|
Reasoning by Consequence: Selective Chain-of-Thought for Driving VLAs
Jacob Yeung,
Wenyuan Zhang,
Nadine Chang,
Michael J. Tarr,
Deva Ramanan,
Jose M. Alvarez,
Pichao Wang,
Ke Chen
Website
We show that chain of thought (CoT) improves driving when the car must act and degrades it when the car should keep going. We propose Clutch, a router that selects between trajectories generated with and without CoT, and counterfactual supervision, which trains the trajectory decoder to follow the CoT.
|
|
Reanimating Images using Neural Representations of Dynamic Stimuli
Jacob Yeung,
Andrew F. Luo,
Gabriel Sarch,
Margaret M. Henderson,
Deva Ramanan,
Michael J. Tarr
CVPR 2025 Oral, (top 0.7% of all submissions)
Paper / Website / bibtex
We propose a way to study visual motion in the brain by decoupling the modeling of static image representations and motion representations in the human brain.
|
|
Brain Mapping with Dense Features: Grounding Cortical Semantic Selectivity in Natural Images With Vision Transformers
Andrew F. Luo,
Jacob Yeung,
Rushikesh Zawar, Shaurya Dewan,
Margaret M. Henderson,
Leila Wehbe,
Michael J. Tarr
ICLR 2025
Paper / bibtex
We propose an efficient gradient-free distillation module capable of extraction high quality dense CLIP embeddings, and utilize these embeddings to understand semantic selectivity in the visual cortex.
|
|
Discovery of Linked Neural and Behavioral Subspaces with External Dynamic Components Analysis
Jacob Yeung,
Ji Hyun Bak,
Kristofer Bouchard
COSYNE, 2023
Paper /
Poster
We introduce external Dynamics Components Analysis
(eDCA), a linear dimensionality reduction method that finds subspaces of past neural data that have maximal mutual
information with subspaces of future behavior data.
|
|
Hangul Fonts Dataset: a Hierarchical and Compositional Dataset for Investigating Learned Representations
Jesse Livezey,
Ahyeon Hwang,
Jacob Yeung,
Kristofer Bouchard
International Conference on Image Analysis and Processing, 2022
Paper / bibtex
We introduce a dataset with known hierarchical and compositional structure and show deep unsupervised models poorly learn the latent generative structures.
|
|
Teaching
|
|
I have been a teaching assistant for the following courses:
|
|
I have been a tutor for the following courses:
|
|