Alex Oesterling
@alex_oesterling
PhD Candidate at Harvard. Fairness, Interpretability, and Information Theory. He/Him/His
Finally, I am pleased to announce 🪢Interpreting CLIP with Sparse Linear Concept Embeddings (SpLiCE)🪢 Joint work with Usha Bhalla, as well as @Suuraj, @FlavioCalmon, and @hima_lakkaraju, which was just accepted to NeurIPS 2024! Check out the paper here: arxiv.org/abs/2402.10376
1/10 ML can solve PDEs – but precision🔬is still a challenge. Towards high-precision methods for scientific problems, we introduce BWLer 🎳, a new architecture for physics-informed learning achieving (near-)machine-precision (up to 10⁻¹² RMSE) on benchmark PDEs. 🧵How it works:
🤝Takeaway: Don't just optimize. Hedge! Work done with my great collaborators: @claudiomverdun @alex_oesterling @hima_lakkaraju @FlavioCalmon Paper: arxiv.org/abs/2506.19248 Website: hadikhalaf.com/blog/hacking
(1/9) Excited to share my recent work on "Alignment reduces LM's conceptual diversity" with @TomerUllman and @jennhu, to appear at #NAACL2025! 🐟 We want models that match our values...but could this hurt their diversity of thought? Preprint: arxiv.org/abs/2411.04427
Very happy that our work won a best paper award at the Generative AI for Health workshop at #NeurIPS2024 ! If you want to know what are the LLM applications that are both useful for healthcare professionals and feasible in practice, check out the paper: openreview.net/pdf?id=JixmrZF…
It's time for #NeurIPS2024 🇨🇦! I won't be attending this time, but check out our work below! Poster: "Interpreting CLIP via Sparse Linear Concept Embeddings (SPLiCE)" 📄 Paper: arxiv.org/abs/2402.10376 📅 Date: Friday, 13th Dec, 16:30 PST 1/n
Join us at the #RegulatableML workshop at #NeurIPS2024 to learn about AI regulations and how to operationalize them in practice. 🗓️ Date: Dec 15, 2024 (East Meeting Room 13) 🕓 Time: 8:15 am - 5:30 pm 🔗 Details: regulatableml.github.io We have an exciting schedule: ⭐️ Six…
I’ll be at NeurIPS next week if anyone wants to hang out!
📢Our latest preprint shows that learning global neuron shapes can help to automatically proofread connectomes and predict neuron types. biorxiv.org/content/10.110… Work done in collaboration with @HHMIJanelia, @srinituraga & @HarvardVCG #connectomics #AI 🧵(1/n)