no code implementations • NeurIPS 2011 • Kristen Grauman, Fei Sha, Sung Ju Hwang
Given a hierarchical taxonomy that captures semantic similarity between the objects, we learn a corresponding tree of metrics (ToM).
no code implementations • NeurIPS 2012 • Sung Ju Hwang, Kristen Grauman, Fei Sha
When learning features for complex visual recognition problems, labeled image exemplars alone can be insufficient.
no code implementations • NeurIPS 2014 • Sung Ju Hwang, Leonid Sigal
We propose a method that learns a discriminative yet semantic space for object categorization, where we also embed auxiliary semantic entities such as supercategories and attributes.
no code implementations • 6 Feb 2015 • Guang-Tong Zhou, Sung Ju Hwang, Mark Schmidt, Leonid Sigal, Greg Mori
We present a hierarchical maximum-margin clustering method for unsupervised data analysis.
no code implementations • CVPR 2015 • Alina Kuznetsova, Sung Ju Hwang, Bodo Rosenhahn, Leonid Sigal
By incrementally detecting object instances in video and adding confident detections into the model, we are able to dynamically adjust the complexity of the detector over time by instantiating new prototypes to span all domains the model has seen.
1 code implementation • ICML 2017 • Jaehong Yoon, Sung Ju Hwang
The number of parameters in a deep neural network is usually very large, which helps with its learning capacity but also hinders its scalability and practicality due to memory/time inefficiency and overfitting.
1 code implementation • ICML 2018 • Hae Beom Lee, Eunho Yang, Sung Ju Hwang
We propose Deep Asymmetric Multitask Feature Learning (Deep-AMTFL) which can learn deep representations shared across multiple tasks while effectively preventing negative transfer that may happen in the feature sharing process.
no code implementations • ICML 2017 • Juyong Kim, Yookoon Park, Gunhee Kim, Sung Ju Hwang
We propose a novel deep neural network that is both lightweight and effectively structured for model parallelization.
3 code implementations • ICLR 2018 • Jaehong Yoon, Eunho Yang, Jeongtae Lee, Sung Ju Hwang
We propose a novel deep network architecture for lifelong learning which we refer to as Dynamically Expandable Network (DEN), that can dynamically decide its network capacity as it trains on a sequence of tasks, to learn a compact overlapping knowledge sharing structure among tasks.
no code implementations • 18 Sep 2017 • Hajin Shim, Sung Ju Hwang, Eunho Yang
We consider the problem of active feature acquisition, where we sequentially select the subset of features in order to achieve the maximum prediction performance in the most cost-effective way.
4 code implementations • NeurIPS 2018 • Hae Beom Lee, Juho Lee, Saehoon Kim, Eunho Yang, Sung Ju Hwang
Moreover, the learning of dropout rates for non-target classes on each instance allows the classifier to focus more on classification against the most confusing classes.
no code implementations • ICLR 2018 • Hae Beom Lee, Juho Lee, Eunho Yang, Sung Ju Hwang
Moreover, the learning of dropout probabilities for non-target classes on each instance allows the classifier to focus more on classification against the most confusing classes.
2 code implementations • NeurIPS 2018 • Jay Heo, Hae Beom Lee, Saehoon Kim, Juho Lee, Kwang Joon Kim, Eunho Yang, Sung Ju Hwang
Attention mechanism is effective in both focusing the deep learning models on relevant features and interpreting them.
2 code implementations • ICLR 2019 • Yanbin Liu, Juho Lee, Minseop Park, Saehoon Kim, Eunho Yang, Sung Ju Hwang, Yi Yang
The goal of few-shot learning is to learn a classifier that generalizes well even when trained with a limited number of training instances per class.
1 code implementation • 28 May 2018 • Juho Lee, Saehoon Kim, Jaehong Yoon, Hae Beom Lee, Eunho Yang, Sung Ju Hwang
With such input-independent dropout, each neuron is evolved to be generic across inputs, which makes it difficult to sparsify networks without accuracy loss.
2 code implementations • 5 Jun 2018 • Ingyo Chung, Saehoon Kim, Juho Lee, Kwang Joon Kim, Sung Ju Hwang, Eunho Yang
We present a personalized and reliable prediction model for healthcare, which can provide individually tailored medical services such as diagnosis, disease treatment, and prevention.
no code implementations • CVPR 2019 • Sangil Jung, Changyong Son, Seohyung Lee, Jinwoo Son, Youngjun Kwak, Jae-Joon Han, Sung Ju Hwang, Changkyu Choi
We demonstrate the effectiveness of our trainable quantizer on ImageNet dataset with various network architectures such as ResNet-18, -34 and AlexNet, on which it outperforms existing methods to achieve the state-of-the-art accuracy.
no code implementations • 27 Sep 2018 • Juho Lee, Saehoon Kim, Jaehong Yoon, Hae Beom Lee, Eunho Yang, Sung Ju Hwang
With such input-independent dropout, each neuron is evolved to be generic across inputs, which makes it difficult to sparsify networks without accuracy loss.
no code implementations • ICLR 2019 • Hanze Dong, Yanwei Fu, Sung Ju Hwang, Leonid Sigal, xiangyang xue
This paper studies the problem of Generalized Zero-shot Learning (G-ZSL), whose goal is to classify instances belonging to both seen and unseen classes at the test time.
no code implementations • NeurIPS 2018 • Hajin Shim, Sung Ju Hwang, Eunho Yang
We consider the problem of active feature acquisition where the goal is to sequentially select the subset of features in order to achieve the maximum prediction performance in the most cost-effective way at test time.
1 code implementation • ICLR 2020 • Jaehong Yoon, Saehoon Kim, Eunho Yang, Sung Ju Hwang
First, a continual learning model should effectively handle catastrophic forgetting and be efficient to train even with a large number of tasks.
