When researching Generative AI or ChatGPT, you frequently see these terms. However, trying to explain their exact meanings and relationships can be surprisingly difficult.

“What is the difference between machine learning and deep learning?”

“Is a neural network the same thing as AI itself?”

“What exactly is AI learning?”

It is not at all strange to have these questions.

In fact, these three terms do not refer to completely separate technologies. Once you understand their relationship, you will gradually see how AI learns, recognizes images, and generates text.

In this article, we will not just memorize technical terms, but use familiar examples to understand the mechanisms of AI learning.

To help with your Generative AI Passport studies, we will explain everything step-by-step, from basic concepts to key points you need to know for the exam.

Chapter 1: Let’s grasp the big picture! The relationship between the three terms

First, let’s cover the most important point in this article.

The relationship between machine learning, neural networks, and deep learning can be summarized as follows:

Machine Learning: A general term for AI technologies and methods that learn patterns and regularities from data
Neural Network: A type of model for processing and learning from data
Deep Learning: A machine learning method that uses neural networks with many layers

In other words, deep learning is a type of machine learning that utilizes neural networks.

The relationship can be illustrated as follows.

人工知能（AI）
  └─ 機械学習
       ├─ 決定木などの手法
       ├─ サポートベクターマシンなどの手法
       └─ ニューラルネットワーク
            └─ ディープラーニング

This is a simplified representation of the typical relationship. There are various methods in machine learning other than neural networks.

Let’s note one thing here.

Neural networks and deep learning do not strictly mean the same thing. Deep learning is a learning method that uses neural networks, specifically those that have been multi-layered.

Keeping this difference in mind will make subsequent explanations easier to understand.

Let’s think about it with a familiar example

For example, suppose you want an AI to determine whether an animal in a photo is a “cat” or a “dog.”

With conventional programming, one might consider a method where humans set detailed judgment rules, such as “if the ears are triangular, it’s a cat” or “if the nose is long, it’s a dog.”

However, actual animal photos have various differences.

Even among cats, the shape of the ears and the color of the fur differ, and there are diverse types of dogs, such as small and large breeds.

It is difficult for humans to write down all the conditions as rules in advance.

Therefore, a method emerges that uses many photos to have the AI learn common features and patterns of cats and dogs from the data. This is the basic concept of machine learning.

One of the ways to realize that machine learning is the neural network, and deep learning is a representative method that makes it possible to learn advanced features by multi-layering it.

Now, let’s look at each one in detail.

Source link