Unsupervised learning

Unsupervised Learning
- Python Automation and Machine Learning for ICs -
- An Online Book -

Python Automation and Machine Learning for ICs http://www.globalsino.com/ICs/

Chapter/Index: Introduction | A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | P | Q | R | S | T | U | V | W | X | Y | Z | Appendix

=================================================================================

Unsupervised learning does not use labels, for instance, Convolutional Autoencoder (CAE) (see page4215). Clustering is the most common unsupervised learning algorithm. Clustering algorithms are widely applied for detecting defect patterns on a wafer (local clustering) as well as for searching for wafer maps with similar defect patterns (wafer bin map, WBM, clustering). Unsupervised learning is more appealing because it makes use of inexpensive unlabeled data. In fact, since 2012, this research direction has gone through a relatively quiet period, because unsupervised learning is less relevant when a lot of labeled data are available. However, there are a few recent research attempts to revive this area, for instance, using variational methods for probabilistic autoencoders [1].

The data input of unsupervised learning is {x⁽¹⁾, x⁽¹⁾, ... x^(m)} instead of (x, y), which is input data of supervised learning.

Comparison between machine learning algorithms (supervised Learning, unsupervised Learning and reinforcement Learning). Error = target output - actual output

Figure 4322. Comparison between machine learning algorithms (supervised Learning, unsupervised Learning and reinforcement Learning). Error = target output - actual output.

============================================

[1] D. P. Kingma and M. Welling. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114, 2013.

=================================================================================