Introduction to Machine Learning

AI Combines Supervised Learning with Other Learning Methods

Oct 17, 2009

Supervised learning, unsupervised learning, and reinforcement learning are methods used by artificial intelligence and explained in this introduction to machine learning.

Machine Learning is a term used to describe various types of artificial intelligence technology in which a computer program has the ability to modify itself. This self-modification can take the form of self-improvement, or error-detection, and is implemented in various ways, including neural networks, and decision tree topologies. While there are many types of learning, it has been found that combining these learning styles improves the program's ability to assimilate information, just as with humans.

Unsupervised Learning in Machine Learning Software

Unsupervised learning and supervised learning in machine learning software was examined in a 2004 study published by Gokhan Tur, Dilek Hakkani-Tür, and Robert E. Schapire, called Combining Active and Semi-supervised Learning for Spoken Language Understanding. In this study, it was found that the combination of supervised and semi-supervised learning improved the ability of an AI program to learn to recognize human speech. The study demonstrated that, by using a combination of supervised and semi-supervised learning, the program learned much faster. This type of directed learning also allowed the program to learn to recognize many different types of speech automatically. This activity would typically require each utterance to be manually labeled in advance, for the understanding of the speech recognition program.

Supervised Learning Algorithms

Supervised learning in machine learning software consists of a carefully calculated set of data, to show the software program the correct patterns to follow. The program must then manipulate the input data in order to fit the patterns given by the trainer, and make adjustments as directed by the trainer. During unsupervised learning, the computer program is given the set of data, and patterns, but must make adjustments without direction. The program essentially learns by trial and error, without human intervention.

Reinforcement Learning Improves AI Performance

In addition to supervised and unsupervised learning, machine learning can also be accomplished through reinforcement. This method can also operate by trial and error, but also by observing an example of how the task should be completed. Reinforcement learning provides positive and negative feedback to the program as it learns, enabling it to improve performance.

Machine Learning Programs Reduce Workload

Machine learning is an integral part of the artificial intelligence movement, contributing to the success of neural networks and other technologies. This type of artificial intelligence software contributes to the ability of call centers to use voice recognition, data mining operations to glean more information, and reduces the amount of time necessary to create and refine computer programs. The essential function of these programs is to find ways to reduce the need for human involvement, through the increased abilities of the computer.

Expert systems applications use small artificial intelligence programs called shells to interpret human knowledge. Shell programming requires quality data for success.

Neural network applications provide an online medical diagnosis by data mining previous patient records to result in automated online health advice in the future.

A neural network's artificial intelligence, with cognitive modeling and structured learning algorithms, bring neural nets closer to independent reasoning than earlier AI.

The copyright of the article Introduction to Machine Learning in Cognitive Science is owned by Victoria Nicks. Permission to republish Introduction to Machine Learning in print or online must be granted by the author in writing.