1 code implementation • ACL 2019 • Moonsu Han, Minki Kang, Hyunwoo Jung, Sung Ju Hwang
We consider a novel question answering (QA) task where the machine needs to read from large streaming data (long documents or videos) without knowing when the questions will be given, which is difficult to solve with existing QA methods due to their lack of scalability.
4 code implementations • 15 May 2019 • Yunhun Jang, Hankook Lee, Sung Ju Hwang, Jinwoo Shin
To address the issue, we propose a novel transfer learning approach based on meta-learning that can automatically learn what knowledge to transfer from the source network to where in the target network.
1 code implementation • ICLR 2020 • Hae Beom Lee, Hayeon Lee, Donghyun Na, Saehoon Kim, Minseop Park, Eunho Yang, Sung Ju Hwang
While tasks could come with varying the number of instances and classes in realistic settings, the existing meta-learning approaches for few-shot classification assume that the number of instances per task and class is fixed.
1 code implementation • 30 May 2019 • Hae Beom Lee, Taewook Nam, Eunho Yang, Sung Ju Hwang
Specifically, we meta-learn a noise generator which outputs a multiplicative noise distribution for latent features, to obtain low errors on the test instances in an input-dependent manner.
1 code implementation • ICLR 2020 • Joonyoung Yi, Juhyuk Lee, Kwang Joon Kim, Sung Ju Hwang, Eunho Yang
Among many approaches, the simplest and most intuitive way is zero imputation, which treats the value of a missing entry simply as zero.
no code implementations • 7 Jun 2019 • Sungyub Kim, Yongsu Baek, Sung Ju Hwang, Eunho Yang
We also introduce an additional form of a regularizer from the perspective of understanding ITE in the semi-supervised learning framework to ensure more reliable representations.
Ranked #1 on Causal Inference on IDHP
no code implementations • 5 Aug 2019 • Hayeon Lee, Donghyun Na, Hae Beom Lee, Sung Ju Hwang
To tackle this issue, we propose a simple yet effective meta-learning framework for metricbased approaches, which we refer to as learning to generalize (L2G), that explicitly constrains the learning on a sampled classification task to reduce the classification error on a randomly sampled unseen classification task with a bilevel optimization scheme.
1 code implementation • ICML 2020 • 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, which makes it challenging to deploy them in real-world safety-critical applications.
no code implementations • 10 Sep 2019 • Byunggill Joe, Sung Ju Hwang, Insik Shin
Yet, most of them cannot effectively defend against whitebox attacks where an adversary has a knowledge of the model and defense.
1 code implementation • ICML 2020 • Hankook Lee, Sung Ju Hwang, Jinwoo Shin
Our main idea is to learn a single unified task with respect to the joint distribution of the original and self-supervised labels, i. e., we augment original labels via self-supervision of input transformation.
no code implementations • 29 Nov 2019 • Jung Hyun Lee, Jihun Yun, Sung Ju Hwang, Eunho Yang
As a natural extension of DropBits, we further introduce the way of learning heterogeneous quantization levels to find proper bit-length for each layer using DropBits.
1 code implementation • 27 Feb 2020 • Seong Min Kye, Hae Beom Lee, Hoirin Kim, Sung Ju Hwang
To tackle this issue, we propose to meta-learn the confidence for each query sample, to assign optimal weights to unlabeled queries such that they improve the model's transductive inference performance on unseen tasks.
1 code implementation • 6 Mar 2020 • Jaehong Yoon, Wonyong Jeong, Giwoong Lee, Eunho Yang, 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 in real-world scenarios.
1 code implementation • 6 Apr 2020 • Seong Min Kye, Youngmoon Jung, Hae Beom Lee, Sung Ju Hwang, Hoirin Kim
By combining these two learning schemes, our model outperforms existing state-of-the-art speaker verification models learned with a standard supervised learning framework on short utterance (1-2 seconds) on the VoxCeleb datasets.
2 code implementations • ICLR 2020 • Hae Beom Lee, Taewook Nam, Eunho Yang, Sung Ju Hwang
Specifically, we meta-learn a noise generator which outputs a multiplicative noise distribution for latent features, to obtain low errors on the test instances in an input-dependent manner.
Ranked #1 on Meta-Learning on OMNIGLOT - 1-Shot, 20-way
1 code implementation • ACL 2020 • Dong Bok Lee, Seanie Lee, Woo Tae Jeong, Donghwan Kim, Sung Ju Hwang
We validate our Information Maximizing Hierarchical Conditional Variational AutoEncoder (Info-HCVAE) on several benchmark datasets by evaluating the performance of the QA model (BERT-base) using only the generated QA pairs (QA-based evaluation) or by using both the generated and human-labeled pairs (semi-supervised learning) for training, against state-of-the-art baseline models.
Ranked #1 on Question Generation on Natural Questions
2 code implementations • 9 Jun 2020 • Jay Heo, Junhyeon Park, Hyewon Jeong, Kwang Joon Kim, Juho Lee, Eunho Yang, Sung Ju Hwang
Moreover, it is almost infeasible for the human annotators to examine attentions on tons of instances and features.
1 code implementation • NeurIPS 2020 • Jinheon Baek, Dong Bok Lee, Sung Ju Hwang
For transductive link prediction, we further propose a stochastic embedding layer to model uncertainty in the link prediction between unseen entities.
2 code implementations • NeurIPS 2020 • Minseon Kim, Jihoon Tack, Sung Ju Hwang
In this paper, we propose a novel adversarial attack for unlabeled data, which makes the model confuse the instance-level identities of the perturbed data samples.
1 code implementation • NeurIPS 2020 • Jeongun Ryu, Jaewoong Shin, Hae Beom Lee, Sung Ju Hwang
As MetaPerturb is a set-function trained over diverse distributions across layers and tasks, it can generalize to heterogeneous tasks and architectures.
no code implementations • 22 Jun 2020 • Minyoung Song, Jaehong Yoon, Eunho Yang, Sung Ju Hwang
As deep neural networks are growing in size and being increasingly deployed to more resource-limited devices, there has been a recent surge of interest in network pruning methods, which aim to remove less important weights or activations of a given network.
