What is Machine Learning?

Intro

If you read my last article on What is AI?, you'll know that AI is a broad field with many applications. At its core, AI is about creating machines that can learn and make decisions equal to or greater than humans.

Machine Learning is a subfield related to AI that focuses on building systems that can learn from data and make predictions or decisions without being explicitly programmed. Lets dive in and explore the fundamentals of machine learning.

What is Machine Learning?

"the use and development of computer systems that are able to learn and adapt without following explicit instructions, by using algorithms and statistical models to analyze and draw inferences from patterns in data."

As humans we learn primarily through observation, this is no different. Machine learning is a way for our machines to learn by example. These examples are broadly categorized into two types based on the nature of the data it works with: Structured and Unstructured Learning.

Structured Learning

Structured learning works with well-organized, tabular data such as spreadsheets and databases. This type of data is easier to process and analyze, making it commonly used in applications like business analytics and financial forecasting.

Unstructured Learning

Unstructured learning deals with complex, raw data such as text, images, audio, and video. It requires more sophisticated algorithms to process and analyze this type of data however, this is where a lot of the generative applications come into play. This approach is commonly used in applications like natural language processing and computer vision.

Know your data

The main catch here, Machine Learning algorithms require a lot of data to train on. The output will only be as good as the foundation it was given. This is where all the hot topic ethical questions come into play. It is very easy to train model with a set of biased data, intentionally or unintentionally. More data is not always better either, it is important to understand the nuances of the data you are working with and how it can affect the outcome. Understand that an Unstructured learning approach will attempt to find patterns you may not have considered, but alas, it is all your data in which the model derives its patterns from.

Learning Approaches

Machine Learning algorithms can also be classified based on their learning approach:

Supervised Learning

Supervised learning uses pre-fabricated, labeled data for training. The algorithm learns to map inputs to known outputs. There are several types of supervised learning:

• Classification: Predicts discrete categories (e.g., spam detection)
• Multi-class Classification: Predicts multiple categories (e.g., image recognition)
• Regression: Predicts continuous values (e.g., house price prediction)

A solid example of this is the Palmer Penguin dataset. You'll see the CSV has quite a bit of data for us to train a model on. From here we can predict the species of penguin, sex, or location based on its physical characteristics.

Unsupervised Learning

Unsupervised learning works with unlabeled data. It finds patterns or structures in data without predefined outputs. Examples of unsupervised learning include clustering, dimensionality reduction, and anomaly detection. "Help me find the patterns in this data"

Each of these approaches has its strengths and is suited for different types of problems and datasets. The choice of which to use depends on the specific task, available data, and desired outcomes.

Deep Learning

You may also hear the term "Deep Learning" which is a specialized subset of Machine Learning that uses neural networks with multiple layers (hence "deep") to model and process complex patterns in data. It's particularly powerful for handling large amounts of unstructured data and has revolutionized fields like image and speech recognition.

Summary

With Machine Learning, we can create systems that can learn and make decisions equal to or greater than humans. This is one of the foundation blocks to building an AI system. These are powerful tools that can be used to solve some of the worlds most complex problems.

However, we must always consider the implications of our creations. "With Great Power, Comes Great Responsibility." We must be mindful of the data we use to train our models, and the decisions they make. It is our responsibility to use them wisely.

Up Next

Machine Learning is just the first step in the AI journey. Not all "input" data is created equal. In the next article, we'll explore other areas of our AI System such as NLP, Computer Vision, and how they all come together to form the AI system we know today.