Tsung-Wei Ke

Assistant Professor

NTU Department of Computer Science

I am an Assistant Professor at NTU CSIE, where I am leading the Embodied Artificial Intelligence Lab. Before joining NTU, I spent two years as a postdoc at CMU MLD, working with Katerina Fragkiadaki. During 2018-2022, I have closely worked with Stella Yu for my Ph.D at UC Berkeley VS Program.

twke (at) csie (dot) ntu (dot) edu (dot) tw

[GitHub] [Google Scholar] [Download CV]

Research

I am interested in the trio of action, perception, and cognition. More particularly, I have been hoping to build machines that perceives to recognize the real world, understands the (physical) outcomes of their actions, explores novel complex behaviors, and learns/plans to achieve goals. I am fascinated by the animal beings that generalize and explore so successfully (see Moravec's Paradox).

Prospective students

I am looking for Ph.D/master/undergraduate students interested in robot learning, computer vision and machine learning. If you are a undergraduate at NTU, interested in research positions, please fill in this form. If you are interested in Mater/Doctoral research positions, please fill in this form.

Publications

3D Diffuser Actor: Policy Diffusion with 3D Scene Representations

CoRL 2024

Tsung-Wei Ke*, Nikolaos Gkanatsios*, and Katerina Fragkiadaki

[Arxiv] [Code] [Project]

Diffusion-ES: Gradient-free Planning with Diffusion for Autonomous and Instruction-guided Driving

CVPR 2024

Brian Yang, Huangyuan Su, Nikolaos Gkanatsios, Tsung-Wei Ke, Ayush Jain, Jeff Schneider, and Katerina Fragkiadaki

[Arxiv] [Code] [Project]

Learning Hierarchical Image Segmentation For Recognition and By Recognition<

ICLR 2024 (Spotlight)

Tsung-Wei Ke*, Sangwoo Mo*, and Stella Yu

[Arxiv] [Code] [Project]

Test time Adaptation with Diffusion Models

NeuRIPS 2023

Mihir Prabhudesai*, Tsung-Wei Ke*, Alexander Cong Li, Deepak Pathak and Katerina Fragkiadaki

[Arxiv] [Code] [Project]

Unsupervised Hierarchical Semantic Segmentation with Multiview Cosegmentation and Clustering Transformers

CVPR 2022 (Oral)

Tsung-Wei Ke, Jyh-Jing Hwang, Yunhui Guo, Xudong Wang and Stella X. Yu

[Arxiv] [Code] [Project]

Universal Weakly Supervised Segmentation by Pixel-to-Segment Contrastive Learning

ICLR 2021

Tsung-Wei Ke, Jyh-Jing Hwang, and Stella X. Yu

[Arxiv] [Code] [Project]

Adversarial Structure Matching for Structured Prediction Tasks

CVPR 2019

Jyh-Jing Hwang, Tsung-Wei Ke, Jianbo Shi and Stella X. Yu

[Arxiv] [Code] [Project]

Adaptive Affinity Field for Semantic Segmentation

ECCV 2018

Tsung-Wei Ke*, Jyh-Jing Hwang*, Ziwei Liu and Stella X. Yu (*equal contribution)

[Arxiv] [Code] [Project]

Neural MultiGrid

CVPR 2017

Tsung-Wei Ke, Michael Maire, and Stella X. Yu

[Arxiv] [Code]