Efficient GAN-Based Anomaly Detection

By Houssam Zenati, Chuan-Sheng Foo, Bruno Lecouat, Gaurav Manek, Vijay Ramaseshan Chandrasekhar. arXiv.org, 2018

607 citations recorded by the source corpus.

Generative adversarial networks (GANs) are able to model the complex highdimensional distributions of real-world data, which suggests they could be effective for anomaly detection. However, few works have explored the use of GANs for the anomaly detection task. We leverage recently developed GAN models for anomaly detection, and achieve state-of-the-art performance on image and network intrusion datasets, while being several hundred-fold faster at test time than the only published GAN-based method.

Subfields: Artificial Intelligence, Computer Networks and Communications, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/b3acb6f183b5f4b651f53c0eec5cb5c805224ac1

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 Computer Networks and Communications

In Computer Vision and Pattern Recognition

Citations in the corpus

Improving Unsupervised Defect Segmentation by Applying Structural Similarity to Autoencoders (2018, methodology): “[13] proposed to use bi-directional GANs [14] to add the missing encoder network for faster inference.”
OpenGAN: Open-Set Recognition via Open Data Generation (2021, neutral): “There are many other GAN-based open-set methods, such as training BiGANs [47, 55, 56] or Table 2: Our technical insights apply to other GAN-based openset discrimination methods: 1) using BiGAN-discriminator as the open likelihood function works better than using reconstruction errors (BiGAN d vs.”
Skip-GANomaly: Skip Connected and Adversarially Trained Encoder-Decoder Anomaly Detection (2019, methodology): “to perform inference within the latent space is one of the novel contributions of the proposed approach compared to the previous approaches [9], [12], [13].”
A Survey on Anomaly Detection for Technical Systems using LSTM Networks (2021, methodology): “Different architectures of deep neural networks, and in particular, architectures based on Long Short-Term Memory (LSTM), have been designed proving to be capable of solving a variety of complex detection tasks, as is described in [1] or [2].”
A Unifying Review of Deep and Shallow Anomaly Detection (2020, background): “to find the latent point z̃ and is used in a variety of ways, usually incorporating the reconstruction error [57], [148], [151], [152].”
Maximum Entropy Generators for Energy-Based Models (2019, methodology): “From Table 4, it can be seen that our energy model outperforms VAEs for outlier detection and is comparable to the SOTA BiGAN-based anomaly detection methods for this dataset (Zenati et al., 2018) which train bidirectional GANs to learn both an encoder and decoder (generator) simultaneously and use ”
Adversarial Machine Learning for Network Intrusion Detection Systems: A Comprehensive Survey (2023, background): “Hashemi et al. [107] assume a WB threat model and attack three NIDS: Kitsune [47], DAGMM [119], and BiGAN [120] on CIC-IDS2018.”
Joint Training of Variational Auto-Encoder and Latent Energy-Based Model (2020, methodology): “We use energy function as our decision function and compare with the BiGAN-based anomaly detection model [40], the recent MEG [24] and the VAE model using area under the precision-recall curve (AUPRC) as in [40].”
Generalized Out-of-Distribution Detection: A Survey (2021, positive): “Existing works leverage GANs to generate OOD training samples and force the model predictions to be uniform (Lee et al., 2018a), generate boundary samples in the low-density region (Vernekar et al., 2019), or similarly, high-conﬁdence OOD samples (Sricharan et al., 2018), or using meta-learning the ”
Unsupervised Brain Abnormality Detection Using High Fidelity Image Reconstruction Networks (2020, background): “This holds even for high-quality image generation in comparison to GAN-based architectures such as BiGAN , which incorporates an inverse mapping function from the sample space to the latent space (Donahue et al., 2017; Zenati et al., 2018).”
Bayesian Variational Autoencoders for Unsupervised Out-of-Distribution Detection (2019, methodology): “We use two benchmarks: (a) FashionMNIST (in-distribution) vs. MNIST (OoD) (Hendrycks et al., 2018; Nalisnick et al., 2018; Ren et al., 2019; Zenati et al., 2018; Akcay et al., 2018), and (b) eight classes of FashionMNIST (in-distribution) vs. the remaining two classes (OoD), using ﬁve different…”
MimicGAN: Robust Projection onto Image Manifolds with Corruption Mimicking (2019, methodology): “We also compare our method with recent manifold-projection based anomaly detection techniques [7, 6, 58].”
A Deep Reinforcement Learning Perspective on Internet Congestion Control (2019, methodology): “This problem is an instance of novelty detection (Schölkopf et al., 2000), and modern approaches employing deep networks could in principle be employed here (Zenati et al., 2018; Mandelbaum & Weinshall, 2017).”
Learning Latent Space Energy-Based Prior Model (2020, methodology): “Following the protocol as in [31, 62], we make each digit class an anomaly and consider the remaining 9 digits as normal examples.”
Deep Learning for Anomaly Detection (2020, methodology): “, an inverse of the generator, resulting in methods like EBGAN [170] and”
Detecting semantic anomalies (2019, background): “Recommended benchmarks A very small number of recent works (Akcay, Atapour-Abarghouei, and Breckon; Zenati et al., 2018; 2018) have considered a case that is more aligned with the goals stated above.”