Zobrazeno 1 - 10
of 58
pro vyhledávání: '"Jeong, Jongheon"'
Adversarial robustness has been conventionally believed as a challenging property to encode for neural networks, requiring plenty of training data. In the recent paradigm of adopting off-the-shelf models, however, access to their training data is oft
Externí odkaz:
http://arxiv.org/abs/2407.18658
Modern alignment techniques based on human preferences, such as RLHF and DPO, typically employ divergence regularization relative to the reference model to ensure training stability. However, this often limits the flexibility of models during alignme
Externí odkaz:
http://arxiv.org/abs/2406.06424
Autor:
Kim, Kyuyoung, Jeong, Jongheon, An, Minyong, Ghavamzadeh, Mohammad, Dvijotham, Krishnamurthy, Shin, Jinwoo, Lee, Kimin
Fine-tuning text-to-image models with reward functions trained on human feedback data has proven effective for aligning model behavior with human intent. However, excessive optimization with such reward models, which serve as mere proxy objectives, c
Externí odkaz:
http://arxiv.org/abs/2404.01863
Autor:
Jeong, Jongheon, Shin, Jinwoo
Along with recent diffusion models, randomized smoothing has become one of a few tangible approaches that offers adversarial robustness to models at scale, e.g., those of large pre-trained models. Specifically, one can perform randomized smoothing on
Externí odkaz:
http://arxiv.org/abs/2310.16779
Despite its practical importance across a wide range of modalities, recent advances in self-supervised learning (SSL) have been primarily focused on a few well-curated domains, e.g., vision and language, often relying on their domain-specific knowled
Externí odkaz:
http://arxiv.org/abs/2310.16318
Generative priors of large-scale text-to-image diffusion models enable a wide range of new generation and editing applications on diverse visual modalities. However, when adapting these priors to complex visual modalities, often represented as multip
Externí odkaz:
http://arxiv.org/abs/2307.04787
Autor:
Jeong, Jongheon, Zou, Yang, Kim, Taewan, Zhang, Dongqing, Ravichandran, Avinash, Dabeer, Onkar
Visual anomaly classification and segmentation are vital for automating industrial quality inspection. The focus of prior research in the field has been on training custom models for each quality inspection task, which requires task-specific images a
Externí odkaz:
http://arxiv.org/abs/2303.14814
In practical scenarios where training data is limited, many predictive signals in the data can be rather from some biases in data acquisition (i.e., less generalizable), so that one cannot prevent a model from co-adapting on such (so-called) "shortcu
Externí odkaz:
http://arxiv.org/abs/2303.14096
An energy-based model (EBM) is a popular generative framework that offers both explicit density and architectural flexibility, but training them is difficult since it is often unstable and time-consuming. In recent years, various training techniques
Externí odkaz:
http://arxiv.org/abs/2303.03023
Any classifier can be "smoothed out" under Gaussian noise to build a new classifier that is provably robust to $\ell_2$-adversarial perturbations, viz., by averaging its predictions over the noise via randomized smoothing. Under the smoothed classifi
Externí odkaz:
http://arxiv.org/abs/2212.09000