서성욱 교수

We propose a simple but effective multi-source domain generalization technique based on deep neural networks by incorporating optimized normalization layers that are specific to individual domains. Our approach employs multiple normalization methods while learning separate affine parameters per domain. For each domain, the activations are normalized by a weighted average of multiple normalization statistics. The normalization statistics are kept track of separately for each normalization type if

We propose a novel contrastive learning framework to effectively address the challenges of data heterogeneity infederated learning. We first analyze the inconsistency of gradient updates across clients during local training and establish its dependence on the distribution of feature representations, leading to the derivation of the supervised contrastive learning (SCL) objective to mitigate local deviations. In addition, we show that a naïve integration of SCL into federated learning incurs repr

Dataset bias is a critical challenge in machine learning since it often leads to a negative impact on a model due to the unintended decision rules captured by spurious correlations. Although existing works often handle this issue based on human supervision, the availability of the proper annotations is impractical and even unrealistic. To better tackle the limitation, we propose a simple but effective unsupervised debiasing technique. Specifically, we first identify pseudo-attributes based on th

We propose an information-theoretic bias measurement technique through a causal interpretation of spurious correlation, which is effective to identify the feature-level algorithmic bias by taking advantage of conditional mutual information. Although several bias measurement methods have been proposed and widely investigated to achieve algorithmic fairness in various tasks such as face recognition, their accuracy- or logit-based metrics are susceptible to leading to trivial prediction score adjus

In the present paper, Computational fluid dynamics (CFD) uncertainty analysis proposed by ITTC was investigated and applied to ship resistance performance using open source libraries, called OpenFOAM. Uncertainties for grid size, time step and iteration number were studied. Wave patterns and hull wave profile were compared for various uncertainty parameters. From results, grid size uncertainty was mainly contributed to simulation numerical uncertainty.

We propose a generic framework to calibrate accuracy and confidence of a prediction in deep neural networks through stochastic inferences. We interpret stochastic regularization using a Bayesian model, and analyze the relation between predictive uncertainty of networks and variance of the prediction scores obtained by stochastic inferences for a single example. Our empirical study shows that the accuracy and the score of a prediction are highly correlated with the variance of multiple stochastic

We propose a generic framework to calibrate accuracy and confidence of a prediction in deep neural networks through stochastic inferences. We interpret stochastic regularization using a Bayesian model, and analyze the relation between predictive uncertainty of networks and variance of the prediction scores obtained by stochastic inferences for a single example. Our empirical study shows that the accuracy and the score of a prediction are highly correlated with the variance of multiple stochastic

Machine unlearning, an emerging research topic focusing on data privacy compliance, enables trained models to erase information learned from specific data. While many existing methods indirectly address this issue by intentionally injecting incorrect supervision, they often result in drastic and unpredictable changes to decision boundaries and feature spaces, leading to training instability and undesired side effects. To address this challenge more fundamentally, we first analyze the changes in

The increasing demand for online advertising leads to a strong competition in Real-Time Biddinng (RTB) industry. It requires Demand-Side Platforms (DSPs) to perform a proper market price modeling that predicts the landscape of competitors’ bids, in order to maximize their profits. Under this circumstance, RTB industry has recently been changing from second-price auctions (SPA) to first-price auctions (FPA), and thus DSPs now face two different auction types simultaneously. Most previous studies

Dataset bias is a critical challenge in machine learning since it often leads to a negative impact on a model due to the unintended decision rules captured by spurious correlations. Although existing works often handle this issue based on human supervision, the availability of the proper annotations is impractical and even unrealistic. To better tackle the limitation, we propose a simple but effective unsupervised debiasing technique. Specifically, we first identify pseudo-attributes based on th

We propose a simple but effective multi-source domain generalization technique based on deep neural networks by incorporating optimized normalization layers specific to individual domains. Our approach employs multiple normalization methods while learning a separate affine parameter per domain. For each domain, the activations are normalized by a weighted average of multiple normalization statistics. The normalization statistics are kept track of separately for each normalization type if necessa