🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

How to do continual learning on a data stream without restrictions on network architecture?🤔 Our proposed framework learns a stream of learners with diverse architectures. Visit our poster to explore this work done during my internship at NVIDIA 📷 youtube.com/watch?v=8Il-bp…

CDS is excited to announce this summer's NYU AI School! A week of hands-on labs & expert lectures, all for FREE. Open to undergrads in the NYC area, no prior experience needed. Join us in person at NYU, June 3rd - 7th, 2024! More info: nyu-mll.github.io/nyu-ai-school-…

How do you know if a method is better, or just has better hyperparameters? He He, Kyunghyun Cho, and I give a new tool to answer this in our #NAACL2024 paper: "Show Your Work with Confidence" arxiv.org/abs/2311.09480. Use it in your own work with just a "pip install opda"! 🧵 1/8

this semester (spring 2024), i created and taught a new introductory course on causal inference in machine learning, aimed at msc and phd students in cs and ds. the whole material was created from scratch, including the lecture note and lab materials; docs.google.com/document/d/1qN… 1/4

CDS has wrapped its 4th annual NYU AI School, introducing AI to undergrads from schools across NYC. With a goal of making AI education accessible to all, 61% of attendees were from non-CS majors, and the program covered ML, vision, & NLP. Till next year! nyudatascience.medium.com/nyu-ai-school-…

2⃣ "A Framework for Multi-modal Learning: Jointly Modeling Inter- & Intra-Modality Dependencies" Divyam Madaan, Taro Makino, Sumit Chopra, Kyunghyun Cho arxiv.org/abs/2405.17613

CDS & NYU Courant researchers Divyam Madaan, Taro Makino, @suchop, & Kyunghyun Cho introduced I2M2, a new framework that improves multi-modal AI performance across healthcare and vision-language tasks. nyudatascience.medium.com/new-framework-…

Join us tomorrow at 10 AM PST for a deep dive into multi-modal learning at our next DLCT event! 🚀 Divyam Madaan will explore "Jointly Modeling Inter- & Intra-Modality Dependencies for Multi-modal Learning" — uncovering new insights into dependencies across modalities. Don’t miss it!

🚨I'm on the faculty job market this year!🚨 My research focuses on AI safety & generalization. I'm interested in developing technical solutions for ensuring reliability and alignment of advanced AI models, particularly LLM agents that are becoming increasingly capable. 🧵1/4

Why do multimodal models sometimes outperform unimodal learners and their combinations, yet other times fall short? 🤔 Join us on Wednesday at 11am for our #NeurIPS poster 3808 in the East Exhibit Hall to find out the answer in my work w/ Taro Makino, @suchop, and Kyunghyun Cho

Join us tomorrow as we present the progress in multi-modal learning, covering benchmarks to model architectures, and share a recipe for meaningful progress in the future. w/ @suchop and Kyunghyun Cho 🔗 iclr-blogposts.github.io/2025/blog/mult… ⏰ 3PM 📍 Hall 3 + Hall 2B, Poster 550 #ICLR25

CDS PhD student Haoxu Huang’s new paper shows AI diagnostic models are more accurate when considering a patient’s past medical reports—not just current scans. With Courant PhD Divyam Madaan, CDS Prof. Kyunghyun Cho, Cem M. Deniz, and CDS-affiliated Sumit Chopra. nyudatascience.medium.com/ai-diagnostic-…

✅ Divyam Madaan will distinguish between different modeling paradigms for multi-modal learning from the perspective of generative models and offer a general recipe for designing models that efficiently leverage multi-modal data. x.com/dmadaan_/statu…