Essentials of Machine Learning Algorithms

Broadly, there are 3 types of Machine Learning Algorithms..

1. Supervised Learning
2. Unsupervised Learning
3. Reinforcement Learning

1. Supervised Learning

This algorithm consist of a target / outcome variable (or dependent variable) which is to be predicted from a given set of predictors (independent variables). Using these set of variables, we generate a function that map inputs to desired outputs. The training process continues until the model achieves a desired level of accuracy on the training data. Examples of Supervised Learning: Regression, Decision Tree, Random Forest, KNN, Logistic Regression etc.

Examples of Supervised Learning

Supervised learning is an approach to machine learning that is based on training data that includes expected answers. An artificial intelligence uses the data to build general models that map the data to the correct answer. The following are illustrative examples.

Visual Recognition

An AI that is learning to identify pedestrians on a street is trained with 2 million short videos of street scenes from self-driving cars. Some of the videos contain no pedestrians at all while others have up to 25. A variety of learning algorithms are trained on the data with each having access to the correct answers. Each algorithm develops a variety of models to identify pedestrians in fast moving scenes. The algorithms are then tested against another set of data to evaluate accuracy and precision.

Sorting

A robot is learning to sort garbage using visual identification. It sits all day picking out recyclable items from garbage as it passes on a conveyor belt. It places items such as glass, plastic and metal into 12 bins. Each item is labeled with an identification number on a sticker. Once a day, human experts examine the bins and inform the robot which items were improperly sorted. The robot uses this feedback to improve.

Decision Support

An AI is learning to estimate investing risk. It is fed a large number of trades that real investors made and asked to estimate a risk/reward ratio for each trade based on company fundamentals, price and other factors such as volume. The estimated risk/reward ratio is then compared to the historical results of the trade at a variety of time intervals such as a day or a year.