Machine Learning in a Nutshell
Categories: Tech | Pubby Cash Received:
0 |
Click to Award
Have you used Siri, Alexa, or any other chatbot? I'm sure you have. Now many people wonder how on earth could a person inside that phone talk to you whenever you want. First of all, there's no person. It's all artificial intelligence. What's artificial intelligence, you ask? It's intelligence demonstrated by machines, unlike the knowledge displayed by humans and animals. The bot in your phone is demonstrating intelligence by communicating with you (picking up and understanding what you tell it). Some people don't stop there. How can a non-living "thing" be able to speak and understand? That all comes down to the art of coding. Which kind of coding? Machine learning is where you should look. It's the process of machine learning (basically how to train your "bot"). First, there is data collection. Your machine learning model will need lots of data to train it. Think of it like this: how would you learn a new language? You would learn each word, one by one until you know the entire language, right? The same goes for training your machine learning model. You feed it thousands (even millions) of samples so it can get a deeper understanding. However, there is one little note: don't feed it all of the data. People recommend using 80% of the data for training and the remaining 20% for testing. It's because when you're testing the data, you want to see how accurate the model is at predicting the values for the data. If the model has already seen the entire data set, then it will get 100% every time. (It's kind of like cheating in a sense) The second step is to choose the model. This step can get tricky, but you can also be creative. A machine learning model is a file that was trained to recognize certain types of patterns. You train a model over a set of data, providing it an algorithm that it can use to reason over and learn from those data. One of the most common models out there is the linear regression model. A linear regression model will draw a line given your data and predict the relationship between two variables or factors. The factor that is being predicted (the factor that the equation solves for) is called the dependent variable. (Just a quick note: if you like machine learning, then you must be into a lot of math!) Anyways, the third step is to evaluate the model. It's when the test data you set aside, in the beginning, will be used. Now you will see how accurate your model is. In most cases, the more you train your model, the more accurate it will be. If your model is performing at where you want it to be, it is time to use that model in your projects. Let's say you wanted to understand the relationship between drug dosage and blood pressure of patients. Here, you would collect samples of drug dosage and blood pressure and train your linear regression model. Then, you can use that model to predict the blood pressure of future patients (feed your model the drug dosage, and it will spit out its prediction for the blood pressure). That's all you need to know to get the general gist of machine learning.
Published from: Pennsylvania US
Liked by: H2O