Anomaly Detection Using Generative Adversarial Networks
Thesis title in Czech: | Detekce anomálií pomocí generativních adversariálních sítí |
---|---|
Thesis title in English: | Anomaly Detection Using Generative Adversarial Networks |
Key words: | Anomaly detection, generative adversarial networks, neural network, deep learning |
English key words: | Anomaly detection, generative adversarial networks, neural network, deep learning |
Academic year of topic announcement: | 2018/2019 |
Thesis type: | Bachelor's thesis |
Thesis language: | angličtina |
Department: | Department of Theoretical Computer Science and Mathematical Logic (32-KTIML) |
Supervisor: | RNDr. Jiří Fink, Ph.D. |
Author: | hidden - assigned and confirmed by the Study Dept. |
Date of registration: | 26.10.2018 |
Date of assignment: | 26.10.2018 |
Confirmed by Study dept. on: | 31.10.2018 |
Date and time of defence: | 27.06.2019 09:00 |
Date of electronic submission: | 09.05.2019 |
Date of submission of printed version: | 17.05.2019 |
Date of proceeded defence: | 27.06.2019 |
Opponents: | Mgr. Martin Pilát, Ph.D. |
Guidelines |
Anomaly detection is an important topic in the automotive industry, medical research, computer security and many other areas. Its goal is to identify novelties and anomalies.
Generative adversarial networks (GANs) [1] are unsupervised learning algorithms. They have been used for generating realistic images based on real images in which they achieved state of the art results. An important feature is the ability to learn the distribution of the supplied images. This thesis will exploit this feature for learning the distribution of normal instances and then, during inference, measure how much given instance (possibly anomalous) differs from the learned distribution. The main objective of this work is to study GAN based anomaly detection methods presented in articles [2] and [3] and with the use of Tensorflow framework [4] set suitable parameters for neural network to achieve good results on credit card fraud dataset used in [5] and compare studied methods. Student may try to modify GAN to obtain better results. |
References |
[1] I. J. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y. Bengio, Generative adversarial networks, 2014. arXiv:1406.2661v1
[2] T. Schlegl, P. Seeb ̈ock, S. M. Waldstein, U. Schmidt-Erfurth, and G. Langs, Unsupervised anomaly detection with generative adversarial networks to guide marker discovery, 2017. arXiv:1703.05921v1 [3] H. Zenati, C. S. Foo, B. Lecouat, G. Manek, and V. R. Chandrasekhar, Efficient GAN-based anomaly detection, 2018. arXiv:1802.06222v1 [4] M. Abadi, A. Agarwal, P. Barham, E. Brevdo, Z. Chen, C. Citro, G. S. Corrado, A. Davis, J. Dean, M. Devin, S. Ghemawat, I. Goodfellow, A. Harp, G. Irving, M. Isard, Y. Jia, R. Jozefowicz, L. Kaiser, M. Kudlur, J. Levenberg, D. Man ́e, R. Monga, S. Moore, D. Murray, C. Olah, M. Schuster, J. Shlens, B. Steiner, I. Sutskever, K. Talwar, P. Tucker, V. Vanhoucke, V. Vasudevan, F. Vi ́egas, O. Vinyals, P. Warden, M. Wattenberg, M. Wicke, Y. Yu, and X. Zheng, TensorFlow: Large-Scale Machine Learning on Heterogeneous Systems, Software available from https://www.tensorflow.org, 2015. https://www.tensorflow.org/ [5] A. D. Pozzolo, O. Caelen, R. A. Johnson, and G. Bontempi, “Calibrating Probability with Undersampling for Unbalanced Classification,” in 2015 IEEE Symposium Series on Computational Intelligence, IEEE, 2015 [6] K. G. Mehrotra, C. K. Mohan, and H. Huang, Anomaly Detection Principles and Algorithms. Springer, 2017, isbn:978-3-319-67526-8 |