Online Coregularization for Multiview Semisupervised Learning

Table 4

Mean test error rates on the rotating two-moons-two-lines synthetic data sequence. The error rates are reported for three different sparse approximations. For gradient ascent, we choose a stationary step size 𝜌_𝑡=0.1. The result shows that our derived online co-regularization algorithms are able to track the changes in the sequence and maintain a smaller error rate compared with batch learning algorithms. Specially, ADA (Buffer- 𝑁) performs better than the other online co-regularization algorithms.

Online co-regularization algorithms	Gradient ascent	ADA 𝑆𝑡={𝑡}	ADA Buffer-𝑁	ADA Buffer-𝐿
Error rates (%)
No sparse approximation	12.75	7.75	6.43	20.35
Absolute threshold (𝜀=0.05)	11.55	10.75	9.27	18.75
𝑘-MC (𝑘=400)	11.55	10.35	9.45	18.35