Filters








2,465 Hits in 3.7 sec

Backpropagated Gradient Representations for Anomaly Detection [article]

Gukyeong Kwon, Mohit Prabhushankar, Dogancan Temel, Ghassan AlRegib
2020 arXiv   pre-print
Hence, we propose the utilization of backpropagated gradients as representations to characterize model behavior on anomalies and, consequently, detect such anomalies.  ...  Most of existing anomaly detection algorithms use activation representations from forward propagation while not exploiting gradients from backpropagation to characterize data.  ...  The gradients as representations have not been actively explored for anomaly detection.  ... 
arXiv:2007.09507v1 fatcat:hd44q5gv3nbyfh2xy44efh2cu4

Novelty Detection Through Model-Based Characterization of Neural Networks [article]

Gukyeong Kwon, Mohit Prabhushankar, Dogancan Temel, Ghassan AlRegib
2020 arXiv   pre-print
To articulate the significance of the model perspective in novelty detection, we utilize backpropagated gradients.  ...  We conduct a comprehensive analysis to compare the representation capability of gradients with that of activation and show that the gradients outperform the activation in novel class and condition detection  ...  This indicates that different classes of anomalies are separated and characterized robustly using the backpropagated gradients.  ... 
arXiv:2008.06094v1 fatcat:e4tpqry5szd5fgehk6ym76igvi

Gradient-Based Adversarial and Out-of-Distribution Detection [article]

Jinsol Lee, Mohit Prabhushankar, Ghassan AlRegib
2022 arXiv   pre-print
We propose to utilize gradients for detecting adversarial and out-of-distribution samples.  ...  state-of-the-art methods for adversarial and out-of-distribution detection.  ...  Experiments In this section, we utilize gradient-based representations generated in response to confounding labels for detecting anomalous inputs: adversarial detection and out-ofdistribution detection  ... 
arXiv:2206.08255v2 fatcat:os55tbr46zaitm2sen4pqmcz24

Towards Visually Explaining Variational Autoencoders

Wenqian Liu, Runze Li, Meng Zheng, Srikrishna Karanam, Ziyan Wu, Bir Bhanu, Richard J. Radke, Octavia Camps
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
In particular, gradient-based visual attention methods have driven much recent effort in using visual attention maps as a means for visual explanations.  ...  We show how these attention maps can be used to localize anomalies in images, demonstrating state-of-the-art performance on the MVTec-AD dataset.  ...  Anomaly Detection. Unsupervised learning for anomaly detection [1] still remains challenging.  ... 
doi:10.1109/cvpr42600.2020.00867 dblp:conf/cvpr/LiuLZKWBRC20 fatcat:ic3yv2knd5aehcmezxyb5s3e4u

Context-encoding Variational Autoencoder for Unsupervised Anomaly Detection [article]

David Zimmerer, Simon A. A. Kohl, Jens Petersen, Fabian Isensee, Klaus H. Maier-Hein
2018 arXiv   pre-print
However, state-of-the-art anomaly scores are still based on the reconstruction error, which lacks in two essential parts: it ignores the model-internal representation employed for reconstruction, and it  ...  In this context, deep learning-based auto encoders have shown great potential in detecting anomalies in medical images.  ...  anomaly-detection for images [7, 11, 15] .  ... 
arXiv:1812.05941v1 fatcat:xnellrlzo5g6tdnj3nhr5jzgva

Anomaly Detection for Solder Joints Using β-VAE [article]

Furkan Ulger, Seniha Esen Yuksel, Atila Yilmaz
2021 arXiv   pre-print
We compare the activation and gradient-based representations that are used to characterize anomalies; and observe the effect of different beta parameters on accuracy and on untwining the feature representations  ...  anomaly detection that can work on both IC and non-IC components.  ...  [26] proposed training autoencoders with gradient constraint to model normal data distributions and using gradient-based representations for anomaly detection, motivated to capture information unavailable  ... 
arXiv:2104.11927v1 fatcat:cmsiso4vongvvfzrqchlpf4idu

Towards Visually Explaining Variational Autoencoders [article]

Wenqian Liu, Runze Li, Meng Zheng, Srikrishna Karanam, Ziyan Wu, Bir Bhanu, Richard J. Radke, Octavia Camps
2020 arXiv   pre-print
In particular, gradient-based visual attention methods have driven much recent effort in using visual attention maps as a means for visual explanations.  ...  We show how these attention maps can be used to localize anomalies in images, demonstrating state-of-the-art performance on the MVTec-AD dataset.  ...  Anomaly Detection. Unsupervised learning for anomaly detection [1] still remains challenging.  ... 
arXiv:1911.07389v7 fatcat:glr3maratbdrzoqgu2g4kxfmny

Auditing Visualizations: Transparency Methods Struggle to Detect Anomalous Behavior [article]

Jean-Stanislas Denain, Jacob Steinhardt
2022 arXiv   pre-print
We find that while existing methods can detect stark anomalies such as shape bias or adversarial training, they struggle to identify more subtle anomalies such as models trained on incomplete data.  ...  For instance, can they diagnose abnormal behavior such as backdoors or shape bias?  ...  Similarly for the Spurious Features anomalies (see Figure 4 for qualitative observations), the AUROCs of Guided Backpropagation and Integrated Gradients are greater than 0.75 for δ ∈ {0.1, 0.3, 0.5},  ... 
arXiv:2206.13498v1 fatcat:phte2jue7fesfd6todapsvnvc4

