Generative OpenMax for Multi-Class Open Set Classification

Zongyuan Ge, Sergey Demyanov and Rahil Garnavi

Abstract

We present a conceptually new and flexible method for multi-class open set classification. Unlike previous methods where unknown classes are inferred with respect to the feature or decision distance to the known classes, our approach is able to provide explicit modelling and decision score for unknown classes. The proposed method, called Generative OpenMax (G-OpenMax), extends OpenMax by employing generative adversarial networks (GANs) for novel category image synthesis. We validate the proposed method on two datasets of handwritten digits and characters, resulting in superior results over previous deep learning based method OpenMax Moreover, G-OpenMax provides a way to visualize samples representing the unknown classes from open space.

Session

Posters

Files

PDF iconPaper (PDF)
PDF iconSupplementary (PDF)

DOI

10.5244/C.31.42
https://dx.doi.org/10.5244/C.31.42

Citation

Zongyuan Ge, Sergey Demyanov and Rahil Garnavi. Generative OpenMax for Multi-Class Open Set Classification. In T.K. Kim, S. Zafeiriou, G. Brostow and K. Mikolajczyk, editors, Proceedings of the British Machine Vision Conference (BMVC), pages 42.1-42.12. BMVA Press, September 2017.

Bibtex

            @inproceedings{BMVC2017_42,
                title={Generative OpenMax for Multi-Class Open Set Classification},
                author={Zongyuan Ge, Sergey Demyanov and Rahil Garnavi},
                year={2017},
                month={September},
                pages={42.1-42.12},
                articleno={42},
                numpages={12},
                booktitle={Proceedings of the British Machine Vision Conference (BMVC)},
                publisher={BMVA Press},
                editor={Tae-Kyun Kim, Stefanos Zafeiriou, Gabriel Brostow and Krystian Mikolajczyk},
                doi={10.5244/C.31.42},
                isbn={1-901725-60-X},
                url={https://dx.doi.org/10.5244/C.31.42}
            }