Machine learning, deep learning, and artificial intelligence are related terms, but quite different. In this video, learn the correct definitions and uses of these terms.
- [Narrator] In this lesson, we're going to try to develop an understanding of how the fields of machine learning, deep learning, and artificial intelligence are related. These terms are wrongfully used interchangeably. As we work through clarifying what machine learning is, I also want to be clear about what it is not. Just doing a simple Google search of these terms will return hundreds of Venn diagrams attempting to describe how these three terms are related. Here's one such example, and I only show this just to illustrate how difficult it is to understand some of these diagrams.
There's just way too much going on here to really understand the certain relationships that we're interested in. What I'm hoping to lay out in the coming slides is a much more simplified explanation of how these terms are related. So beginning with machine learning. This is the definition that we saw back in lesson one. Machine learning is the process of fitting functions to examples and then using that function to generalize and make predictions about new examples. And then we also discussed a simpler related definition that is just pattern matching.
So a model will basically learn a pattern from past examples and then it will fit a function to it and then it'll use that function to pick up on similar patterns in future data to make a prediction about it. So let this circle on the right-hand side represent all things machine learning. Now deep learning. Deep learning resides entirely within machine learning. That is, it's a subset of machine learning. The definition of deep learning is very similar to the definition of machine learning. It is still fitting a function to examples, but the difference comes in the fact that these functions are then organized as connected layers of notes.
In other words, you'll have many functions connected together in one network where each function is responsible for a very specific thing. This is still all with the goal to generalize and make predictions about new examples using our network of functions. An alternative definition is just more sophisticated connected pattern matching. So again, instead of just using one function to match a pattern, you're just using many functions to match a pattern. Now this may seem very vague, and that's quite okay.
Deep learning is not the focus of this course. The one important thing to take away is just that deep learning is a type of machine learning. It is a subset of machine learning. There are things that are machine learning that are not deep learning, but there's nothing that is deep learning that is not machine learning. Now the last layer that we're going to add on top is artificial intelligence. Artificial intelligence is a superset of machine learning and deep learning. In defining it, we're going to split it into two separate definitions: weak AI and strong AI.
Weak AI is intelligence specifically designed to focus on a very narrow task. The key take away here is that weak AI is machine learning, intelligence designed to focus on a very narrow task like identifying fraudulent credit card charges, but then you're not going to take that model that was built to detect fraudulent credit card charges and recommend Netflix movies to you. Strong or general AI is a machine with consciousness, sentience, and a mind. General intelligence is capable of any and all cognitive functions and reasoning that a human is capable of.
So general AI is a superset of many things. Only one of those things happens to be machine learning, but there are many things under the umbrella of AI that is not machine learning. So this is just a high-level overview of how these three terms are related. Deep learning is a subset of machine learning which is a subset of general artificial intelligence. They should not necessarily be used interchangeably without specifying what type of AI we're talking about.
- What is machine learning (ML)?
- ML vs. deep learning vs. AI
- Handling common challenges in ML
- Plotting continuous features
- Continuous and categorical data cleaning
- Measuring success
- Overfitting and underfitting
- Tuning hyperparameters
- Evaluating a model