Lecture 1 Machine Learning (Stanford)
-
http://www.youtube.com/watch?v=UzxYlbK2c7E
Machine Learning approached from diverse fields.
Inter-disciplinary topic
having large impact // lots of implication // to science and industry
Early work in AI - viewed as a new capability for computers.
- reading hand-written letters, translating - extract characters
- fly a helicopter
: using learning algorithm. pretty much the only approach.
- database mining.
- checks -> processed by learning algorithm / ??
- credit card // fraud, stolen check
- guess what clients wants
state of art machine learning algorithm
Using Matlab, Octave.
-
Machine Learning definition
Arthur Samuel: Field of study that gives computers the ability to learn w/o being explicitly programmed.
- The computer learns how to play Checker game
Tom Mitchell: Well-posed Learning Problem: A computer program is said to learn from experience E with respect to some task T and some performance measure P, if its performance on T, as measured by P, improved with experience E.
1. Supervised Learning
Cartesian coordinate - predicting based on events accumulated
- Regression. (A: continuous)
what about discrete data set?
-
Classification (discrete values: 0, 1)
what about diverse input data? what if its infinite?
2. Learning Theory
reading zip code?
will it work? how many training data will be needed?
3. Unsupervised Learning
unlike supervised learning that data gives right answers,
just a data set // having no right answers..? 'classification'? grouping
image processing, -> vision, image processing (clustering)