Yuki Wang
@YukiWang_hw
Third Year CS Ph.D. Student at Cornell University @CornellCIS A member of the PoRTaL group @PortalCornell!
Come see the pitfalls of aligning video frames, and my first ever paper at #ICML2025. Huge thanks to Yuki for helping me with all the nuances of academic research, she was an outstanding mentor and co-lead on this.
Using OT to define rewards for imitating video demos is popular, but it breaks down when demos are temporally misaligned—a frequent challenge in practice. We present ORCA at #ICML2025 , which defines rewards by aligning sequences, rather than matching individual frames via OT.
Curious about a simple and scalable approach to multi-turn code generation? Come check out μCode — our framework built on one-step recoverability and multi-turn BoN search. Stop by and say hi during Poster Session 4 at #ICML2025 today at East Hall A-B # E-2600.
Coding agents can debug their own outputs, but what if none of the fixes are correct? We overcome sparse rewards by making them continuous📈 Instead of having binary execution rewards, we introduce a learned verifier to measure how close the current solution is to a correct one📏
I will be presenting this work at ICML 2025: ⏰ Thu 17 Jul 11am-1:30pm PDT 📍West Exhibition Hall B2-B3 W-316 DM if you want to chat about multimodal models / interactions / anything!
🗣️We can listen and speak simultaneously when we talk, and so should the spoken dialogue models (SDMs)! 💬Unlike typical "walkie-talkie" voice AIs, full-duplex SDMs let both sides talk at once - more like real, natural conversation. But this makes alignment harder: - No…
How to learn from temporally misaligned demo videos? Come and chat on Wed 16 11-1.30pm @ W. Exhibition Hall B2-B3 W-709! #ICML2025
Using OT to define rewards for imitating video demos is popular, but it breaks down when demos are temporally misaligned—a frequent challenge in practice. We present ORCA at #ICML2025 , which defines rewards by aligning sequences, rather than matching individual frames via OT.
Come check out our work on reward learning with sequence data at ICML! Will be there :)
Using OT to define rewards for imitating video demos is popular, but it breaks down when demos are temporally misaligned—a frequent challenge in practice. We present ORCA at #ICML2025 , which defines rewards by aligning sequences, rather than matching individual frames via OT.
Most assistive robots live in labs. We want to change that. FEAST enables care recipients to personalize mealtime assistance in-the-wild, with minimal researcher intervention across diverse in-home scenarios. 🏆 Outstanding Paper & Systems Paper Finalist @RoboticsSciSys 🧵1/8
Really excited to share that FEAST won the Best Paper Award at #RSS2025! Huge thanks to everyone who’s shaped this work, from roboticists to care recipients, caregivers, and occupational therapists. ❤️
Most assistive robots live in labs. We want to change that. FEAST enables care recipients to personalize mealtime assistance in-the-wild, with minimal researcher intervention across diverse in-home scenarios. 🏆 Outstanding Paper & Systems Paper Finalist @RoboticsSciSys 🧵1/8
Huge thanks to my co-lead @prithwish_dan, collaborators Angela Chao, Edward Duan & Maximus Pace, and co-advisors @weichiuma @sanjibac! Thrilled that our X-Sim paper received Best Paper (Runner-Up) at the EgoAct Workshop @RoboticsSciSys — winning a cool pair of @Meta Ray-Bans! 😎
Teleoperation is slow, expensive, and difficult to scale. So how can we train our robots instead? Introducing X-Sim: a real-to-sim-to-real framework that trains image-based policies 1) learned entirely in simulation 2) using rewards from human videos. portal-cornell.github.io/X-Sim
Say ahoy to 𝚂𝙰𝙸𝙻𝙾𝚁⛵: a new paradigm of *learning to search* from demonstrations, enabling test-time reasoning about how to recover from mistakes w/o any additional human feedback! 𝚂𝙰𝙸𝙻𝙾𝚁 ⛵ out-performs Diffusion Policies trained via behavioral cloning on 5-10x data!
I also want to call out a few great related papers that inspired this project: 1. Text2Interaction (Thumm et al., CoRL 2024). 2. APRICOT (Wang et al., CoRL 2024). 3. Trust the PRoC3S (Curtis*, Kumar*, et al., CoRL 2024). And I hope more work in this direction will continue.