Machine learning is a field of study that gives computers the ability to learn without being explicitly programmed - Arthur Samuel

Tom Mitchell's Definition

A computer program is said to learn from experience E wiht respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measrued by P, improves with experience E. - Tom Mitchel

T: classification, regression, detection
P: error rate, accuracy, likelihood, margin...(performance measure)
E: data

Example

T: Playing chess
P: Percentage of games won against an opponent
E: Playing practice games against itself.

Generalization: The goal of the machine learning

A form of abstraction extracting essential similarities.

No Free Lunch Theorem

• No machine learning algorithm is universally any better than any other
• Do not try to seek a universal learning algorithm(No absolute best algorithm)
  

Semi-supervised Learning

Black elements are unlabeled data.

LU Learning

Learning with a small set of Labeled examples and a large set of Unlabeled examples

PU Learning

Learning wiht Positive and Unlabeled exmamples(no labeled negative examples)

Reinforcement Learning

A feedback loop between the learning system and its environment

• does not experience a fixed dataset
• No supervisor, but only rewards(Learning "fills in the details")
• Feedback could be delayed, not instantaneous
• Agent's actions affect the subsequent data it receives.

References