2023-06-09 更新
Linking generative semi-supervised learning and generative open-set recognition
Authors:Emile Reyn Engelbrecht, Johan du Preez
This study investigates the relationship between semi-supervised learning (SSL) and open-set recognition (OSR) in the context of generative adversarial networks (GANs). Although no previous study has formally linked SSL and OSR, their respective methods share striking similarities. Specifically, SSL-GANs and OSR-GANs require their generators to produce samples in the complementary space. Subsequently, by regularising networks with generated samples, both SSL and OSR classifiers generalize the open space. To demonstrate the connection between SSL and OSR, we theoretically and experimentally compare state-of-the-art SSL-GAN methods with state-of-the-art OSR-GAN methods. Our results indicate that the SSL optimised margin-GANs, which have a stronger foundation in literature, set the new standard for the combined SSL-OSR task and achieves new state-of-other art results in certain general OSR experiments. However, the OSR optimised adversarial reciprocal point (ARP)-GANs still slightly out-performed margin-GANs at other OSR experiments. This result indicates unique insights for the combined optimisation task of SSL-OSR.
PDF
点此查看论文截图
Open Set Relation Extraction via Unknown-Aware Training
Authors:Jun Zhao, Xin Zhao, Wenyu Zhan, Qi Zhang, Tao Gui, Zhongyu Wei, Yunwen Chen, Xiang Gao, Xuanjing Huang
The existing supervised relation extraction methods have achieved impressive performance in a closed-set setting, where the relations during both training and testing remain the same. In a more realistic open-set setting, unknown relations may appear in the test set. Due to the lack of supervision signals from unknown relations, a well-performing closed-set relation extractor can still confidently misclassify them into known relations. In this paper, we propose an unknown-aware training method, regularizing the model by dynamically synthesizing negative instances. To facilitate a compact decision boundary, ``difficult’’ negative instances are necessary. Inspired by text adversarial attacks, we adaptively apply small but critical perturbations to original training instances and thus synthesizing negative instances that are more likely to be mistaken by the model as known relations. Experimental results show that this method achieves SOTA unknown relation detection without compromising the classification of known relations.
PDF Accepted by ACL2023
点此查看论文截图