottim@mila.quebec

Pure Math may shine a light in understanding theoretical Deep Learning! My research is focused on the application of Random Matrix Theory and Integrable Systems to generalization analysis and optimization techniques.

Supervisor : Ioannis Mitliagkas

Fun Fact : I met my PhD mentor when singing in a Gregorian choir in Milano.

Martin Weiss

American / PhD / martin.clyde.weiss@gmail.com

I have somewhat diverse interests, but try to work on high-impact social good projects. In my primary research project, I am building a mobile navigational system that uses multi-modal inputs (images, text, gps, etc) to assist blind and visually impaired people with an outdoor address finding task. This past year, I put that project on hold to build COVID-19 contact tracing applications with Yoshua Bengio. Previously, I interned with Yoshua and Joseph P. Cohen on Gene Graph Convolution Neural Networks for predicting clinical outcomes in cancer patients. I’m also interested in systematic generalization.

Supervisor : Christopher Pal

Fun Fact : I produce electronic music for fun. https://soundcloud.com/rmfrstar/accelerando

Nikolaus H. R. Howe

Canadian / M.Sc. / howeniko@mila.quebec

My research focuses on learning to model and behave optimally in dynamical systems using RL, constrained optimization, and Neural ODE. Among other applications, this could help us improve the efficiency of energy generation and HVAC as we move towards a sustainable future.

Supervisor : Pierre-Luc Bacon

Fun Fact : I am a huge fan of stationery, especially pens.

Raghav Gupta

Indian / Professional M.Sc. / raghav0296@gmail.com

I am interested in applied research in Representation Learning and Computer Vision. I am currently adapting self-supervised approaches to segmentation and object detection for obtaining better pre-trained models, making downstream task training more efficient and yielding high-performing models for devices with limited computing capability.

Fun Fact : I’m a dribbler and playmaker in football, a competitive gamer in Call of Duty, and cofounder of a startup venture.

Rim Assouel

French/Moroccan / PhD / assouelr@mila.quebec

I’m mainly interested in object-centric representations in computer vision! A core level of abstraction at which humans perceive and understand the world. I plan on giving an overview of current approaches to learn object-centric representations, if you’d like to help please reach out!

Supervisor : Yoshua Bengio

Fun Fact : Back in Paris, I used to take shifts as a firefighter once every 2 weeks. I miss the adrenaline!

Sékou-Oumar Kaba

Canadian / M.Sc. / kabaseko@mila.quebec

My research focuses on how to make machine learning algorithms that can help us understand physics problems. More specifically, predicting the properties of materials before having to perform experiments in a laboratory is a challenge for which artificial intelligence could help in a significant way. These algorithms could greatly accelerate the design process of new materials. I am also interested in studying how machine learning could help us model complex systems evolving in time.

Supervisor : Siamak Ravanbakhsh

Fun Fact : I hosted a radio show in Montréal! 

Tegan Maharaj

Canadian / PhD / tegan.jrm@gmail.com

My research interests are to understand how and why learning behavior varies with learning environment (data, regularization, losses/meta-losses, state space, etc.), in order to ensure AI is developed safely and responsibly. I am especially interested in time-varying data (nonstationarity, out-of-distribution generalization, dynamical systems modeling), and in determining/quantifying what constitutes or has an important influence on the learning environment. I’m involved with applications of my research to climate, ecology/environment, epidemiology, and responsible AI policy.

Supervisor : Christopher Pal

Fun Fact : I used to be very awkward and clumsy, but loved climbing trees. Finally, I discovered a sport for me (bouldering!)

Victor Schmidt

French / PhD / schmidtv@mila.quebec

My research interest is about AI vs Climate change: generative models and practical tools. I am co-leading a project named Visualizing the Impacts of Climate Change, to create vivid and relatable visualization of the potential climate-related extreme events: floods, smog, wildfire etc. Also, i’m working on CodeCarbon project to measure the carbon emissions of ML and cloud modeling with GANs.

Supervisor : Yoshua Bengio

Fun Fact : I tend to spend 3h automating tasks which could have taken me 10 minutes to do manually.

array(1) { ["wp-wpml_current_language"]=> string(2) "en" }