1 code implementation • 22 Jun 2020 • Divyam Madaan, Jinwoo Shin, Sung Ju Hwang
Adversarial learning has emerged as one of the successful techniques to circumvent the susceptibility of existing methods against adversarial perturbations.
1 code implementation • ICLR 2021 • Wonyong Jeong, Jaehong Yoon, Eunho Yang, Sung Ju Hwang
Through extensive experimental validation of our method in the two different scenarios, we show that our method outperforms both local semi-supervised learning and baselines which naively combine federated learning with semi-supervised learning.
2 code implementations • 23 Jun 2020 • A. Tuan Nguyen, Hyewon Jeong, Eunho Yang, Sung Ju Hwang
Existing asymmetric multi-task learning methods tackle this negative transfer problem by performing knowledge transfer from tasks with low loss to tasks with high loss.
no code implementations • 25 Jun 2020 • Bruno Andreis, Seanie Lee, A. Tuan Nguyen, Juho Lee, Eunho Yang, Sung Ju Hwang
Deep models are designed to operate on huge volumes of high dimensional data such as images.
no code implementations • 28 Jun 2020 • Youngwan Lee, Joong-won Hwang, Hyung-Il Kim, Kimin Yun, Yongjin Kwon, Yuseok Bae, Sung Ju Hwang
To tackle these limitations, we propose a new localization uncertainty estimation method called UAD for anchor-free object detection.
Ranked #116 on Object Detection on COCO test-dev
no code implementations • ICLR 2021 • Youngmin Oh, Kimin Lee, Jinwoo Shin, Eunho Yang, Sung Ju Hwang
Experience replay, which enables the agents to remember and reuse experience from the past, has played a significant role in the success of off-policy reinforcement learning (RL).
1 code implementation • NeurIPS 2020 • Jaehyung Kim, Youngbum Hur, Sejun Park, Eunho Yang, Sung Ju Hwang, Jinwoo Shin
While semi-supervised learning (SSL) has proven to be a promising way for leveraging unlabeled data when labeled data is scarce, the existing SSL algorithms typically assume that training class distributions are balanced.
1 code implementation • NeurIPS 2020 • In Huh, Eunho Yang, Sung Ju Hwang, Jinwoo Shin
Time-reversal symmetry, which requires that the dynamics of a system should not change with the reversal of time axis, is a fundamental property that frequently holds in classical and quantum mechanics.
no code implementations • NeurIPS 2020 • Youngsung Kim, Jinwoo Shin, Eunho Yang, Sung Ju Hwang
While humans can solve a visual puzzle that requires logical reasoning by observing only few samples, it would require training over large amount of data for state-of-the-art deep reasoning models to obtain similar performance on the same task.
1 code implementation • NeurIPS 2020 • Juho Lee, Yoonho Lee, Jungtaek Kim, Eunho Yang, Sung Ju Hwang, Yee Whye Teh
While this "data-driven" way of learning stochastic processes has proven to handle various types of data, NPs still rely on an assumption that uncertainty in stochastic processes is modeled by a single latent variable, which potentially limits the flexibility.
1 code implementation • NeurIPS 2020 • Yoonho Lee, Juho Lee, Sung Ju Hwang, Eunho Yang, Seungjin Choi
While various complexity measures for deep neural networks exist, specifying an appropriate measure capable of predicting and explaining generalization in deep networks has proven challenging.
1 code implementation • EMNLP 2020 • Minki Kang, Moonsu Han, Sung Ju Hwang
We propose a method to automatically generate a domain- and task-adaptive maskings of the given text for self-supervised pre-training, such that we can effectively adapt the language model to a particular target task (e. g. question answering).
1 code implementation • NeurIPS 2020 • Geondo Park, June Yong Yang, Sung Ju Hwang, Eunho Yang
Neural networks embedded in safety-sensitive applications such as self-driving cars and wearable health monitors rely on two important techniques: input attribution for hindsight analysis and network compression to reduce its size for edge-computing.
no code implementations • 7 Dec 2020 • Byunggill Joe, Jihun Hamm, Sung Ju Hwang, Sooel Son, Insik Shin
Although deep neural networks have shown promising performances on various tasks, they are susceptible to incorrect predictions induced by imperceptibly small perturbations in inputs.
1 code implementation • ICLR 2021 • Seanie Lee, Dong Bok Lee, Sung Ju Hwang
In this work, we propose to mitigate the conditional text generation problem by contrasting positive pairs with negative pairs, such that the model is exposed to various valid or incorrect perturbations of the inputs, for improved generalization.
1 code implementation • ICLR 2021 • Dong Bok Lee, Dongchan Min, Seanie Lee, Sung Ju Hwang
Then, the learned model can be used for downstream few-shot classification tasks, where we obtain task-specific parameters by performing semi-supervised EM on the latent representations of the support and query set, and predict labels of the query set by computing aggregated posteriors.
no code implementations • 1 Jan 2021 • Seong Min Kye, Hae Beom Lee, Hoirin Kim, Sung Ju Hwang
A popular transductive inference technique for few-shot metric-based approaches, is to update the prototype of each class with the mean of the most confident query examples, or confidence-weighted average of all the query samples.
no code implementations • 1 Jan 2021 • Jeffrey Ryan Willette, Juho Lee, Sung Ju Hwang
We demonstrate the effectiveness of our method and validate its performance on both classification and regression problems by applying it to the training of recent state-of-the-art neural network models.
no code implementations • 1 Jan 2021 • Jung Hyun Lee, Jihun Yun, Sung Ju Hwang, Eunho Yang
As a natural extension of DropBits, we further introduce the way of learning heterogeneous quantization levels to find proper bit-length for each layer using DropBits.
