2 code implementations • 28 Nov 2022 • Dongjun Kim, Yeongmin Kim, Se Jung Kwon, Wanmo Kang, Il-Chul Moon
In sample generation, we add an auxiliary term to the pre-trained score to deceive the discriminator.
Ranked #1 on Conditional Image Generation on CIFAR-10
2 code implementations • ICLR 2020 • Cheolhyoung Lee, Kyunghyun Cho, Wanmo Kang
We empirically evaluate the proposed mixout and its variants on finetuning a pretrained language model on downstream tasks.
1 code implementation • 10 Jun 2021 • Dongjun Kim, Seungjae Shin, Kyungwoo Song, Wanmo Kang, Il-Chul Moon
This paper investigates with sufficient empirical evidence that such inverse correlation happens because density estimation is significantly contributed by small diffusion time, whereas sample generation mainly depends on large diffusion time.
Ranked #2 on Image Generation on CIFAR-10 (Inception score metric)
1 code implementation • 27 May 2022 • Dongjun Kim, Byeonghu Na, Se Jung Kwon, Dongsoo Lee, Wanmo Kang, Il-Chul Moon
Whereas diverse variations of diffusion models exist, extending the linear diffusion into a nonlinear diffusion process is investigated by very few works.
Ranked #4 on Image Generation on CelebA 64x64
1 code implementation • 27 Feb 2024 • Byeonghu Na, Yeongmin Kim, HeeSun Bae, Jung Hyun Lee, Se Jung Kwon, Wanmo Kang, Il-Chul Moon
This paper proposes Transition-aware weighted Denoising Score Matching (TDSM) for training conditional diffusion models with noisy labels, which is the first study in the line of diffusion models.
1 code implementation • Proceedings of the 40th International Conference on Machine Learning 2023 • Yoon-Yeong Kim, Youngjae Cho, JoonHo Jang, Byeonghu Na, Yeongmin Kim, Kyungwoo Song, Wanmo Kang, Il-Chul Moon
Specifically, our proposed method, Sharpness-Aware Active Learning (SAAL), constructs its acquisition function by selecting unlabeled instances whose perturbed loss becomes maximum.
1 code implementation • 2 Mar 2024 • Yeongmin Kim, Byeonghu Na, Minsang Park, JoonHo Jang, Dongjun Kim, Wanmo Kang, Il-Chul Moon
While directly applying it to score-matching is intractable, we discover that using the time-dependent density ratio both for reweighting and score correction can lead to a tractable form of the objective function to regenerate the unbiased data density.
1 code implementation • 15 Oct 2020 • Dongjun Kim, Kyungwoo Song, YoonYeong Kim, Yongjin Shin, Wanmo Kang, Il-Chul Moon, Weonyoung Joo
This paper introduces a new sampling approach, called Neural Proposal (NP), of the simulation input that resolves the biased data collection as it guarantees the i. i. d.
1 code implementation • 15 Feb 2021 • Dongjun Kim, Kyungwoo Song, Seungjae Shin, Wanmo Kang, Il-Chul Moon, Weonyoung Joo
A simulation is useful when the phenomenon of interest is either expensive to regenerate or irreproducible with the same context.
no code implementations • ICLR 2019 • Cheolhyoung Lee, Kyunghyun Cho, Wanmo Kang
We empirically verify our result using deep convolutional networks and observe a higher correlation between the gradient stochasticity and the proposed directional uniformity than that against the gradient norm stochasticity, suggesting that the directional statistics of minibatch gradients is a major factor behind SGD.
no code implementations • 24 Nov 2020 • Hyemi Kim, Seungjae Shin, JoonHo Jang, Kyungwoo Song, Weonyoung Joo, Wanmo Kang, Il-Chul Moon
Therefore, this paper proposes Disentangled Causal Effect Variational Autoencoder (DCEVAE) to resolve this limitation by disentangling the exogenous uncertainty into two latent variables: either 1) independent to interventions or 2) correlated to interventions without causality.
no code implementations • 29 Sep 2021 • Dongjun Kim, Seungjae Shin, Kyungwoo Song, Wanmo Kang, Il-Chul Moon
From the theory side, the difficulty arises in estimating the high precision diffusion because the data score goes to $\infty$ as $t \rightarrow 0$ of the diffusion time.
no code implementations • 29 Sep 2021 • Dongjun Kim, Byeonghu Na, Se Jung Kwon, Dongsoo Lee, Wanmo Kang, Il-Chul Moon
Specifically, PDM utilizes the flow to non-linearly transform a data variable into a latent variable, and PDM applies the diffusion process to the transformed latent distribution with the linear diffusing mechanism.