scholarly journals Robustness of Autoencoders for Anomaly Detection Under Adversarial Impact

2020 ◽  
Author(s):  
Adam Goodge ◽  
Bryan Hooi ◽  
See Kiong Ng ◽  
Wee Siong Ng
Keyword(s):  
Deep Learning ◽  
Anomaly Detection ◽  
Defense Mechanisms ◽  
Model Performance ◽  
Real Data ◽  
Learning Methods ◽  
Small Perturbations ◽  
Unsupervised Anomaly Detection ◽  
Latent Representations ◽  
Supervised Image Classification

Detecting anomalies is an important task in a wide variety of applications and domains. Deep learning methods have achieved state-of-the-art performance in anomaly detection in recent years; unsupervised methods being particularly popular. However, deep learning methods can be fragile to small perturbations in the input data. This can be exploited by an adversary to deliberately hinder model performance; an adversarial attack. This phenomena has been widely studied in the context of supervised image classification since its discovery, however such studies for an anomaly detection setting are sorely lacking. Moreover, the plethora of defense mechanisms that have been proposed are often not applicable to unsupervised anomaly detection models. In this work, we study the effect of adversarial attacks on the performance of anomaly-detecting autoencoders using real data from a Cyber physical system (CPS) testbed with intervals of controlled, physical attacks as anomalies. An adversary would attempt to disguise these points as normal through adversarial perturbations. To combat this, we propose the Approximate Projection Autoencoder (APAE), which incorporates two defenses against such attacks into a general autoencoder. One of these involves a novel technique to improve robustness under adversarial impact by optimising latent representations for better reconstruction outputs.

scholarly journals FuseAD: Unsupervised Anomaly Detection in Streaming Sensors Data by Fusing Statistical and Deep Learning Models

Sensors ◽  
2019 ◽  
Vol 19 (11) ◽  
pp. 2451 ◽  
Author(s):  
Mohsin Munir ◽  
Shoaib Ahmed Siddiqui ◽  
Muhammad Ali Chattha ◽  
Andreas Dengel ◽  
Sheraz Ahmed
Keyword(s):  
Deep Learning ◽  
Anomaly Detection ◽  
Internal State ◽  
Arima Model ◽  
Streaming Data ◽  
Detection Technique ◽  
Detection Methods ◽  
Smart Devices ◽  
Detection Techniques ◽  
Unsupervised Anomaly Detection

The need for robust unsupervised anomaly detection in streaming data is increasing rapidly in the current era of smart devices, where enormous data are gathered from numerous sensors. These sensors record the internal state of a machine, the external environment, and the interaction of machines with other machines and humans. It is of prime importance to leverage this information in order to minimize downtime of machines, or even avoid downtime completely by constant monitoring. Since each device generates a different type of streaming data, it is normally the case that a specific kind of anomaly detection technique performs better than the others depending on the data type. For some types of data and use-cases, statistical anomaly detection techniques work better, whereas for others, deep learning-based techniques are preferred. In this paper, we present a novel anomaly detection technique, FuseAD, which takes advantage of both statistical and deep-learning-based approaches by fusing them together in a residual fashion. The obtained results show an increase in area under the curve (AUC) as compared to state-of-the-art anomaly detection methods when FuseAD is tested on a publicly available dataset (Yahoo Webscope benchmark). The obtained results advocate that this fusion-based technique can obtain the best of both worlds by combining their strengths and complementing their weaknesses. We also perform an ablation study to quantify the contribution of the individual components in FuseAD, i.e., the statistical ARIMA model as well as the deep-learning-based convolutional neural network (CNN) model.

scholarly journals A Hybrid Deep Learning-Based Unsupervised Anomaly Detection in High Dimensional Data

2022 ◽  
Vol 70 (3) ◽  
pp. 5363-5381
Author(s):  
Amgad Muneer ◽  
Shakirah Mohd Taib ◽  
Suliman Mohamed Fati ◽  
Abdullateef O. Balogun ◽  
Izzatdin Abdul Aziz
Keyword(s):  
Deep Learning ◽  
Anomaly Detection ◽  
High Dimensional Data ◽  
High Dimensional ◽  
Unsupervised Anomaly Detection

scholarly journals DeepAnT: A Deep Learning Approach for Unsupervised Anomaly Detection in Time Series

IEEE Access ◽  
2019 ◽  
Vol 7 ◽  
pp. 1991-2005 ◽  
Author(s):  
Mohsin Munir ◽  
Shoaib Ahmed Siddiqui ◽  
Andreas Dengel ◽  
Sheraz Ahmed
Keyword(s):  
Time Series ◽  
Deep Learning ◽  
Anomaly Detection ◽  
Learning Approach ◽  
Unsupervised Anomaly Detection

scholarly journals HMD-AMP: Protein Language-Powered Hierarchical Multi-label Deep Forest for Annotating Antimicrobial Peptides