no code implementations • 1 Jan 2021 • Minyoung Song, Jaehong Yoon, Eunho Yang, Sung Ju Hwang
As deep neural networks are growing in size and being increasingly deployed to more resource-limited devices, there has been a recent surge of interest in network pruning methods, which aim to remove less important weights or activations of a given network.
no code implementations • 7 Feb 2021 • Youngmin Oh, Jinwoo Shin, Eunho Yang, Sung Ju Hwang
We show that the proposed scheme, called Model-augmented $Q$-learning (MQL), obtains a policy-invariant solution which is identical to the solution obtained by learning with true reward.
no code implementations • 14 Feb 2021 • Jaewoong Shin, Hae Beom Lee, Boqing Gong, Sung Ju Hwang
Meta-learning of shared initialization parameters has shown to be highly effective in solving few-shot learning tasks.
no code implementations • 22 Feb 2021 • Jeffrey Willette, Juho Lee, Sung Ju Hwang
Neural networks have proven successful at learning from complex data distributions by acting as universal function approximators.
1 code implementation • ICLR 2021 • Jinheon Baek, Minki Kang, Sung Ju Hwang
Graph neural networks have been widely used on modeling graph data, achieving impressive results on node classification and link prediction tasks.
Ranked #1 on Graph Classification on ToxCast
1 code implementation • NeurIPS 2021 • Wonyong Jeong, Hayeon Lee, Gun Park, Eunyoung Hyung, Jinheon Baek, Sung Ju Hwang
To address such limitations, we introduce a novel problem of \emph{Neural Network Search} (NNS), whose goal is to search for the optimal pretrained network for a novel dataset and constraints (e. g. number of parameters), from a model zoo.
no code implementations • NeurIPS 2021 • Bruno Andreis, Jeffrey Willette, Juho Lee, Sung Ju Hwang
The proposed method adheres to the required symmetries of invariance and equivariance as well as maintaining MBC for any partition of the input set.
1 code implementation • ICML Workshop AML 2021 • Jihoon Tack, Sihyun Yu, Jongheon Jeong, Minseon Kim, Sung Ju Hwang, Jinwoo Shin
Adversarial training (AT) is currently one of the most successful methods to obtain the adversarial robustness of deep neural networks.
1 code implementation • ACL 2021 • Seanie Lee, Minki Kang, Juho Lee, Sung Ju Hwang
QA models based on pretrained language mod-els have achieved remarkable performance on various benchmark datasets. However, QA models do not generalize well to unseen data that falls outside the training distribution, due to distributional shifts. Data augmentation (DA) techniques which drop/replace words have shown to be effective in regularizing the model from overfitting to the training data. Yet, they may adversely affect the QA tasks since they incur semantic changes that may lead to wrong answers for the QA task.
no code implementations • ICLR 2022 • Jaehong Yoon, Divyam Madaan, Eunho Yang, Sung Ju Hwang
We validate the effectiveness of our coreset selection mechanism over various standard, imbalanced, and noisy datasets against strong continual learning baselines, demonstrating that it improves task adaptation and prevents catastrophic forgetting in a sample-efficient manner.
2 code implementations • 6 Jun 2021 • Dongchan Min, Dong Bok Lee, Eunho Yang, Sung Ju Hwang
In this work, we propose StyleSpeech, a new TTS model which not only synthesizes high-quality speech but also effectively adapts to new speakers.
1 code implementation • 11 Jun 2021 • Jongmin Yoon, Sung Ju Hwang, Juho Lee
Recently, an Energy-Based Model (EBM) trained with Markov-Chain Monte-Carlo (MCMC) has been highlighted as a purification model, where an attacked image is purified by running a long Markov-chain using the gradients of the EBM.
1 code implementation • 16 Jun 2021 • Hayeon Lee, Sewoong Lee, Song Chong, Sung Ju Hwang
To overcome such limitations, we propose Hardware-adaptive Efficient Latency Predictor (HELP), which formulates the device-specific latency estimation problem as a meta-learning problem, such that we can estimate the latency of a model's performance for a given task on an unseen device with a few samples.
no code implementations • ICML Workshop AML 2021 • Minseon Kim, Jihoon Tack, Jinwoo Shin, Sung Ju Hwang
Adversarial training methods, which minimizes the loss of adversarially-perturbed training examples, have been extensively studied as a solution to improve the robustness of the deep neural networks.
1 code implementation • NeurIPS 2021 • Jaehyeong Jo, Jinheon Baek, Seul Lee, DongKi Kim, Minki Kang, Sung Ju Hwang
This dual hypergraph construction allows us to apply message-passing techniques for node representations to edges.
1 code implementation • ICLR 2021 • Tehrim Yoon, Sumin Shin, Sung Ju Hwang, Eunho Yang
Federated learning (FL) allows edge devices to collectively learn a model without directly sharing data within each device, thus preserving privacy and eliminating the need to store data globally.
1 code implementation • ICLR 2021 • Hayeon Lee, Eunyoung Hyung, Sung Ju Hwang
Despite the success of recent Neural Architecture Search (NAS) methods on various tasks which have shown to output networks that largely outperform human-designed networks, conventional NAS methods have mostly tackled the optimization of searching for the network architecture for a single task (dataset), which does not generalize well across multiple tasks (datasets).
no code implementations • ICCV 2021 • Jung Hyun Lee, Jihun Yun, Sung Ju Hwang, Eunho Yang
Network quantization, which aims to reduce the bit-lengths of the network weights and activations, has emerged for their deployments to resource-limited devices.
no code implementations • ICLR 2022 • Youngmin Oh, Jinwoo Shin, Eunho Yang, Sung Ju Hwang
Experience replay is an essential component in off-policy model-free reinforcement learning (MfRL).
no code implementations • ICLR 2022 • Jeffrey Ryan Willette, Hae Beom Lee, Juho Lee, Sung Ju Hwang
Numerous recent works utilize bi-Lipschitz regularization of neural network layers to preserve relative distances between data instances in the feature spaces of each layer.
