Graduate School and Research Center in Digital Sciences

Deep Gaussian process autoencoders for novelty detection

Domingues, Rémi; Michiardi, Pietro; Zouaoui, Jihane; Filippone, Maurizio

Machine Learning, June 2018, Special Issue of the ECML PKDD 2018 Journal Track, ISSN: 0885-6125

Novelty detection is one of the classic problems in Machine Learning that has applications across several domains. This paper proposes a novel autoencoder based on Deep Gaussian Processes for novelty detection tasks. The learning of the proposed model is made tractable and scalable through the use of random feature approximations and stochastic variational inference. The result is a  exible model that is easy to implement and train, and can be applied to general novelty detection tasks, including large-scale problems and data with mixed-type features. The experiments indicate that the proposed model achieves competitive results with state-of-the-art novelty detection methods.

Document Doi Bibtex

Title:Deep Gaussian process autoencoders for novelty detection
Keywords:Novelty detection, Deep Gaussian, Processes Autoencoder, Unsupervised learning, Stochastic variational inference
Type:Journal
Language:English
City:
Date:
Department:Data Science
Eurecom ref:5589
Copyright: © Springer. Personal use of this material is permitted. The definitive version of this paper was published in Machine Learning, June 2018, Special Issue of the ECML PKDD 2018 Journal Track, ISSN: 0885-6125 and is available at : http://doi.org/10.1007/s10994-018-5723-3
Bibtex: @article{EURECOM+5589, doi = {http://doi.org/10.1007/s10994-018-5723-3}, year = {2018}, month = {06}, title = {{D}eep {G}aussian process autoencoders for novelty detection}, author = {{D}omingues, {R}{\'e}mi and {M}ichiardi, {P}ietro and {Z}ouaoui, {J}ihane and {F}ilippone, {M}aurizio}, journal = {{M}achine {L}earning, {J}une 2018, {S}pecial {I}ssue of the {ECML} {PKDD} 2018 {J}ournal {T}rack, {ISSN}: 0885-6125 }, url = {http://www.eurecom.fr/publication/5589} }
See also: