Abstract: Multi-view datasets are frequently encountered in learning tasks. However, existing multi-view approaches mainly focus on the visible views and ignore the hidden information shared by the visible views. To address this problem, a multi-view fuzzy system (FS) which utilizes the hidden information shared by the multiple visible views is proposed in this paper. More specifically, using the ridge regression extreme learning fuzzy system (RR-EL-FS) as the base model, an approach with the cooperation of visible and hidden views (RR-EL-FS-CVH) is proposed for multi-view learning. This method can make better use of the effective information of the hidden space than the previous multi-view methods, and thus can further improve the generalization performance of the trained model. Extensive experiments are conducted to validate its effectiveness.

Key words: cooperation between visible and hidden views, multi-view learning, shared hidden space, fuzzy system;extreme learning

摘要： 多视角数据正在越来越多地应用于各种建模任务，但当前的多视角模糊系统建模方法，主要集中于实现各个显性视角的合作，还未能充分探讨和利用各视角间共享的隐信息。针对此，对如何引入各个显性视角共享的隐空间信息来实现显隐视角协同的模糊系统建模进行了研究。具体地，基于岭回归极限学习模糊系统（ridge regression extreme learning fuzzy system，RR-EL-FS）模型，引入隐空间信息实现显隐视角协同学习来对RR-EL-FS进行学习，最终开发出具有显隐视角协同功能的岭回归极限学习模糊系统预测模型（ridge regression extreme learning fuzzy system with cooperation between visible and hidden views，RR-EL-FS-CVH）。该方法较之以往相关的多视角建模方法能更好地利用隐空间的有效信息，从而能够进一步提高受训模型的泛化性能。大量的实验结果亦验证了所提方法的有效性。

关键词: 显隐视角协同, 多视角学习, 共享隐空间, 模糊系统, 极限学习