LSTM-based Encoder-Decoder for Multi-sensor Anomaly Detection

Malhotra, Pankaj; Ramakrishnan, A. G.; Anand, Gaurangi; Vig, Lovekesh; Agarwal, Puneet; Shroff, Gautam

doi:10.48550/arxiv.1607.00148

Cited by 121 publications

(194 citation statements)

References 0 publications

Supporting

Mentioning

193

Contrasting

Order By: Relevance

“…• LSTM-AE [8], which is a reconstruction-based method using a single-layer LSTM on both encoder and decoder.…”

Section: Comparison Methodsmentioning

confidence: 99%

“…As the model is trained exclusively on normal data, whenever it is not able to reconstruct a given input with equal quality compared to the reconstruction of normal data, the instance is treated as an anomaly. The LSTM Encoder-Decoder model [8] is proposed to learn temporal representations of the time series via the LSTM networks and use reconstruction errors to detect anomalies. Despite its effectiveness, LSTM does not take spatial correlation into consideration.…”

Section: Related Workmentioning

confidence: 99%

“…AE is usually trained by minimizing the reconstruction error on the normal data, and then using such errors as indicators or thresholds of anomalies. Based on the AE structure, LSTM-AE [8], Convolutional AE (CAE) [6], and ConvLSTM-AE [9] have emerged as promising anomaly detection approaches.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Adaptive Memory Networks with Self-supervised Learning for Unsupervised Anomaly Detection

Zhang¹,

Wang²,

Chen³

et al. 2022

Preprint

View full text Add to dashboard Cite

Unsupervised anomaly detection aims to build models to effectively detect unseen anomalies by only training on the normal data. Although previous reconstruction-based methods have made fruitful progress, their generalization ability is limited due to two critical challenges. First, the training dataset only contains normal patterns, which limits the model generalization ability. Second, the feature representations learned by existing models often lack representativeness which hampers the ability to preserve the diversity of normal patterns. In this paper, we propose a novel approach called Adaptive Memory Network with Self-supervised Learning (AMSL) to address these challenges and enhance the generalization ability in unsupervised anomaly detection. Based on the convolutional autoencoder structure, AMSL incorporates a self-supervised learning module to learn general normal patterns and an adaptive memory fusion module to learn rich feature representations. Experiments on four public multivariate time series datasets demonstrate that AMSL significantly improves the performance compared to other state-of-the-art methods. Specifically, on the largest CAP sleep stage detection dataset with 900 million samples, AMSL outperforms the second-best baseline by 4%+ in both accuracy and F1 score. Apart from the enhanced generalization ability, AMSL is also more robust against input noise.

show abstract

“…• LSTM-AE [8], which is a reconstruction-based method using a single-layer LSTM on both encoder and decoder.…”

Section: Comparison Methodsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Adaptive Memory Networks with Self-supervised Learning for Unsupervised Anomaly Detection

Zhang¹,

Wang²,

Chen³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…LSTM models are a type of recurrent neural networks which possess the ability to retain short term memory. They are capable of learning complex and inter-related input features and predict output sequences [36]. LSTMs are useful in processing long sequences as they are able to comprehend and retain the overall global distribution as well as map that to local changes.…”

Section: B Long Short-term Memorymentioning

confidence: 99%

RoboMal: Malware Detection for Robot Network Systems

Kaur¹,

Zhou²,

Shen³

et al. 2022

Preprint

View full text Add to dashboard Cite

Robot systems are increasingly integrating into numerous avenues of modern life. From cleaning houses to providing guidance and emotional support, robots now work directly with humans. Due to their far-reaching applications and progressively complex architecture, they are being targeted by adversarial attacks such as sensor-actuator attacks, data spoofing, malware, and network intrusion. Therefore, security for robotic systems has become crucial. In this paper, we address the underserved area of malware detection in robotic software. Since robots work in close proximity to humans, often with direct interactions, malware could have life-threatening impacts. Hence, we propose the RoboMal framework of static malware detection on binary executables to detect malware before it gets a chance to execute. Additionally, we address the great paucity of data in this space by providing the RoboMal dataset 1 comprising controller executables of a small-scale autonomous car. The performance of the framework is compared against widely used supervised learning models: GRU, CNN, and ANN. Notably, the LSTMbased RoboMal model outperforms the other models with an accuracy of 85% and precision of 87% in 10-fold cross-validation, hence proving the effectiveness of the proposed framework.

show abstract

“…The selection of similarity measurement is very important for clustering-based methods. The second category of approaches is reconstruction-based methods ( [19,20,21,11]). These methods usually construct an auto-encoder network, which model the distribution of the entire time series and reconstruct the original input based on latent representations.…”

Section: Attack Detection For Cpssmentioning

confidence: 99%

TFDPM: Attack detection for cyber-physical systems with diffusion probabilistic models

Yan¹,

Zhou²,

Zhan³

et al. 2021

Preprint

View full text Add to dashboard Cite

With the development of AIoT, data-driven attack detection methods for cyber-physical systems (CPSs) have attracted lots of attention. However, existing methods usually adopt tractable distributions to approximate data distributions, which are not suitable for complex systems. Besides, the correlation of the data in different channels does not attract sufficient attention. To address these issues, we use energy-based generative models, which are less restrictive on functional forms of the data distribution. In addition, graph neural networks are used to explicitly model the correlation of the data in different channels. In the end, we propose TFDPM, a general framework for attack detection tasks in CPSs. It simultaneously extracts temporal pattern and feature pattern given the historical data. Then extract features are sent to a conditional diffusion probabilistic model. Predicted values can be obtained with the conditional generative network and attacks are detected based on the difference between predicted values and observed values. In addition, to realize real-time detection, a conditional noise scheduling network is proposed to accelerate the prediction process. Experimental results show that TFDPM outperforms existing state-of-the-art attack detection methods. The noise scheduling network increases the detection speed by three times.

show abstract

LSTM-based Encoder-Decoder for Multi-sensor Anomaly Detection

Cited by 121 publications

References 0 publications

Adaptive Memory Networks with Self-supervised Learning for Unsupervised Anomaly Detection

Adaptive Memory Networks with Self-supervised Learning for Unsupervised Anomaly Detection

RoboMal: Malware Detection for Robot Network Systems

TFDPM: Attack detection for cyber-physical systems with diffusion probabilistic models

Contact Info

Product

Resources

About