Machine Learning: Everything you need to know

Machine Learning
Machine Learning

What is machine learning?

Machine learning is a branch of artificial intelligence – it refers to the use and development of computer systems which are able to learn and adapt without following explicit instructions.

By using algorithms and statistical models, machine learning is able to analyse and draw inferences from data, without the need for human intervention.

This powerful software is behind chatbots and predictive text, language translation apps, Netflix suggestions and how your social media feed is presented to show you what you want to see. It provides computers with the ability to learn without being programmed.

There are various types of single-board computers that are ideal for machine learning, including the ROCK single board computer.

How machine learning works?

The machine learning algorithm is broken up into three main parts:

  1. Decision Process: The machine learning algorithm produces an estimate about a pattern in data based on what has been inputted into the system. In general, this algorithm is designed to make a prediction or decision itself.
  2. Error function: This evaluates the prediction made by the model. An error function can make a comparison to assess the accuracy of the decision made.
  3. Model Optimisation Process: If the model can fit better to the data points in the original training set, then weights are adjusted to reduce any discrepancies or errors between the known example and the model’s prediction. The algorithm will then repeat the “evaluate and optimise” process, updating consistently until full accuracy has been achieved.

Types of machine learning

There are four basic approaches to machine learning. The type of machine learning algorithm data that is used depends on the type of data that needs to be predicted. These include:

  • Supervised learning: Data scientists supply algorithms with labelled training data and define the variables they want the algorithm to assess for correlations. The input and output is specified.
  • Unsupervised learning: This is where algorithms train on unlabelled data, scanning through data sets looking for meaningful connections. This data is predetermined, as well as the predictions and recommendations that the algorithm trains on.
  • Semi-supervised learning: This involves a mix of supervised and unsupervised learning. Data scientists may feed an algorithm mostly labelled training data, but the model is free to explore the data on its own accord to develop an understanding of the data set.
  • Reinforcement learning: This is used to teach a machine to complete a multi-step process for which there are clearly defined rules. Data scientists will program an algorithm to complete a task, giving it positive and negative cues as it works out how to complete the task. The algorithm however, decides on what steps to take along the way.