no code implementations • 29 Sep 2021 • Wonyong Jeong, Sung Ju Hwang
We then study two essential challenges of the agnostic personalized federated learning, which are (1) Label Heterogeneity where local clients learn from the same single domain but labeling schemes are not synchronized with each other and (2) Domain Heterogeneity where the clients learn from the different datasets which can be semantically similar or dissimilar for each other.
no code implementations • 29 Sep 2021 • Andreis Bruno, Seanie Lee, A. Tuan Nguyen, Juho Lee, Eunho Yang, Sung Ju Hwang
Deep Learning algorithms are designed to operate on huge volumes of high dimensional data such as images.
no code implementations • 29 Sep 2021 • Seul Lee, Dong Bok Lee, Sung Ju Hwang
To validate the ability to explore the chemical space beyond the known molecular distribution, we experiment with MOG to generate molecules of high absolute values of docking score, which is the affinity score based on a physical binding simulation between a target protein and a given molecule.
no code implementations • NeurIPS 2021 • Soojung Yang, Doyeong Hwang, Seul Lee, Seongok Ryu, Sung Ju Hwang
We further show with ablation studies that our method, predictive error-PER (FREED(PE)), significantly improves the model performance.
no code implementations • ICLR 2022 • Hae Beom Lee, Hayeon Lee, Jaewoong Shin, Eunho Yang, Timothy Hospedales, Sung Ju Hwang
Many gradient-based meta-learning methods assume a set of parameters that do not participate in inner-optimization, which can be considered as hyperparameters.
no code implementations • ICLR 2022 • Seanie Lee, Hae Beom Lee, Juho Lee, Sung Ju Hwang
Thanks to the gradients aligned between tasks by our method, the model becomes less vulnerable to negative transfer and catastrophic forgetting.
no code implementations • 12 Oct 2021 • Jeffrey Willette, Hae Beom Lee, Juho Lee, Sung Ju Hwang
Numerous recent works utilize bi-Lipschitz regularization of neural network layers to preserve relative distances between data instances in the feature spaces of each layer.
2 code implementations • ICLR 2022 • Divyam Madaan, Jaehong Yoon, Yuanchun Li, Yunxin Liu, Sung Ju Hwang
Continual learning (CL) aims to learn a sequence of tasks without forgetting the previously acquired knowledge.
no code implementations • 22 Nov 2021 • Taesik Gong, Yewon Kim, Adiba Orzikulova, Yunxin Liu, Sung Ju Hwang, Jinwoo Shin, Sung-Ju Lee
However, various factors such as different users, devices, and environments impact the performance of such applications, thus making the domain shift (i. e., distributional shift between the training domain and the target domain) a critical issue in mobile sensing.
1 code implementation • NeurIPS 2021 • Hayeon Lee, Sewoong Lee, Song Chong, Sung Ju Hwang
To overcome such limitations, we propose Hardware-adaptive Efficient Latency Predictor (HELP), which formulates the device-specific latency estimation problem as a meta-learning problem, such that we can estimate the latency of a model's performance for a given task on an unseen device with a few samples.
no code implementations • 16 Dec 2021 • Joonhyung Park, June Yong Yang, Jinwoo Shin, Sung Ju Hwang, Eunho Yang
However, they now suffer from lack of sample diversification as they always deterministically select regions with maximum saliency, injecting bias into the augmented data.
3 code implementations • CVPR 2022 • Youngwan Lee, Jonghee Kim, Jeff Willette, Sung Ju Hwang
While Convolutional Neural Networks (CNNs) have been the dominant architectures for such tasks, recently introduced Vision Transformers (ViTs) aim to replace them as a backbone.
Ranked #38 on Instance Segmentation on COCO minival
no code implementations • 1 Feb 2022 • Wonyong Jeong, Sung Ju Hwang
In real-world federated learning scenarios, participants could have their own personalized labels which are incompatible with those from other clients, due to using different label permutations or tackling completely different tasks or domains.
1 code implementation • 5 Feb 2022 • Jaehyeong Jo, Seul Lee, Sung Ju Hwang
Specifically, we propose a new graph diffusion process that models the joint distribution of the nodes and edges through a system of stochastic differential equations (SDEs).
1 code implementation • 7 Feb 2022 • DongKi Kim, Jinheon Baek, Sung Ju Hwang
Contrastive learning, while it can learn global graph-level similarities, its objective to maximize the similarity between two differently perturbed graphs may result in representations that cannot discriminate two similar graphs with different properties.
no code implementations • 23 Feb 2022 • Jaehong Yoon, Geon Park, Wonyong Jeong, Sung Ju Hwang
We introduce a pragmatic FL scenario with bitwidth heterogeneity across the participating devices, dubbed as Bitwidth Heterogeneous Federated Learning (BHFL).
1 code implementation • ACL 2022 • Soyeong Jeong, Jinheon Baek, Sukmin Cho, Sung Ju Hwang, Jong C. Park
Dense retrieval models, which aim at retrieving the most relevant document for an input query on a dense representation space, have gained considerable attention for their remarkable success.
Ranked #1000000000 on Passage Retrieval on Natural Questions
1 code implementation • NAACL 2022 • Minki Kang, Jinheon Baek, Sung Ju Hwang
Pre-trained language models (PLMs) have achieved remarkable success on various natural language understanding tasks.
no code implementations • ICLR 2022 • Taewook Nam, Shao-Hua Sun, Karl Pertsch, Sung Ju Hwang, Joseph J Lim
While deep reinforcement learning methods have shown impressive results in robot learning, their sample inefficiency makes the learning of complex, long-horizon behaviors with real robot systems infeasible.
no code implementations • 20 May 2022 • Seanie Lee, Bruno Andreis, Kenji Kawaguchi, Juho Lee, Sung Ju Hwang
Recently, several task augmentation methods have been proposed to tackle this issue using domain-specific knowledge to design augmentation techniques to densify the meta-training task distribution.
