Results 61 to 70 of about 34,476 (259)

Squeeze Training for Adversarial Robustness

, 2022
The vulnerability of deep neural networks (DNNs) to adversarial examples has attracted great attention in the machine learning community. The problem is related to non-flatness and non-smoothness of normally obtained loss landscapes. Training augmented with adversarial examples (a.k.a., adversarial training) is considered as an effective remedy.
Qizhang Li, Yiwen Guo, Wangmeng Zuo, Hao Chen 0003   +3 more
openaire   +3 more sources

Imbalanced Adversarial Training with Reweighting

2022 IEEE International Conference on Data Mining (ICDM), 2022
Adversarial training has been empirically proven to be one of the most effective and reliable defense methods against adversarial attacks. However, almost all existing studies about adversarial training are focused on balanced datasets, where each class has an equal amount of training examples.
Wentao Wang 0006, Han Xu 0002, Xiaorui Liu, Yaxin Li 0001, Bhavani Thuraisingham, Jiliang Tang   +5 more
openaire   +2 more sources

From the Discovery of the Giant Magnetocaloric Effect to the Development of High‐Power‐Density Systems

Advanced Materials Technologies, EarlyView.
The article overviews past and current efforts on caloric materials and systems, highlighting the contributions of Ames National Laboratory to the field. Solid‐state caloric heat pumping is an innovative method that can be implemented in a wide range of cooling and heating applications.
Agata Czernuszewicz, Yaroslav Mudryk, Jun Cui, Lucas Griffith, Duane D. Johnson, Julie Slaughter   +5 more
wiley   +1 more source

A Two-Stage Adversarial Training Method Based on Stability Contrastive Learning to Enhance Adversarial Robustness

Applied Sciences
Neural network models are highly susceptible to adversarial sample attacks, causing significant differences in model predictions with even minor perturbations to the samples.
Wenjuan Ren, Zhanpeng Yang, Guangzuo Li
doaj   +1 more source

Adversarial Training for Relation Extraction [PDF]

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, 2017
Adversarial training is a mean of regularizing classification algorithms by generating adversarial noise to the training data. We apply adversarial training in relation extraction within the multi-instance multi-label learning framework. We evaluate various neural network architectures on two different datasets.
Yi Wu 0013, David Bamman, Stuart Russell 0001   +2 more
openaire   +1 more source

Learning Highly Dynamic Skills Transition for Quadruped Jumping Through Constrained Space

Advanced Robotics Research, EarlyView.
A quadruped robot masters dynamic jumps through constrained spaces with animal‐inspired moves and intelligent vision control. This hierarchical learning approach combines imitation of biological agility with real‐time trajectory planning. Although legged animals are capable of performing explosive motions while traversing confined spaces, replicating ...
Zeren Luo, Jiahui Zhang, Yimin Han, Ji Ma, Minghao Lu, Ioannis Havoutis, Peng Lu   +6 more
wiley   +1 more source

ATVis: Understanding and diagnosing adversarial training processes through visual analytics

Visual Informatics
Adversarial training has emerged as a major strategy against adversarial perturbations in deep neural networks, which mitigates the issue of exploiting model vulnerabilities to generate incorrect predictions.
Fang Zhu, Xufei Zhu, Xumeng Wang, Yuxin Ma, Jieqiong Zhao   +4 more
doaj   +1 more source

Stable Imitation of Multigait and Bipedal Motions for Quadrupedal Robots Over Uneven Terrains

Advanced Robotics Research, EarlyView.
How are quadrupedal robots empowered to execute complex navigation tasks, including multigait and bipedal motions? Challenges in stability and real‐world adaptation persist, especially with uneven terrains and disturbances. This article presents an imitation learning framework that enhances adaptability and robustness by incorporating long short‐term ...
Erdong Xiao, Yinzhao Dong, Ji Ma, Peng Lu   +3 more
wiley   +1 more source

Adversarial Training: A Survey

CoRR
Adversarial training (AT) refers to integrating adversarial examples -- inputs altered with imperceptible perturbations that can significantly impact model predictions -- into the training process. Recent studies have demonstrated the effectiveness of AT in improving the robustness of deep neural networks against diverse adversarial attacks. However, a
Mengnan Zhao 0001, Lihe Zhang, Jingwen Ye, Huchuan Lu, Baocai Yin, Xinchao Wang   +5 more
openaire   +2 more sources

On the Properties of Adversarially-Trained CNNs

CoRR, 2022
Adversarial Training has proved to be an effective training paradigm to enforce robustness against adversarial examples in modern neural network architectures. Despite many efforts, explanations of the foundational principles underpinning the effectiveness of Adversarial Training are limited and far from being widely accepted by the Deep Learning ...
Mattia Carletti, Matteo Terzi, Gian Antonio Susto   +2 more
openaire   +2 more sources