ML101: Types of Learning
- Supervised Learning:
When labels or information is provided with respect to data set that has been provided. It is expensive and time consuming to label data and sometimes require expertise to access this kind of data. Divided into 2 kinds of problem:
a. Classification: when we have categorized output.
b. Regression: when we have real value output.
- Unsupervised Learning:
When no labels or information is provided with respect to the data provided. Unlabeled data is easily available, cheap and easy to store. Divided into 2 kinds of problem:
a. Clustering: when we need to discover the inherent groupings in data.
b. Association: such problems require discovering rules that describe large portions of data.
- Semi-Supervised Learning:
Problems where huge data set is present and only some data is labeled and major part is not labeled.
- Reinforcement Learning:
These algorithms train system to make specific decision. Machine is exposed to an environment where it trains itself continuously using trial and error method. These algorithms learn from past experience and try to capture best possible knowledge to make accurate decision.