Long Short Term Memory Networks for Anomaly Detection in Time Series

By Pankaj Malhotra, Lovekesh Vig, Gautam M. Shroff, Puneet Agarwal. The European Symposium on Artificial Neural Networks, 2015

1,566 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/8e54dd2b426b8d656a77c155818a87dd34c40754

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

LSTM-based Encoder-Decoder for Multi-sensor Anomaly Detection (2016, methodology): “Similar to (Malhotra et al., 2015), we divide the normal time-series into four sets of time-series: The error vectors for the points in the sequences in set v N 1 are used to estimate the parameters µ and Σ of a Normal distribution N ( µ , Σ ) using Maximum Likelihood Estimation.”
A Roadmap Toward the Resilient Internet of Things for Cyber-Physical Systems (2018, background): “For instance, the authors in [84] use recurrent neural networks to detect anomalies in real-time data.”
Detecting Cyber Attacks in Industrial Control Systems Using Convolutional Neural Networks (2018, background): “In the past few years research has been performed applying deep and recurrent neural networks in the task of ICS anomaly detection [32], [7], [19].”
Anomaly Detection in Cyber Physical Systems Using Recurrent Neural Networks (2017, methodology): “In [11], the authors used RNN for modelling several real-world time series data.”
On scientific understanding with artificial intelligence (2022, background): “, see reference [50]) to this type of problem will be an important future research direction.”
Artificial Intelligence and COVID-19: Deep Learning Approaches for Diagnosis and Treatment (2020, background): “sequences with longer-term patterns of unknown length [36].”
Anomaly Detection in Time Series: A Comprehensive Evaluation (2022, methodology): “SE M I-S U PE RV IS ED LSTM-AD [89] 14 % 50 % 3 %”
Unsupervised Anomaly Detection With LSTM Neural Networks (2017, methodology): “timation technique [10], [21]. Thus, the conventional LSTM based approaches require careful selection of several additional parameters, which signiﬁcantly degrades their performance in real life [1], [10]. Furthermore, both the OC-SVM (or SVDD) and LSTM based methods are able to process only ﬁxed le”
DÏoT: A Federated Self-learning Anomaly Detection System for IoT (2018, background): “Most applications leverage Long Short-Term Memory (LSTM) networks for detecting anomalies in time series [45], aircraft data [46] or system logs [47].”
Detecting Spacecraft Anomalies Using LSTMs and Nonparametric Dynamic Thresholding (2018, methodology): “prediction performance will more likely come from providing increasingly refined command-based features to the model. Given the rise in prediction-based anomaly detection methods and related research [30, 31], we place increased emphasis on post-prediction error evaluation methods that have received”
TON_IoT Telemetry Dataset: A New Generation Dataset of IoT and IIoT for Data-Driven Intrusion Detection Systems (2020, methodology): “Finally, Long Short-Term Memory (LSTM) [73] has been chosen as the deep learning method based on the reviewer’s suggestion; moreover, it achieves state-of-the-art performance in dealing with sensor data and learning long-term dependencies from observations [74].”
Anomaly Detection for IoT Time-Series Data: A Survey (2020, background): “Malhotra et al. [71] presented two approaches using stacked layers of recurrent sigmoid units (RSUs) and LSTMs to capture long term dynamics in a variety of univariate systems.”
Deep Anomaly Detection for Time-Series Data in Industrial IoT: A Communication-Efficient On-Device Federated Learning Approach (2020, methodology): “Furthermore, the supervised DAD method has poor performance for data with class imbalance (the total number of negative class data is much less than the total number of positive class data) [12].”
Multivariate Time-series Anomaly Detection via Graph Attention Network (2020, background): “The first category of approaches analyzes each individual time-series by applying univariate models [1], [6]–[8], while the second one models multiple time-series as a unified entity [2], [4], [9]–[13].”
Deep reinforcement learning for time series: playing idealized trading games (2018, background): “Malhotra et al. [6] used stacked LSTM networks for anomaly detection in time series.”