TY - GEN
T1 - SAN
T2 - 27th International Conference on Neural Information Processing, ICONIP 2020
AU - Tang, Chenwei
AU - Kuang, Yangzhu
AU - Lv, Jiancheng
AU - Hu, Jinglu
N1 - Funding Information:
Thank Yangzhu Kuang for his contribution to this article. This paper is supported by the National Natural Science Fund for Distinguished Young Scholar under Grant No. 61625204 and the Key Program of National Science Foundation of China under Grant No. 61836006.
Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - In this paper, we propose a Sampling Adversarial Networks (SAN) framework to improve Zero-Shot Learning (ZSL) by mitigating the hubness and semantic gap problem. The SAN framework incorporates a sampling model and a discriminating model, and corresponds them to the minimax two-player game. Specifically, given the semantic embedding, the sampling model samples the visual features from the training set to approach the discriminator’s decision boundary. Then, the discriminator distinguishes the matching visual-semantic pairs from the sampled data. On the one hand, by the measurement of the matching degree of visual-semantic pairs and the adversarial training way, the visual-semantic embedding built by the proposed SAN decreases the intra-class distance and increases the inter-class separation. Then, the reduction of universal neighbours in the visual-semantic embedding subspace alleviates the hubness problem. On the other, the sampled rather than directly generated visual features maintain the same manifold as the real data, mitigating the semantic gap problem. Experiments show that the sampler and discriminator of the SAN framework outperform state-of-the-art methods both in conventional and generalized ZSL settings.
AB - In this paper, we propose a Sampling Adversarial Networks (SAN) framework to improve Zero-Shot Learning (ZSL) by mitigating the hubness and semantic gap problem. The SAN framework incorporates a sampling model and a discriminating model, and corresponds them to the minimax two-player game. Specifically, given the semantic embedding, the sampling model samples the visual features from the training set to approach the discriminator’s decision boundary. Then, the discriminator distinguishes the matching visual-semantic pairs from the sampled data. On the one hand, by the measurement of the matching degree of visual-semantic pairs and the adversarial training way, the visual-semantic embedding built by the proposed SAN decreases the intra-class distance and increases the inter-class separation. Then, the reduction of universal neighbours in the visual-semantic embedding subspace alleviates the hubness problem. On the other, the sampled rather than directly generated visual features maintain the same manifold as the real data, mitigating the semantic gap problem. Experiments show that the sampler and discriminator of the SAN framework outperform state-of-the-art methods both in conventional and generalized ZSL settings.
KW - Hubness problem
KW - Sampling Adversarial Networks
KW - Semantic gap
KW - Zero-Shot Learning
UR - http://www.scopus.com/inward/record.url?scp=85097436896&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85097436896&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-63833-7_53
DO - 10.1007/978-3-030-63833-7_53
M3 - Conference contribution
AN - SCOPUS:85097436896
SN - 9783030638320
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 626
EP - 638
BT - Neural Information Processing - 27th International Conference, ICONIP 2020, Proceedings
A2 - Yang, Haiqin
A2 - Pasupa, Kitsuchart
A2 - Leung, Andrew Chi-Sing
A2 - Kwok, James T.
A2 - Chan, Jonathan H.
A2 - King, Irwin
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 18 November 2020 through 22 November 2020
ER -