1 code implementation • 6 Jun 2022 • Seul Lee, Jaehyeong Jo, Sung Ju Hwang
A well-known limitation of existing molecular generative models is that the generated molecules highly resemble those in the training set.
no code implementations • 20 Jun 2022 • Hyunsu Rhee, Dongchan Min, Sunil Hwang, Bruno Andreis, Sung Ju Hwang
Real-time video segmentation is a crucial task for many real-world applications such as autonomous driving and robot control.
1 code implementation • 21 Jun 2022 • Jinheon Baek, Wonyong Jeong, Jiongdao Jin, Jaehong Yoon, Sung Ju Hwang
To this end, we introduce a new subgraph FL problem, personalized subgraph FL, which focuses on the joint improvement of the interrelated local GNNs rather than learning a single global model, and propose a novel framework, FEDerated Personalized sUBgraph learning (FED-PUB), to tackle it.
1 code implementation • International Conference on Machine Learning 2022 • Haeyong Kang, Rusty John Lloyd Mina, Sultan Rizky Hikmawan Madjid, Jaehong Yoon, Mark Hasegawa-Johnson, Sung Ju Hwang, Chang D. Yoo
Inspired by Lottery Ticket Hypothesis that competitive subnetworks exist within a dense network, we propose a continual learning method referred to as Winning SubNetworks (WSN), which sequentially learns and selects an optimal subnetwork for each task.
no code implementations • 4 Jul 2022 • Geon Park, Jaehong Yoon, Haiyang Zhang, Xing Zhang, Sung Ju Hwang, Yonina C. Eldar
Neural network quantization aims to transform high-precision weights and activations of a given neural network into low-precision weights/activations for reduced memory usage and computation, while preserving the performance of the original model.
1 code implementation • 21 Aug 2022 • Hae Beom Lee, Dong Bok Lee, Sung Ju Hwang
In this paper, we introduce a novel approach for systematically solving dataset condensation problem in an efficient manner by exploiting the regularity in a given dataset.
no code implementations • 23 Aug 2022 • Dongchan Min, Minyoung Song, Eunji Ko, Sung Ju Hwang
We propose StyleTalker, a novel audio-driven talking head generation model that can synthesize a video of a talking person from a single reference image with accurately audio-synced lip shapes, realistic head poses, and eye blinks.
no code implementations • 23 Aug 2022 • Jongha Kim, Jinheon Baek, Sung Ju Hwang
To achieve this, we first detect objects and then measure their semantic and spatial distances to construct an object graph, which is then represented by a graph neural network (GNN) for refining visual CNN features for objects.
1 code implementation • 26 Aug 2022 • Jeffrey Willette, Seanie Lee, Bruno Andreis, Kenji Kawaguchi, Juho Lee, Sung Ju Hwang
Recent work on mini-batch consistency (MBC) for set functions has brought attention to the need for sequentially processing and aggregating chunks of a partitioned set while guaranteeing the same output for all partitions.
2 code implementations • 15 Sep 2022 • Haeyong Kang, Jaehong Yoon, Sultan Rizky Hikmawan Madjid, Sung Ju Hwang, Chang D. Yoo
Inspired by Regularized Lottery Ticket Hypothesis (RLTH), which hypothesizes that there exist smooth (non-binary) subnetworks within a dense network that achieve the competitive performance of the dense network, we propose a few-shot class incremental learning (FSCIL) method referred to as \emph{Soft-SubNetworks (SoftNet)}.
no code implementations • 30 Sep 2022 • Seanie Lee, Minki Kang, Juho Lee, Sung Ju Hwang, Kenji Kawaguchi
Pre-training a large transformer model on a massive amount of unlabeled data and fine-tuning it on labeled datasets for diverse downstream tasks has proven to be a successful strategy, for a variety of vision and natural language processing tasks.
1 code implementation • 5 Oct 2022 • Youngwan Lee, Jeffrey Willette, Jonghee Kim, Juho Lee, Sung Ju Hwang
Masked image modeling (MIM) has become a popular strategy for self-supervised learning~(SSL) of visual representations with Vision Transformers.
no code implementations • 19 Oct 2022 • Minseon Kim, Hyeonjeong Ha, Dong Bok Lee, Sung Ju Hwang
Despite the success on few-shot learning problems, most meta-learned models only focus on achieving good performance on clean examples and thus easily break down when given adversarially perturbed samples.
1 code implementation • 19 Oct 2022 • Jin Myung Kwak, Minseon Kim, Sung Ju Hwang
Transformer-based Language Models (LMs) have achieved impressive results on natural language understanding tasks, but they can also generate toxic text such as insults, threats, and profanity, limiting their real-world applications.
no code implementations • 17 Nov 2022 • Minki Kang, Dongchan Min, Sung Ju Hwang
There has been a significant progress in Text-To-Speech (TTS) synthesis technology in recent years, thanks to the advancement in neural generative modeling.
1 code implementation • 19 Nov 2022 • Sunil Hwang, Jaehong Yoon, Youngwan Lee, Sung Ju Hwang
Masked Video Autoencoder (MVA) approaches have demonstrated their potential by significantly outperforming previous video representation learning methods.
Ranked #1 on Object State Change Classification on Ego4D
Object State Change Classification Object State Change Classification on Ego4D +4
1 code implementation • 7 Feb 2023 • Jaehyeong Jo, DongKi Kim, Sung Ju Hwang
Generation of graphs is a major challenge for real-world tasks that require understanding the complex nature of their non-Euclidean structures.
1 code implementation • 10 Feb 2023 • Soyeong Jeong, Jinheon Baek, Sung Ju Hwang, Jong C. Park
Conversational Question Answering (ConvQA) models aim at answering a question with its relevant paragraph and previous question-answer pairs that occurred during conversation multiple times.
