Picture for Taejong Joo

Taejong Joo

Improving self-training under distribution shifts via anchored confidence with theoretical guarantees

Add code
Nov 01, 2024
Viaarxiv icon

IW-GAE: Importance weighted group accuracy estimation for improved calibration and model selection in unsupervised domain adaptation

Add code
Oct 16, 2023
Viaarxiv icon

Deep Learning Requires Explicit Regularization for Reliable Predictive Probability

Add code
Jun 11, 2020
Figure 1 for Deep Learning Requires Explicit Regularization for Reliable Predictive Probability
Figure 2 for Deep Learning Requires Explicit Regularization for Reliable Predictive Probability
Figure 3 for Deep Learning Requires Explicit Regularization for Reliable Predictive Probability
Figure 4 for Deep Learning Requires Explicit Regularization for Reliable Predictive Probability
Viaarxiv icon

Being Bayesian about Categorical Probability

Add code
Feb 19, 2020
Figure 1 for Being Bayesian about Categorical Probability
Figure 2 for Being Bayesian about Categorical Probability
Figure 3 for Being Bayesian about Categorical Probability
Figure 4 for Being Bayesian about Categorical Probability
Viaarxiv icon

Regularizing activations in neural networks via distribution matching with the Wasserstein metric

Add code
Feb 13, 2020
Figure 1 for Regularizing activations in neural networks via distribution matching with the Wasserstein metric
Figure 2 for Regularizing activations in neural networks via distribution matching with the Wasserstein metric
Figure 3 for Regularizing activations in neural networks via distribution matching with the Wasserstein metric
Figure 4 for Regularizing activations in neural networks via distribution matching with the Wasserstein metric
Viaarxiv icon