Taesup (TS) Kim
PhD Candidate

My name is Taesup Kim, also simply called TS. I'm currently a PhD candidate at Université de Montréal, and supervised by professor Yoshua Bengio. Previously, I was a computer vision researcher at Intel Korea and LG electronics. I also studied machine learning during my master degree at KAIST in Korea. My current research interests focus on deep generative model and meta-learning (learn-to-learn). 


  • Sep. 2019: Our 'Variational Temporal  Abstraction' and 'Fast AutoAugment' are accepted in NeurIPS 2019.
  • Jun. 2019: Our 'Scalable Neural Architecture Search for 3D Medical Image Segmentation' is accepted in MICCAI 2019
  • Jun. 2019: Our 'Variational Temporal  Abstraction' is accepted in ICML Workshop 2019 as SPOTLIGHT
  • Jun. 2019: Our 'Fast AutoAugment' is accepted in ICML Workshop 2019 as POSTER
  • Feb. 2019: Our 'Edge-Labeling Graph Neural Network for Few-shot Learning' is accepted in CVPR 2019 as ORAL
  • Sep. 2018: Our 'Bayesian MAML' is accepted in NIPS 2018 as SPOTLIGHT
  • Feb. 2018: Two papers accepted in ICLR 2018 and ICASSP 2018
  • Aug. 2017: ORAL, "Dynamic Layer Normalization for Adaptive Neural Acoustic Modeling in Speech Recognition" (INTERSPEECH 2017)
  • Jun. 2017: Starting research internship at Microsoft Research (Cambridge, UK)
  • Sep. 2017: Joining MILA team for 'The Alexa Prize'
  • Jun. 2016: Starting research internship at Microsoft Research (Redmond, WA)
  • Sep. 2015: Starting PhD at Université de Montréal

  • Sep.2015~Present, Ph.D. candidate, MILA, DIRO, Université de Montréal, Montreal, QC, Canada
  • Sep.2009~Aug.2011, M.S., Electrical Engineering, KAIST (Korea Advanced Institute of Science and Technology), Daejeon, Korea
  • Mar.2004~Aug.2009, B.S., Electrical Engineering, Korea University, Seoul, Korea

  • Oct.2017~ Aug.2018, research intern, ElementAI, Montreal, Canada
  • Jun.2017~Sep.2017, research intern, Machine Intelligence and Perception Group, Microsoft Research, Cambridge, UK
  • Jun.2016~Sep.2016, research intern, Cognition Group, Microsoft Research, Redmond, WA, USA
  • Apr.2015~Aug.2015, machine learning scientist, Solidware (startup company), Seoul, Korea
  • Feb.2014~Apr.2015, computer vision researcher, Imaging & Camera Technology Group, Intel, Seoul, Korea
  • Oct.2011~Jan.2014, researcher, Advanced Technology & System Group, LG Electronics, Seoul, Korea
  • Jul.2010~Aug.2010, visiting researcher, Machine Learning & Perception Group, Microsoft Research Cambridge, Cambridge, UK
  • Jun. 2019, 'Variational Temporal Abstraction',
    • T. Kim, S. Ahn, Y. Bengio, (Spotlight), ICML Generative Modeling and Model-Based Reasoning for Robotics and AI Workshop 2019, Longbeach, CA, USA
  • Jun. 2019, 'Fast AutoAugment',
    • S. Lim*, I. Kim*, T.Kim, C.Kim, S.Kim, ICML AutoML Workshop 2019, Longbeach, CA, USA
  • Jun. 2019, 'Edge-Labeling Graph Neural Network for Few-shot Learning',
    • J. Kim, T. Kim, S. Kim, C. Yoo, (Oral) CVPR 2019, Longbeach, CA, USA
  • Jul. 2018, 'Noise-Adaptive Deep Neural Network for Single-Channel Speech Enhancement',
    • H. Chung, T. Kim, E. Plourde, B. Champagne, MLSP 2018, Aalborg, Denmark
  • Jun. 2018, 'Bayesian Model Agnostic Meta Learning',
    • T. Kim*, J. Yoon*, O. Dia , S. Kim, Y. Bengio, S. Ahn, (Spotlight) NIPS 2018, Montreal, Canada
  • Oct. 2017, 'Dynamic Frame Skipping for Fast Speech Recognition in Recurrent',
    • I. Song, J. Chung, T. Kim ,Y. Bengio, ICASSP 2018, Calgary, Canada
  • Oct. 2017, 'PixelDefend: Leveraging Generative Models to Understand and Defend against Adversarial Examples',
    • Y. Song, T. Kim, S. Nowozin, S. Ermon, N. Kushman, ICLR 2018, Vancouver, Canada
  • Sep. 2017, 'A Deep Reinforcement Learning Chatbot',
    • I. Serban, C. Sankar, M. Germain, S. Zhang, Z. Lin, S. Subramanian, T. Kim, M. Pieper, S. Chandar, N. Ke, S. Mudumba, A. Brebisson, J. Sotelo, D. Suhubdy, V. Michalski, A. Nguyen, J. Pineau, Y. Bengio, Alexa Prize (arXiv:1709.02349)
  • Aug. 2017, 'Dynamic Layer Normalization for Adaptive Neural Acoustic Modeling in Speech Recognition', 
    • T. Kim, I. Song, Y. Bengio, (Oral), INTERSPEECH 2017, Stockholm, Sweden
  • May. 2016, 'Deep Directed Generative Models with Energy-Based Probability Estimation'
    • T. Kim, Y. Bengio, ICLR Workshop 2016, San Juan, Puerto Rico
  • Jun. 2011, 'Variable Grouping for Energy Minimization'
    • T. Kim, S. Nowozin, P. Kohli, C. Yoo, CVPR 2011, Colorado, USA

  • Email : taesup dot kim at umontreal dot ca