1 code implementation • CVPR 2023 • Beomyoung Kim, JoonHyun Jeong, Dongyoon Han, Sung Ju Hwang
In this paper, we introduce a novel learning scheme named weakly semi-supervised instance segmentation (WSSIS) with point labels for budget-efficient and high-performance instance segmentation.
1 code implementation • 27 Mar 2023 • Haeyong Kang, Jaehong Yoon, Sultan Rizky Madjid, Sung Ju Hwang, Chang D. Yoo
Inspired by Regularized Lottery Ticket Hypothesis (RLTH), which states that competitive smooth (non-binary) subnetworks exist within a dense network in continual learning tasks, we investigate two proposed architecture-based continual learning methods which sequentially learn and select adaptive binary- (WSN) and non-binary Soft-Subnetworks (SoftNet) for each task.
no code implementations • ICCV 2023 • Jaewoong Lee, Sangwon Jang, Jaehyeong Jo, Jaehong Yoon, Yunji Kim, Jin-Hwa Kim, Jung-Woo Ha, Sung Ju Hwang
Token-based masked generative models are gaining popularity for their fast inference time with parallel decoding.
no code implementations • 21 May 2023 • Jinheon Baek, Alham Fikri Aji, Jens Lehmann, Sung Ju Hwang
There has been a surge of interest in utilizing Knowledge Graphs (KGs) for various natural language processing/understanding tasks.
no code implementations • 23 May 2023 • Minki Kang, Wooseok Han, Sung Ju Hwang, Eunho Yang
Emotional Text-To-Speech (TTS) is an important task in the development of systems (e. g., human-like dialogue agents) that require natural and emotional speech.
1 code implementation • 26 May 2023 • Sohyun An, Hayeon Lee, Jaehyeong Jo, Seanie Lee, Sung Ju Hwang
To tackle such limitations of existing NAS methods, we propose a paradigm shift from NAS to a novel conditional Neural Architecture Generation (NAG) framework based on diffusion models, dubbed DiffusionNAG.
no code implementations • 26 May 2023 • Bruno Andreis, Soro Bedionita, Sung Ju Hwang
We propose an approach to neural network weight encoding for generalization performance prediction that utilizes set-to-set and set-to-vector functions to efficiently encode neural network parameters.
1 code implementation • 26 May 2023 • Hayeon Lee, Rui Hou, Jongpil Kim, Davis Liang, Sung Ju Hwang, Alexander Min
Distillation from Weak Teacher (DWT) is a method of transferring knowledge from a smaller, weaker teacher model to a larger student model to improve its performance.
1 code implementation • 26 May 2023 • Hayeon Lee, Sohyun An, Minseon Kim, Sung Ju Hwang
Previous DaNAS methods have mostly tackled the search for the neural architecture for fixed datasets and the teacher, which are not generalized well on a new task consisting of an unseen dataset and an unseen teacher, thus need to perform a costly search for any new combination of the datasets and the teachers.
1 code implementation • NeurIPS 2023 • Minki Kang, Seanie Lee, Jinheon Baek, Kenji Kawaguchi, Sung Ju Hwang
Large Language Models (LLMs) have shown promising performance in knowledge-intensive reasoning tasks that require a compound understanding of knowledge.
no code implementations • 30 May 2023 • Minki Kang, Jin Myung Kwak, Jinheon Baek, Sung Ju Hwang
To overcome this limitation, we propose SUbgraph Retrieval-augmented GEneration (SURGE), a framework for generating context-relevant and knowledge-grounded dialogues with the KG.
no code implementations • 30 May 2023 • Doyeon Kim, Eunji Ko, Hyunsu Kim, Yunji Kim, Junho Kim, Dongchan Min, Junmo Kim, Sung Ju Hwang
Portrait stylization, which translates a real human face image into an artistically stylized image, has attracted considerable interest and many prior works have shown impressive quality in recent years.
1 code implementation • 7 Jun 2023 • Soyeong Jeong, Jinheon Baek, Sung Ju Hwang, Jong C. Park
To address this problem, we further introduce a novel contrastive learning strategy, making sure to reflect previous turns when retrieving the phrase for the current context, by maximizing representational similarities of consecutive turns in a conversation while minimizing irrelevant conversational contexts.
2 code implementations • 20 Jun 2023 • Haeyong Kang, Jaehong Yoon, Dahyun Kim, Sung Ju Hwang, Chang D Yoo
Motivated by continual learning, this work investigates how to accumulate and transfer neural implicit representations for multiple complex video data over sequential encoding sessions.
no code implementations • 21 Jun 2023 • Jaehong Yoon, Sung Ju Hwang, Yue Cao
We believe this paper breaks the barriers between pre-training and fine-tuning steps and leads to a sustainable learning framework in which the continual learner incrementally improves model generalization, yielding better transfer to unseen tasks.
no code implementations • 2 Oct 2023 • Seul Lee, Seanie Lee, Kenji Kawaguchi, Sung Ju Hwang
Additionally, the existing fragment-based generative models cannot update the fragment vocabulary with goal-aware fragments newly discovered during the generation.
1 code implementation • 3 Oct 2023 • Heejun Lee, Jina Kim, Jeffrey Willette, Sung Ju Hwang
SEA estimates the attention matrix with linear complexity via kernel-based linear attention, then subsequently creates a sparse attention matrix with a top-k selection to perform a sparse attention operation.
2 code implementations • 10 Oct 2023 • Dong Bok Lee, Seanie Lee, Joonho Ko, Kenji Kawaguchi, Juho Lee, Sung Ju Hwang
To achieve this, we also introduce the MSE between representations of the inner model and the self-supervised target model on the original full dataset for outer optimization.
no code implementations • 11 Oct 2023 • Jaehyeong Jo, Sung Ju Hwang
Learning the distribution of data on Riemannian manifolds is crucial for modeling data from non-Euclidean space, which is required by many applications from diverse scientific fields.
no code implementations • 12 Oct 2023 • Jaewoo Lee, Jaehong Yoon, Wonjae Kim, Yunji Kim, Sung Ju Hwang
Continuously learning a variety of audio-video semantics over time is crucial for audio-related reasoning tasks in our ever-evolving world.
