Jisoo Lee
Hi, I am Jisoo Lee, currently working as a research intern of voice synthesis team at Krafton Inc. Before joining the team, I worked at the KAIST interaction lab (KIXLAB) as a grad researcher to learn to build ML algorithms that are oriented towards user needs.


I received my bachelor’s degree in Electrical Engineering (with a CS minor) from KAIST in 2017. I completed my master’s degree in Electrical Engineering at KAIST in 2019 under the supervision of Prof. Sae-Young Chung.
Research Interest

I am particularly interested in Fairness/Accountability/Transparency (FAT) and Reinforcement Learning. I believe that research on that area will facilitate more interaction between ML models and the users, and thus efficiently increase user satisfaction towards ML-backed tools.

Robust training with ensemble consensus (ICLR 2020)

Jisoo Lee and Sae-Young Chung

The motivational insight is “neural networks are hard to generalize low correlated instances”. We regard low correlated instances as noisy and removed them for robust training. We develop a method to identify such low correlated ones via ensemble consensus.

[Paper] [Slides] [Code] [OpenReview]

I won the EE department outstanding TA award (EE807 Deep Reinforcement Learning and AlphaGo).

2017.3 ~ 2017.6 : Undergraduate Teaching Assistant, CS101 Introduction to Programming, KAIST
2018.3 ~ 2018.8 : Mentor, EE496 Individual Study, KAIST
2018.9 ~ 2018.12 : Teaching Assistant, EE807 Deep Reinforcement Learning and AlphaGo, KAIST
2019.3 ~ 2019.6 : Teaching Assistant, EE485 Philosophical issues in AI, KAIST