What is Machine Learning?

2 minute read

image-left

Introduction

Arificial intelligence (AI) and machine learning is gaining popularity and use in our daily lives and filtering into industries including the actuarial profession. They are opening up doors and convenience to our day-to-day lives. The potential scope for improvement to existing actuarial processes and existing business solutions seems boundless.

Overview of AI, Machine Learning & Deep Learning

The below schematic gives an overview of the relationship between artificial intelligence, machine learning and deep learning:

image-left

Artificial intelligence Overall area where human intelligence and behaviour is mimicked by computers and machines.
Machine learning Computers can “learn” from the data, without a pre-programmed set of rules.
Deep learning Area is inspired by the way the human brain works e.g. network of neurons.

What is Machine Learning?

’’The field of machine learning is concerned with the question of how to construct computer programs that automatically improve with experience.” - Tom Mitchell, Machine learning 1st Edition (1997) .

Machine learning is a sub-category of artificial intelligence and refers to the process by which computers develop pattern recognition, or the ability to continuously learn from and predictions based on data, then adjust without being specifically programmed to do so.

There are two types of problems that machine learning can solve:

  • Regression analysis: The problem we are trying to solve is if we can predict a “response variable” (i.e. observed outputs) based on a “predictor variable” (i.e. adjustable input) , for example, predicting a house price from the square metres of the house; and
  • Classification: The problem we are trying to solve is if we can determine what a particular instance belongs to, for example, is a particular email spam or not.

The three major machine learning algorithms are as follows:

  • 1. Supervised learning: Supervised learning uses the dataset which contains both the predictor and response variables where the data is labelled, and it uses a known dataset to make future predictions or classifications;
  • 2. Unsupervised learning: Unsupervised learning uses the dataset which does not have a labelled response variable and is commonly used for clustering analysis i.e. to identify similar groups or patterns;
  • 3. Semi-supervised learning: Semi-supervised learning uses the dataset which is partially labelled and the algorithm uses the labelled data to understand the parameters for interpreting the unlabelled data.

Machine learning algorithm – Selection process

The selection of machine learning algorithm depends on the data available and the question we want to ask.

image-left

Further reading