Machine learning

What is machine learning?

As a subfield of artificial intelligence, it is essentially an algorithm that is constantly improving itself and getting better at its task. Machine learning works on the simple approach of “find patterns, apply patterns”.

Artificial intelligence encompasses the subfields of machine learning, deep learning, and neural networks.

We all already use machine learning in our daily lives without even knowing it, for example, Google Maps, Alexa, Amazon, YouTube or Netflix.

Key components of machine learning

Every machine learning algorithm consists of the following 4 key components:

• Training Data: Refers to the text, image, video, or time series information that the machine learning system must learn from. Training data is often labeled to show the machine learning system what the “correct answer” is.

• Representation: it refers to the encoded representations of objects in the training data, e.g., a face represented by features such as “eyes”. Some models are easier to encode than others, and this is crucial for model selection. For example, neural networks form one type of representation, while support vector machines form another. Most modern approaches use neural networks.

• Evaluation: This is about how we evaluate or prefer one model over another. It is often referred to as a utility function, loss function, or evaluation function.

• Optimization: this refers to how we use the space of the represented models or improve the classification in the training data to get better results. Optimization means updating the model parameters. It helps the model to improve its accuracy faster.