Perceptual Image Anomaly Detection [article]

Nina Tuluptceva, Bart Bakker, Irina Fedulova, Anton Konushin
2019 arXiv   pre-print
We present a novel method for image anomaly detection, where algorithms that use samples drawn from some distribution of "normal" data, aim to detect out-of-distribution (abnormal) samples.  ...  It leverages Generative Adversarial Networks to learn these data distributions and uses perceptual loss for the detection of image abnormality.  ...  Reset gradients. 16: Backpropagate L rec (G, E); multiply gradients of G by γ G , E by γ E . 17: Backpropagate L adv (G), L adv (E).  ... 
arXiv:1909.05904v1 fatcat:dlzcp7s6nvf3rndw3vvbnrerqq

Deep Vision for Breast Cancer Classification and Segmentation

Lawrence Fulton, Alex McLeod, Diane Dolezel, Nathaniel Bastian, Christopher P. Fulton
2021 Cancers  
A small image representation from the fitted training model is returned to evaluate the portion of the loss function gradient with respect to the image that maximizes the classification probability.  ...  This gradient is then re-mapped back to the original images, highlighting the areas of the original image that are most influential for classification (perhaps masses or boundary areas). (3) Results: initial  ...  The address for access follows: https://www.kaggle.com/skooch/ddsm-mammography, accessed on 5 January 2021.  ... 
doi:10.3390/cancers13215384 pmid:34771547 pmcid:PMC8582536 fatcat:pop73fnxuzhi3g3kliepmyzrzy

Unsupervised Representation Learning by Predicting Random Distances

Hu Wang, Guansong Pang, Chunhua Shen, Congbo Ma
2020 Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  
Empirical results on 19 real-world datasets show that our learned representations substantially outperform a few state-of-the-art methods for both anomaly detection and clustering tasks.  ...  However, they often require large-scale labelled data to successfully learn such features, which significantly hinders their adaption in unsupervised learning tasks, such as anomaly detection and clustering  ...  Acknowledgments We thank Mitsuru Kusumoto and Kohei Hayashi for their helpful comments. Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  ... 
doi:10.24963/ijcai.2020/404 dblp:conf/ijcai/ZhaoVAL20 fatcat:dff4nxcmajhgpelzji6hfngedu

Anomalous Situation Detection in Complex Scenes [article]

Michalis Voutouris, Giovanni Sachi, Hina Afridi
2019 arXiv   pre-print
The experimental tests are conducted on a set of benchmark video sequences commonly used for anomaly situation detection.  ...  The MLP neural network is subsequently explored to consider these features that can detect the anomalous situation.  ...  This function computes the squared error between the desired and actual output vectors and the backpropagation is gradient descent on the cost function in Eq. (3) .  ... 
arXiv:1902.10016v1 fatcat:ag3o5gvq5fae5ag2pbppjf6cjq

Real-Time Predictive Maintenance using Autoencoder Reconstruction and Anomaly Detection [article]

Sean Givnan, Carl Chalmers, Paul Fergus, Sandra Ortega, Tom Whalley
2021 arXiv   pre-print
Real-time monitoring offers a solution for detecting faults without the need for manual observation.  ...  However, manual interpretation for threshold anomaly detection is often subjective and varies between industrial experts. This approach is ridged and prone to a large number of false positives.  ...  Figure 5 : 5 Overall Framework design for anomaly detection on the edge for data transmission.  ... 
arXiv:2110.01447v1 fatcat:trmermipdbggdnkhwhw6qcwshu

Research on Healthy Anomaly Detection Model Based on Deep Learning from Multiple Time-Series Physiological Signals

Kai Wang, Youjin Zhao, Qingyu Xiong, Min Fan, Guotan Sun, Longkun Ma, Tong Liu
2016 Scientific Programming  
detection method to detect anomaly data.  ...  Our experiment is shown to have a significant performance in physiological signals anomaly detection.  ...  In this paper, we propose a lightweight approach for detecting the anomaly data by analyzing the physiological signals.  ... 
doi:10.1155/2016/5642856 fatcat:4f5lxxusi5aelmlaubrvzyfqve

Gradient Techniques To Predict Distributed Denial-Of-Service Attack

Roheen Qamar
2022 Iraqi Journal for Computer Science and Mathematics  
This network was developed using five distinct training algorithms: 1) Fletcher–Powell conjugate gradient, 2) Polak–Ribiére conjugate gradient of, 3) resilient backpropagation, 4) gradient conjugation  ...  The artificial neural network toolset in MATLAB was used to investigate the detection of DDoS attacks.  ...  The authors introduced an anomaly intrusion detection system that scans and detects various network attacks.  ... 
doi:10.52866/ijcsm.2022.02.01.006 fatcat:f7vssi6zorhzlil6hdfwis76hq
« Previous Showing results 1 — 15 out of 2,465 results