Semi-supervised learning is a paradigm that uses a large number of unlabeled data and a small number of labeled data. We analyze the dynamical behaviors of semi-supervised learning in the framework of on-line learning by the statistical-mechanical method. A student uses several correlated input vectors in each update. The student is given a desired output for only one input vector out of these correlated input vectors. In this model, we derive simultaneous differential equations with deterministic forms that describe the dynamical behaviors of order parameters using the self-averaging property in the thermodynamic limit. We treat three well-known learning rules, that is, the Hebbian, Perceptron, and AdaTron learning rules. As a result, it is shown that using unlabeled data is effective in the early stages for all three learning rules. In addition, we show that the three learning rules have qualitatively different dynamical behaviors. Furthermore, we propose a new algorithm that improves the generalization...
[1]
Koji Hukushima,et al.
On-line Learning of an Unlearnable True Teacher through Mobile Ensemble Teachers
,
2008,
ArXiv.
[2]
Alexander Zien,et al.
Semi-Supervised Learning
,
2006
.
[3]
Masato Okada,et al.
Statistical Mechanics of On-line Learning when a Moving Teacher Goes around an Unlearnable True Teacher
,
2006,
ArXiv.
[4]
H. Nishimori.
Statistical Physics of Spin Glasses and Information Processing
,
2001
.
[5]
Seiji Miyoshi,et al.
Statistical Mechanics of On-Line Learning Using Correlated Examples
,
2011,
IEICE Trans. Inf. Syst..