Wonjun Jo

I am a Ph.D. candidate in the Department of Electrical Engineering at POSTECH. I conduct my research at the Advanced Machine Intelligence (AMI) Lab at KAIST, advised by Prof. Tae-Hyun Oh.

I am passionate about developing Embodied AI agents for real-world applications.
I believe that the efficient use of human labels and the integration of information from multiple modalities are essential for achieving this goal.
Consequently, my main areas of interest include Self-supervised Learning, Multi-modal Learning, World Modeling for Embodied AI.
I am also pursuing theoretical inquiries from an information-theoretic perspective that unify these studies, in order to explore the emergence of intelligence.

Recently, I have been working with egocentric video data for robotics to advance Embodied AI. I am also interested in extending to diverse modalities (e.g., sound, touch sensors).

Email   /  CV  /  Scholar  /  LinkedIn

Research

Improving robustness of self-supervised 2D representation learning.

Improving pre-trained 2D image encoder's lighting robustness and 3D awareness.

Proposing simple remedies for self-supervised 3D representation learning with 2D representation.

Template adapted from here.