About me

I’m an individual machine learning researcher. Prior to this, I was a postdoctoral researcher at Mila and Université de Montréal under the supervision of Prof. Yoshua Bengio. I earned Ph.D degree from KAIST under the supervision of Prof. Sung Ju Hwang.

Research Interest

My research interest include

• System 2 Deep Learning
• Generative Flow Networks (GFlowNet)
• Bayesian Inference and Learning
• Meta-Learning / Multi-Task Learning / Transfer Learning
• AutoML

Contact

haebeom dot lee at kaist dot ac dot kr

Awards

• Global Ph.D Fellowship Program, 2019-2021
• Google Ph.D Fellowship Program 2021
• Outstanding reviewer (ICML2020 - Top 33%, ICML2022 - Top 10%)

New Preprints

• Dataset Condensation with Latent Space Knowledge Factorization and Sharing

  [paper]
  Hae Beom Lee*, Dong Bok Lee*, Sung Ju Hwang
  (*: equal contribution)
  arXiv, 2022

• Cost-Sensitive Multi-Fidelity Bayesian Optimization with Transfer of Learning Curve Extrapolation

  [paper]
  Dong Bok Lee, Aoxuan Silvia Zhang*, Byungjoo Kim*, Junhyeon Park*, Juho Lee, Sung Ju Hwang, Hae Beom Lee
  (*: equal contribution)
  arXiv, 2024

Conference Publications

• Delta-AI: Local Objectives for Amortized Inference in Sparse Graphical Models

  [paper]
  Jean-Pierre René Falet*, Hae Beom Lee*, Nikolay Malkin*, Chen Sun, Dragos Secrieru, Dinghuai Zhang, Guillaume Lajoie, Yoshua Bengio
  (*: equal contribution)
  ICLR 2024

• Online Hyperparameter Meta-Learning with Hypergradient Distillation

  [paper]
  Hae Beom Lee, Hayeon Lee, Jaewoong Shin, Eunho Yang, Timothy M. Hospedales, Sung Ju Hwang
  ICLR 2022 (spotlight)

• Sequential Reptile: Inter-Task Gradient Alignment for Multilingual Learning

  [paper]
  Seanie Lee*, Hae Beom Lee*, Juho Lee, Sung Ju Hwang
  (*: equal contribution)
  ICLR 2022

• Meta-Learning Low Rank Covariance Factors for Energy-Based Deterministic Uncertainty

  [paper]
  Jeffrey Ryan Willette, Hae Beom Lee, Juho Lee, Sung Ju Hwang
  ICLR 2022

• Large-Scale Meta-Learning with Continual Trajectory Shifting

  [paper] [code]
  Jaewoong Shin*, Hae Beom Lee*, Boqing Gong, Sung Ju Hwang
  (*: equal contribution)
  ICML 2021

• MetaPerturb: Transferable Regularizer for Heterogeneous Tasks and Architectures

  [paper] [code]
  Jeongun Ryu*, Jaewoong Shin*, Hae Beom Lee*, Sung Ju Hwang
  (*: equal contribution)
  NeurIPS 2020 (spotlight)

• Meta-Learning for Short Utterance Speaker Recognition with Imbalance Length Pairs

  [paper] [code]
  Seong Min Kye, Youngmoon Jung, Hae Beom Lee, Sung Ju Hwang, and Hoirin Kim
  Interspeech 2020

• Meta Variance Transfer: Learning to Augment from the Others

  [paper]
  Seong Jin Park, Seungju Han, Ji-won Baek, Insoo Kim, Juhwan Song, Hae Beom Lee, Jae-Joon Han and Sung Ju Hwang
  ICML 2020

• Learning to Balance: Bayesian Meta-Learning for Imbalanced and Out-of-distribution Tasks

  [paper] [code]
  Hae Beom Lee*, Hayeon Lee*, Donghyun Na*, Saehoon Kim, Minseop Park, Eunho Yang, Sung Ju Hwang
  (*: equal contribution)
  ICLR 2020 (oral presentation)

• Meta Dropout: Learning to Perturb Latent Features for Generalization

  [paper] [code]
  Hae Beom Lee, Taewook Nam, Eunho Yang, Sung Ju Hwang
  ICLR 2020

• DropMax: Adaptive Variational Softmax

  [paper][code]
  Hae Beom Lee, Juho Lee, Saehoon Kim, Eunho Yang, Sung Ju Hwang
  NeurIPS 2018

• Uncertainty-Aware Attention for Reliable Interpretation and Prediction

  [paper][code]
  Jay Heo*, Hae Beom Lee*, Saehoon Kim, Juho Lee, Kwang Joon Kim, Eunho Yang, Sung Ju Hwang
  (*: equal contribution)
  NeurIPS 2018

• Deep Asymmetric Multi-task Feature Learning

  [paper][code]
  Hae Beom Lee, Eunho Yang, Sung Ju Hwang
  ICML 2018

Old Preprints

• Meta-Learned Confidence for Few-shot Learning

  [paper][code]
  Sung Min Kye, Hae Beom Lee, Hoirin Kim, Sung Ju Hwang
  arXiv, 2020

• Adaptive Network Sparsification with Dependent Variational Beta-Bernoulli Dropout

  [paper][code]
  Juho Lee, Saehoon Kim, Jaehong Yoon, Hae Beom Lee, Eunho Yang, Sung Ju Hwang
  arXiv, 2018