Semi-Supervised Learning in Machine Learning

Semi-Supervised Learning:

Learning in order to understand the nature of semi-supervised learning, it’ll be useful first to take a look at supervised learning. Traditionally, there have been two fundametally different types of tasks in machine learning.

  • Supervised
  • Unsupervised

But semi-supervised learning is an idea in between these two processes.

Semi-supervised learning is halfway between supervised and unsupervised learning. In addition to unlabeled data, the algorithm is provided with some super vision information – but not necessarily for all examples.

1. Continuity Assumption: Point which are close to each other are more likely to share a label.

2. Semi-supervised Smoothness Assumption:If two points x1, x2, in a high-density region-are close, then so should be the corresponding output y1, y2.

3. Cluster Assumption: If points are in the same cluster, they are likely to be of the same class.

4. Manifold Assumption: The high-dimensional data lie on a low-dimensional manifold.

5. Transduction: Some algorithm naturally operate in a transductive setting. According to the philosophy put forward by Vapnik, high-dimensional estimation problems should attempt to follow the following principle:

Vapnik Principle: When you trying solve some problem, one should not solve a more difficult problem as an intermediate step.