Meta Variance Transfer: Learning to Augment from The Others

ICML 2020 Meta Variance Transfer: Learning to Augment from The Others SeongJin Park, Seungju Han, Jiwon Baek, Insoo Kim, Juhwan Song, Hae Beom Lee, Jae-Joon Han, Sung Ju Hwang  Humans have the ability to robustly recognize objects with various factors of variations such as n...

Federated Semi-Supervised Learning with Inter-Client Consistency

ICML 2020 (Workshop in Federated Learning) Federated Semi-Supervised Learning with Inter-Client Consistency & Disjoint Learning Wonyong Jeong, Jaehong Yoon, Eunho Yang, Sung Ju Hwang  While existing federated learning approaches mostly require that clients have fully-labeled ...

Federated Continual Learning with Weighted Inter-client Transfer

ICML 2021 Federated Continual Learning with Weighted Inter-client Transfer Jaehong Yoon, Wonyong Jeong, Giwoong Lee, Eunho Yang and Sung Ju Hwang  There has been a surge of interest in continual learning and federated learning, both of which are important in deep neural networks ...

Cost-effective Interactive Attention Learning with Neural Attention Processes

ICML 2020 Cost-effective Interactive Attention with Neural Attention Processes Jay Heo, Junhyeon Park, Hyewon Jeong, Kwang Joon Kim, Juho Lee, Eunho Yang, Sung Ju Hwang  We propose a novel interactive learning framework which we refer to as Interactive Attention Learning (IA...

Adversarial Neural Pruning with Latent Vulnerability Suppression

ICML 2020 Adversarial Neural Pruning with Latent Vulnerability Suppression Divyam Madaan, Jinwoo Shin, Sung Ju Hwang  Despite the remarkable performance of deep neural networks on various computer vision tasks, they are known to be susceptible to adversarial perturbations, w...

A benchmark study on reliable molecular supervised learning via Bayesian learning

ICML 2020 Workshop on Uncertainty and Robustness in Deep Learning A benchmark study on reliable molecularsupervised learning via Bayesian learning Doyeong Hwang, Grace Lee, Hanseok Jo, Seyoul Yoon, Seongok Ryu  Virtual screening aims to find desirable compounds from chemical libr...

Trimming the ℓ 1 Regularizer: Statistical Analysis, Optimization, and Applications to Deep Learning

ICML 2019 (full oral presentation) Trimming the ℓ 1 Regularizer: Statistical Analysis, Optimization, and Applications to Deep Learning Jihun Yun, Peng Zheng, Aurelie Lozano, Aleksandr Aravkin, Eunho Yang  We study high-dimensional estimators with the trimmed ℓ1 penalty,...

Training CNNs with Selective Allocation of Channels

ICML 2019 Training CNNs with Selective Allocation ofChannels Jongheon Jeong, Jinwoo Shin  Recent progress in deep convolutional neural networks (CNNs) have enabled a simple paradigm of architecture design: larger models typically achieve better accuracy. Due to this, in modern CN...

Stochastic Gradient Methods with Block Diagonal Matrix Adaptation

ICML 2019 Stochastic Gradient Methods with Block Diagonal Matrix Adaptation Jihun Yun, Aurelie C. Lozano, Eunho Yang  Adaptive gradient approaches that automatically adjust the learning rate on a per-feature basis have been very popular for training deep networks. This rich class...

Spectral Approximate Inference

ICML 2019 Spectral Approximate Inference Sejun Park, Eunho Yang, Se-Young Yun, Jinwoo Shin  Given a graphical model (GM), computing its partition function is the most essential inference task, but it is computationally intractable in general. To address the issue, iterative ...