1 code implementation • 19 Oct 2023 • Jinheon Baek, Soyeong Jeong, Minki Kang, Jong C. Park, Sung Ju Hwang
Recent Language Models (LMs) have shown impressive capabilities in generating texts with the knowledge internalized in parameters.
1 code implementation • 20 Oct 2023 • Soyeong Jeong, Jinheon Baek, Sukmin Cho, Sung Ju Hwang, Jong C. Park
Moreover, further finetuning LMs with labeled datasets is often infeasible due to their absence, but it is also questionable if we can transfer smaller LMs having limited knowledge only with unlabeled test data.
no code implementations • 6 Nov 2023 • Hayeon Lee, Rui Hou, Jongpil Kim, Davis Liang, Hongbo Zhang, Sung Ju Hwang, Alexander Min
2) The enhanced performance of the larger model further boosts the performance of the smaller model.
no code implementations • 13 Nov 2023 • Jin Myung Kwak, Minseon Kim, Sung Ju Hwang
Recent language models have achieved impressive performance in natural language tasks by incorporating instructions with task input during fine-tuning.
no code implementations • 14 Nov 2023 • Yujin Kim, Jaehong Yoon, Seonghyeon Ye, Sangmin Bae, Namgyu Ho, Sung Ju Hwang, Se-Young Yun
The dynamic nature of knowledge in an ever-changing world presents challenges for language models trained on static data; the model in the real world often requires not only acquiring new knowledge but also overwriting outdated information into updated ones.
no code implementations • 7 Dec 2023 • Youngwan Lee, KwanYong Park, Yoorhim Cho, Yong-Ju Lee, Sung Ju Hwang
We hope that due to its balanced speed-performance tradeoff, our KOALA models can serve as a cost-effective alternative to SDXL in resource-constrained environments.
no code implementations • 14 Dec 2023 • Taewook Nam, Juyong Lee, Jesse Zhang, Sung Ju Hwang, Joseph J. Lim, Karl Pertsch
We propose a framework that leverages foundation models as teachers, guiding a reinforcement learning agent to acquire semantically meaningful behavior without human feedback.
2 code implementations • 19 Dec 2023 • Haeyong Kang, Jaehong Yoon, Sung Ju Hwang, Chang D. Yoo
Inspired by the Lottery Ticket Hypothesis (LTH), which highlights the existence of efficient subnetworks within larger, dense networks, a high-performing Winning Subnetwork (WSN) in terms of task performance under appropriate sparsity conditions is considered for various continual learning tasks.
no code implementations • 13 Feb 2024 • Daeun Lee, Jaehong Yoon, Sung Ju Hwang
We validate our method outperforms multiple CTTA scenarios including disjoint and gradual domain shits, while only requiring ~98% fewer trainable parameters.
no code implementations • 21 Feb 2024 • Minju Seo, Jinheon Baek, James Thorne, Sung Ju Hwang
Many existing works tackle this problem by generating synthetic data from the training data and then training models on them, recently using Large Language Models (LLMs).
no code implementations • 28 Feb 2024 • Bedionita Soro, Bruno Andreis, Hayeon Lee, Song Chong, Frank Hutter, Sung Ju Hwang
By learning the distribution of a neural network on a variety pretrained models, our approach enables adaptive sampling weights for unseen datasets achieving faster convergence and reaching competitive performance.
1 code implementation • 21 Mar 2024 • Soyeong Jeong, Jinheon Baek, Sukmin Cho, Sung Ju Hwang, Jong C. Park
Retrieval-Augmented Large Language Models (LLMs), which incorporate the non-parametric knowledge from external knowledge bases into LLMs, have emerged as a promising approach to enhancing response accuracy in several tasks, such as Question-Answering (QA).
1 code implementation • 29 Mar 2024 • Beomyoung Kim, Joonsang Yu, Sung Ju Hwang
Panoptic segmentation, combining semantic and instance segmentation, stands as a cutting-edge computer vision task.
no code implementations • 1 Apr 2024 • Beomyoung Kim, Myeong Yeon Yi, Joonsang Yu, Young Joon Yoo, Sung Ju Hwang
To address this challenge, we introduce a new learning paradigm, weakly semi-supervised human matting (WSSHM), which leverages a small amount of expensive matte labels and a large amount of budget-friendly segmentation labels, to save the annotation cost and resolve the domain generalization problem.
no code implementations • 1 Apr 2024 • Beomyoung Kim, Donghyun Kim, Sung Ju Hwang
This paper presents a fresh perspective on the role of saliency maps in weakly-supervised semantic segmentation (WSSS) and offers new insights and research directions based on our empirical findings.
no code implementations • 5 Apr 2024 • Sangwon Jang, Jaehyeong Jo, Kimin Lee, Sung Ju Hwang
Our experiments demonstrate that MuDI can produce high-quality personalized images without identity mixing, even for highly similar subjects as shown in Figure 1.
no code implementations • 11 Apr 2024 • Jinheon Baek, Sujay Kumar Jauhar, Silviu Cucerzan, Sung Ju Hwang
Scientific Research, vital for improving human life, is hindered by its inherent complexity, slow pace, and the need for specialized experts.
1 code implementation • ICML 2020 • Jay Heo, Junhyeon Park, Hyewon Jeong, Kwang Joon Kim, Juho Lee, Eunho Yang, Sung Ju Hwang
Moreover, it is almost infeasible for the human annotators to examine attentions on tons of instances and features.
no code implementations • ICML 2020 • Seong-Jin Park, Seungju Han, Ji-won 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 nonrigid transformation, background noise, and change in lighting conditions.