Outlier Detection with Autoencoder Ensembles

By Jinghui Chen, Saket K. Sathe, Charu C. Aggarwal, Deepak S. Turaga. 2017

455 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Civil and Structural Engineering, Computer Networks and Communications

How researchers see this paper

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

In Artificial Intelligence

In Civil and Structural Engineering

In Computer Networks and Communications

Citations in the corpus

A practical tutorial on autoencoders for nonlinear feature fusion: Taxonomy, models, software and guidelines (2018, methodology): “The technique introduced in [128] aims to improve the detection of outliers.”
Deep Isolation Forest for Anomaly Detection (2022, methodology): “The literature [31], [32] combines a group of autoencoders and uses the median of reconstruction errors as anomaly scores.”
Statistical Analysis of Nearest Neighbor Methods for Anomaly Detection (2019, methodology): “Many empirical methods have been developed in the unsupervised setup, which can be roughly classified into four categories: density based methods such as the Robust KDE (RKDE) [4], Local Outlier Factor (LOF) [5], and mixture models (EGMM); distance based methods such as kNN [6] and Angle-based Outli”
Neural Networks and Deep Learning (2018, unknown): “The work in [64] uses multiple autoencoders with randomized connections, and then aggregates the outlier scores from these different components in order to create the score of a single data point.”
Deep Anomaly Detection with Deviation Networks (2019, methodology): “her, which poses significant challenges to widely-used optimization objectives that generally assume the data objects within each class are similar to each other. Existingdeepanomalydetection1 methods[2,7,19,20,22,29,30] address these two challenges by using unsupervised deep learning to model the n”
Progress in Outlier Detection Techniques: A Survey (2019, background): “In most of unsupervised deep OD models, autoencoders play a central role [210], [238].”
Deep Learning for Time Series Anomaly Detection: A Survey (2022, unknown): “Since the outputs of separate basic models can be combined to improve accuracy through ensembles [29], a diversity-driven ensemble based on CAEs is proposed along with a parameter-transfer-based training strategy instead of training each CAE model separately.”
Classification-Based Anomaly Detection for General Data (2020, methodology): “To compare against ensemble methods e.g. Chen et al. (2017), we implemented the Feature Bagging Autoencoder (FB-AE) with autoencoders as the base classiﬁer, feature bagging as the source of randomization, and average reconstruction error as the anomaly score.”
ECOD: Unsupervised Outlier Detection Using Empirical Cumulative Distribution Functions (2022, background): “Some representative examples from this line of work include the use of generative adversarial networks (GANs) [20], autoencoders [41] (including variational autoencoders), and reinforcement learning [42].”
Deep Learning for Anomaly Detection (2020, background): “ive DA DP PT Archit. Activation # layers Loss Data OADA [66] (4) Semi Reconstruction Yes No No AE ReLU 3 MSE Video Replicator [58] (5.1.1) Unsup. Reconstruction No No No AE Tanh 2 MSE Tabular RandNet [29] (5.1.1) Unsup. Reconstruction No Yes Yes AE ReLU 3 MSE Tabular RDA [176] (5.1.1) Semi Reconstru”
Zero-Shot Out-of-Distribution Detection Based on the Pre-trained Model CLIP (2021, methodology): “Other generative methods also exist (Andrews, Morton, and Grifﬁn 2016; Chen et al. 2017), which detect OOD samples based on the reconstruction error of the trained generative model for un-seen samples.”
Deep Semi-Supervised Anomaly Detection (2019, background): “, 1986; Hinton & Salakhutdinov, 2006), which are the predominant approach to deep AD (Hawkins et al., 2002; Sakurada & Yairi, 2014; Andrews et al., 2016; Erfani et al., 2016; Zhai et al., 2016; Chen et al., 2017; Chalapathy & Chawla, 2019), can be understood as implicitly maximizing the mutual infor”
Deep One-Class Classification (2018, methodology): “Current approaches to deep AD have shown promising results (Hawkins et al., 2002; Sakurada & Yairi, 2014; Xu et al., 2015; Erfani et al., 2016; Andrews et al., 2016; Chen et al., 2017), but none of these methods are trained by optimizing an AD based objective function and typically rely on reconstru”
A Unifying Review of Deep and Shallow Anomaly Detection (2020, background): “Finally, other AE variants that have been applied to AD include RNN-based AEs [194], [231], [397], [398], convolutional AEs [54], AE ensembles [126], [398], and variants that constrain the gradients [399] or actively control the latent code topology [400] of an AE.”
HRN: A Holistic Approach to One Class Learning (2020, methodology): “…56, 6], mixture models [3, 43], one-class classiﬁcation based methods [75, 78, 39], deep learning based representation learning using auto-encoders [10, 89, 69, 7, 94] and adversarial learning [74, 15, 64, 20], ensemble methods [53, 10], graphs and random walks [58, 31], transfer learning [45, 2],…”