BangLiu
@BangL93
I am an associate professor at University of Montreal and Mila.
Glad to share our work on AI for Earthquake😁
🚀Check out the new paper! 🚀Authors from University of Montreal, Woods Hole Oceanographic Institution, and UC Berkeley proposed DASFormer, a novel self-supervised pretraining technique on DAS data for next-generation earthquake monitoring. link.springer.com/article/10.100… @SpringerEng
One of the biggest challenges in foundation agents is memory and reasoning. How can we improve them to truly enable general purpose agents? @buZZrobot guest @BangL93 Bang Liu from @UMontreal shares his recent research work and proposed path forward to solving those fundamental…
I’m grateful to share that I officially start as a tenured Associate Professor at Université de Montréal and Mila today. Merci à tous les mentors, collègues et étudiant·e·s qui m’ont accompagné sur ce chemin. Hâte de poursuivre cette belle aventure !
You may be interested in the following summer school / Bootcamp for AI & Science if your research is related to GenAI and DL for Science: 1. Duke May 12-16: ai-bootcamp.cs.duke.edu 2. Berkeley June 23-27: dl4sci-school.lbl.gov
Big thanks to Dr @BangL93 from @UMontreal for the presentation on « Advances and challenges in foundation agents ». He introduced the direction to draw insights from human brain functions and social behaviours . What does the AI future holds for us ? #SANER2025 #KeynoteSpeaker
🎉 AFLOW accepted for #ICLR2025 Oral! 🔧Easy to use for closed & open tasks! 📉Low inference costs with DeepSeek vs. larger models! ✨Promising research: Automatic Agentic Workflow/Systems! Paper: openreview.net/pdf?id=z5uVAKw… Code: github.com/geekan/MetaGPT…
We're thrilled to introduce FACT. It can improve the accuracy of retrieval problems such as NIAH from ~70% to 99.4%. It strives to find all the useful context and make the most of it. 💡 Performance improved by more than 20% on some QA datasets, and interesting findings were…
🤖 Can an AI agent design another AI? 🌟 Yes. SELA uses MCTS to design AI, achieving SoTA performance on 20 machine learning datasets. It can learn from past designs and experiments to create better AIs. It's fully open-source. 🚀 Try SELA yourself 📄 Paper:…