2021 ◽  
Author(s):  
Qinze Yu ◽  
Zhihang Dong ◽  
Xingyu Fan ◽  
Licheng Zong ◽  
Yu Li
Keyword(s):  
Deep Learning ◽  
Antimicrobial Peptides ◽  
Binary Classification ◽  
Classification Task ◽  
Protein Sequence Database ◽  
Learning Methods ◽  
Small Perturbations ◽  
Label Protein ◽  
Deep Forest ◽  
Wet Lab

Identifying the targets of an antimicrobial peptide is a fundamental step in studying the innate immuneresponse and combating antibiotic resistance, and more broadly, precision medicine and public health. Therehave been extensive studies on the statistical and computational approaches to identify (i) whether a peptide is anantimicrobial peptide (AMP) or a non-AMP and (ii) which targets are these sequences effective to (Gram-positive,Gram-negative, etc.). Despite the existing deep learning methods on this problem, most of them are unable tohandle the small AMP classes (anti-insect, anti-parasite, etc.). And more importantly, some AMPs can havemultiple targets, which the previous methods fail to consider. In this study, we build a diverse and comprehensivemulti-label protein sequence database by collecting and cleaning amino acids from various AMP databases.To generate efficient representations and features for the small classes dataset, we take advantage of a proteinlanguage model trained on 250 million protein sequences. Based on that, we develop an end-to-end hierarchicalmulti-label deep forest framework, HMD-AMP, to annotate AMP comprehensively. After identifying an AMP, itfurther predicts what targets the AMP can effectively kill from eleven available classes. Extensive experimentssuggest that our framework outperforms state-of-the-art models in both the binary classification task and themulti-label classification task, especially on the minor classes. Compared with the previous deep learning methods,our method improves the performance on macro-AUROC by 11%. The model is robust against reduced featuresand small perturbations and produces promising results. We believe HMD-AMP contribute to both the future wet-lab investigations of the innate structural properties of different antimicrobial peptides and build promising empirical underpinnings for precise medicine with antibiotics.

scholarly journals Detection of Lung Nodules in Micro-CT Imaging Using Deep Learning

Tomography ◽  
2021 ◽  
Vol 7 (3) ◽  
pp. 358-372
Author(s):  
Matthew D. Holbrook ◽  
Darin P. Clark ◽  
Rutulkumar Patel ◽  
Yi Qi ◽  
Alex M. Bassil ◽  
...  
Keyword(s):  
Deep Learning ◽  
Lung Nodule ◽  
Model Performance ◽  
Area Under The Curve ◽  
Simulated Data ◽  
Real Data ◽  
Lung Tumors ◽  
Training Data ◽  
Micro Ct ◽  
Micro Computed Tomography

We are developing imaging methods for a co-clinical trial investigating synergy between immunotherapy and radiotherapy. We perform longitudinal micro-computed tomography (micro-CT) of mice to detect lung metastasis after treatment. This work explores deep learning (DL) as a fast approach for automated lung nodule detection. We used data from control mice both with and without primary lung tumors. To augment the number of training sets, we have simulated data using real augmented tumors inserted into micro-CT scans. We employed a convolutional neural network (CNN), trained with four competing types of training data: (1) simulated only, (2) real only, (3) simulated and real, and (4) pretraining on simulated followed with real data. We evaluated our model performance using precision and recall curves, as well as receiver operating curves (ROC) and their area under the curve (AUC). The AUC appears to be almost identical (0.76–0.77) for all four cases. However, the combination of real and synthetic data was shown to improve precision by 8%. Smaller tumors have lower rates of detection than larger ones, with networks trained on real data showing better performance. Our work suggests that DL is a promising approach for fast and relatively accurate detection of lung tumors in mice.

scholarly journals RCA: A Deep Collaborative Autoencoder Approach for Anomaly Detection

2021 ◽  
Author(s):  
Boyang Liu ◽  
Ding Wang ◽  
Kaixiang Lin ◽  
Pang-Ning Tan ◽  
Jiayu Zhou
Keyword(s):  
Neural Networks ◽  
Deep Learning ◽  
Anomaly Detection ◽  
Missing Values ◽  
Deep Neural Networks ◽  
Reconstruction Error ◽  
Feature Representation ◽  
High Complexity ◽  
Unsupervised Anomaly Detection ◽  
Robust Framework

Unsupervised anomaly detection plays a crucial role in many critical applications. Driven by the success of deep learning, recent years have witnessed growing interests in applying deep neural networks (DNNs) to anomaly detection problems. A common approach is using autoencoders to learn a feature representation for the normal observations in the data. The reconstruction error of the autoencoder is then used as outlier scores to detect the anomalies. However, due to the high complexity brought upon by the over-parameterization of DNNs, the reconstruction error of the anomalies could also be small, which hampers the effectiveness of these methods. To alleviate this problem, we propose a robust framework using collaborative autoencoders to jointly identify normal observations from the data while learning its feature representation. We investigate the theoretical properties of the framework and empirically show its outstanding performance as compared to other DNN-based methods. Our experimental results also show the resiliency of the framework to missing values compared to other baseline methods.

Sign in / Sign up

Export Citation Format

Share Document