Félix Therrien

Sharing this AMAZING opportunity:

A postdoc in AI for Materials Discovery at @mila-quebec.bsky.social, with @alexhergar.bsky.social & @drolnick.bsky.social -- the group that took me in during my PhD. Exciting research topic, fantastic Professors, climate change mitigation = strongly recommended :)

📣 Postdoc call!

We are looking for a postdoctoral researcher to work on machine learning research for materials discovery.

All details here: assets-v2.circle.so/4htxv8ljrkzr...

A critical examination of robustness and generalizability of machine learning prediction of materials properties - npj Computational Materials npj Computational Materials - A critical examination of robustness and generalizability of machine learning prediction of materials properties

I would like to formalize that with a more in-depth study, but I think it aligns with some of the litterature on generalizability:
doi.org/10.1038/s415...

We found that ML proxies perform significantly worse on generated molecules than on dataset molecules. A large SOTA model who did great on benchmarks was more unreliable than a small GNN on these molecules, regardless of how they were generated.

Personally, one of the main takeaways I got from writing this paper is that *evaluating the performance of generative models is really hard because their performance is super dependent on the ML proxy you use for evaluation.*

The paper shows how you can use molecule property predictors as generators through a simple input optimization if you carefully restrict the input space. It performs surprisingly well and generates super diverse molecules.

Using GNN property predictors as molecule generators - Nature Communications Graph neural networks (GNNs) have become ubiquitous as molecular property predictors. Here, authors propose a method to use them in reverse to directly generate diverse functional molecules with desir...

My paper from my time at UofT just came out in Nature Communications! 🎉 A small thread about *my* takeways from writing it:

doi.org/10.1038/s414...

🚨 Call for a POSTDOC in machine learning for drug discovery at @umontreal.ca 💊

This is a large collaboration with @brunlabcaulo.bsky.social @audurand.bsky.social @dom-beaini.bsky.social Anne Marinier @mila-quebec.bsky.social IRIC

All details here: brunlab.com/jobs/open-po...

Please repost 💜

Internship opportunity: Spatiotemporal Graph applications for Smart Buildings⚡⚡⚡ working closely with me, starting in early 2025. Considering applying or/and sharing this with your network please. Apply here: forms.gle/N3kwFxM3yEhS... LinkedIn ad here: www.linkedin.com/feed/update/...

Big News in AI4Science! ✨
We are thrilled to launch LeMaterial, an open-source project in collaboration with @hf.co to accelerate materials discovery ⚛️🤗

Discover LeMat-Bulk: a 6.7M-entry dataset standardizing and unifying Materials Project, Alexandria and OQMD

Great opportunity to work on solving climate change in a super cool work environment!