Now that you know what machine learning is, let's take a look at how it's being used in the real world. In this video, learn how machine learning is currently being used.
- [Narrator] In lesson two, we discussed some common problems that machine learning can help you solve. In this lesson, we're going to run through some real examples of machine learning in our everyday lives. The first example is personal assistants whether it's Siri, Google Home, or Amazon Echo. All of the speech recognition software is built using machine learning. This small diagram illustrates how Siri takes what it thinks it hears, runs it through a model to find out what was likely said, and then executes some query to return the response. More specifically, the part where it takes what it thinks it hears, maybe bet a timer, and smooths that out into the most likely thing the person actually said, so maybe that's set a timer, is the core of the machine learning functionality here. The next example is recommendation systems, and we're literally surrounded by these from shopping on Amazon to Netflix to Facebook friend recommendations to Twitter. All of these recommendation systems are driven by machine learning. This is a very simple example of one method called collaborative filtering. Collaborative filtering uses similarities between users to make recommendations. So in this example, both Bob and Tom have bought and enjoyed pizza and salad. Based on their purchase and rating history, we know that Bob and Tom have similar taste. Bob has also bought and enjoyed soda, but Tom hasn't, so the system should recommend soda to Tom. So that's the 10,000 foot view of how collaborative filtering works. Another example of machine learning is with ride sharing like Uber or Lyft. There are a number of ways in which machine learning can be used. From route efficiency based on traffic to driver-ride pairings based on rider location and driving location and maybe traffic to determining efficient ways to pair two riders together for UberPool and then even optimal pricing based on demand of users and supply of drivers to keep the appropriate balance. Uber has made all of their traffic data publicly available and has built this really neat tool called Uber Movement to show how long it would take to get to different areas of a number of different cities given time of day, day of week, and things like that. So the colors correspond to travel times, and you can click around to see how long it takes to get to different areas outside of downtown LA. Lastly, one of the newest and most popular areas of research and implementation is in self-driving cars, and this is using deep learning which remembers just a subset of machine learning. Think about everything that goes through your mind when you're driving a car, most of which comes instinctively once you've been driving for a couple years. So that might be driving the speed limit, tracking all the cars around you, making sure you're not tailgating, and watching for brake lights. You effectively need a deep learning network to learn all these patterns that our brain has learned over years of driving.
- 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