Results 41 to 50 of about 8,712 (199)

Vax-a-Net: Training-Time Defence Against Adversarial Patch Attacks [PDF]

, 2021
We present Vax-a-Net; a technique for immunizing convolutional neural networks (CNNs) against adversarial patch attacks (APAs). APAs insert visually overt, local regions (patches) into an image to induce misclassification. We introduce a conditional Generative Adversarial Network (GAN) architecture that simultaneously learns to synthesise patches for ...
Thomas Gittings, Steve Schneider, John P. Collomosse   +2 more
openaire   +2 more sources

PatchGuard++: Efficient Provable Attack Detection against Adversarial Patches

CoRR, 2021
ICLR 2021 Workshop on Security and Safety in Machine Learning ...
Chong Xiang 0001, Prateek Mittal
openaire   +2 more sources

Learnable Diffusion Framework for Mouse V1 Neural Decoding

Advanced Science, EarlyView.
We introduce Sensorium‐Viz, a diffusion‐based framework for reconstructing high‐fidelity visual stimuli from mouse primary visual cortex activity. By integrating a novel spatial embedding module with a Diffusion Transformer (DiT) and a synthetic‐response augmentation strategy, our model outperforms state‐of‐the‐art fMRI‐based baselines, enabling robust
Kaiwen Deng, Peter S. Schwendeman, Yuanfang Guan   +2 more
wiley   +1 more source

Multimodal Wearable Biosensing Meets Multidomain AI: A Pathway to Decentralized Healthcare

Advanced Science, EarlyView.
Multimodal biosensing meets multidomain AI. Wearable biosensors capture complementary biochemical and physiological signals, while cross‐device, population‐aware learning aligns noisy, heterogeneous streams. This Review distills key sensing modalities, fusion and calibration strategies, and privacy‐preserving deployment pathways that transform ...
Chenshu Liu, Haolin Fan, Minwoo Kim, Tong Zhou, Pinyi Yang, Lingdi Zhao, Yiran Wang, Ziyuan Che, Chia‐Wei Liu, Bingbing Li, Yangzhi Zhu   +10 more
wiley   +1 more source

Adaptive Rotation-Scaling Ensemble Patch Attack for Ship Detection in Remote Sensing Images

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Existing adversarial patch generation methods for remote sensing ship detection suffer from several limitations, including insufficient adaptability to patch size, inability to handle orientation variations of rotated ships, and poor transferability of ...
Qi Wang, Jiahao Cui, Keyi Zhang, Lingquan Meng, Haifeng Li, Chengli Peng   +5 more
doaj   +1 more source

PAD: Patch-Agnostic Defense against Adversarial Patch Attacks

2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Accepted by CVPR ...
Lihua Jing, Rui Wang 0032, Wenqi Ren, Xin Dong 0015, Cong Zou   +4 more
openaire   +2 more sources

Machine Learning Interatomic Potentials for Energy Materials: Architectures, Training Strategies, and Applications

Advanced Energy Materials, EarlyView.
Machine learning interatomic potentials bridge quantum accuracy and computational efficiency for materials discovery. Architectures from Gaussian process regression to equivariant graph neural networks, training strategies including active learning and foundation models, and applications in solid‐state electrolytes, batteries, electrocatalysts ...
In Kee Park, Saeed Pourasad, Jinhong Mun, Arunkumar Chitteth Rajan, Tae Hyeon Park, Anand Rohit, Mohammad Zafari, Tan‐Lien Pham, Eunjae Jung, Bonseung Koo, Miran Ha, Gi Beom Sim, Hyun Gyu Park, Won Woong Choi, Kwanghyeon Jo, Soohaeng Yoo Willow, David ChangMo Yang, Chang Woo Myung, Geunsik Lee, Kwang S. Kim   +19 more
wiley   +1 more source

A Cascade Defense Method for Multidomain Adversarial Attacks under Remote Sensing Detection

Remote Sensing, 2022
Deep neural networks have been widely used in detection tasks based on optical remote sensing images. However, in recent studies, deep neural networks have been shown to be vulnerable to adversarial examples.
Wei Xue, Zhiming Chen, Weiwei Tian, Yunhua Wu, Bing Hua   +4 more
doaj   +1 more source

Deep Learning‐Assisted Coherent Raman Scattering Microscopy

Advanced Intelligent Discovery, EarlyView.
The analytical capabilities of coherent Raman scattering microscopy are augmented through deep learning integration. This synergistic paradigm improves fundamental performance via denoising, deconvolution, and hyperspectral unmixing. Concurrently, it enhances downstream image analysis including subcellular localization, virtual staining, and clinical ...
Jianlin Liu, Rujia Liu, Xinyan Zhang, Xun Chen, Shuhua Yue   +4 more
wiley   +1 more source

Adversarial Halftone QR Code

IEEE Access
Recent studies have shown that machine-learning models are vulnerable to adversarial attacks. Adversarial attacks are deliberate attempts to modify the input data of a machine learning model in a way that causes it to produce incorrect predictions.
Palakorn Kamnounsing, Karin Sumongkayothin, Prarinya Siritanawan, Kazunori Kotani   +3 more
doaj   +1 more source