Definition
- Systems that automatically learn programs from data [Domingos 2012]
- Teaching a computer about the world [Mark Dredze]
The field called “machine learning” should really be called “applied statistics”
Babies learn through prediction. If something happens that they do not expect, then they are interested. Mistakes are an opportunity to learn.
Machine learning builds non-parametric model as they assume that the parameter set (or feature set)) is not fixed and can increase, or even decrease if new relevant information is collected.
Articles Related
Usage
- Recognizing patterns:
- Objects in real scenes
- Facial identities or facial expressions
- Spoken words
- Recognizing anomalies:
- Unusual sequences of credit card transactions
- Unusual patterns of sensor readings in a nuclear power plant
- Prediction:
- Future stock prices or currency exchange rates
- Which movies will a person like?
Cost
Machine Learning: The High Interest Credit Card of Technical Debt - tech debt (ie neural net are hard to understand and maintain)