Online Coregularization for Multiview Semisupervised Learning

Table 2

Mean test error rates on the two-moons-two-lines synthetic data set. The error rates are reported for three different sparse approximations. For gradient ascent, we choose a decaying step size 𝜌_𝑡=0.1/√𝑡. The result shows that our derived online co-regularization algorithms achieve test accuracy comparable to offline co-regularization (CoLapSVM). The online co-regularization algorithms based on aggressive dual ascending procedures perform better than those based on gradient ascent.

Online co-regularization; algorithms	Gradient ascent	ADA 𝑆𝑡={𝑡}	ADA 𝑆𝑡=[𝑡]	ADA Buffer-𝑁	ADA Buffer-𝐿
Error rates (%)
No sparse approximation	12.15	8.95	3.95	4.25	4.05
Absolute threshold (𝜀=0.05)	10.80	9.65	3.95	4.05	4.05
𝑘-MC (𝑘=400)	13.30	9.85	3.95	7.00	5.35