Search Results for author:
Found 14 papers, 2 papers with code
Analysis of Linear Mode Connectivity via Permutation-Based Weight Matching
This finding shows that permutations found by WM mainly align the directions of singular vectors associated with large singular values across models.
One-Shot Machine Unlearning with Mnemonic Code
A naive MU approach is to re-train the whole model with the training data from which the undesirable data has been removed.
Revisiting Permutation Symmetry for Merging Models between Different Datasets
We also show that merged models require datasets for merging in order to achieve a high accuracy.
ARDIR: Improving Robustness using Knowledge Distillation of Internal Representation
In addition to the output of the teacher model, ARDIR uses the internal representation of the teacher model as a label for adversarial training.
One-vs-the-Rest Loss to Focus on Important Samples in Adversarial Training
This paper proposes a new loss function for adversarial training.
Smoothness Analysis of Adversarial Training
We reveal that the constraint of adversarial attacks is one cause of the non-smoothness and that the smoothness depends on the types of the constraints.
Adversarial Training Makes Weight Loss Landscape Sharper in Logistic Regression
We theoretically and experimentally confirm that the weight loss landscape becomes sharper as the magnitude of the noise of adversarial training increases in the linear logistic regression model.
Constraining Logits by Bounded Function for Adversarial Robustness
We theoretically and empirically reveal that small logits by addition of a common activation function, e. g., hyperbolic tangent, do not improve adversarial robustness since input vectors of the function (pre-logit vectors) can have large norms.
Absum: Simple Regularization Method for Reducing Structural Sensitivity of Convolutional Neural Networks
Furthermore, we reveal that robust CNNs with Absum are more robust against transferred attacks due to decreasing the common sensitivity and against high-frequency noise than standard regularization methods.
FAVAE: SEQUENCE DISENTANGLEMENT USING IN- FORMATION BOTTLENECK PRINCIPLE
Previous works succeed in disentangling static factors and dynamic factors by explicitly modeling the priors of latent variables to distinguish between static and dynamic factors.
Autoencoding Binary Classifiers for Supervised Anomaly Detection
Since our approach becomes able to reconstruct the normal data points accurately and fails to reconstruct the known and unknown anomalies, it can accurately discriminate both known and unknown anomalies from normal data points.
Macro Action Reinforcement Learning with Sequence Disentanglement using Variational Autoencoder
Macro actions, a sequence of primitive actions, have been studied to diminish the dimensionality of the action space with regard to the time axis.
FAVAE: Sequence Disentanglement using Information Bottleneck Principle
Previous models disentangle static and dynamic factors by explicitly modeling the priors of latent variables to distinguish between these factors.
Variational Autoencoder with Implicit Optimal Priors
However, KL divergence with the aggregated posterior cannot be calculated in a closed form, which prevents us from using this optimal prior.
