Getting Started with Machine Learning

Machine Learning has become increasingly popular in recent years, due to the vast amount of data that is available and its ability to produce reliable results. It is not uncommon to see machine learning being used in many applications such as fraud detection, predicting stock market prices, and building a self-driving car. But what IS machine learning?

What is Machine Learning?

Machine Learning (ML) is a process by which a computer learns a set of patterns (without being explicitly programmed) from the data that it is given. There are multiple definitions of ML, all of which expresses and emphasizes the idea that a computer is able to recognize patterns given a set of data. Here’s an example:

Suppose that you wanted to teach a child to differentiate a dog from a cat. In order to do this, they would have to know what a dog looks like as well as the cat. Thus you would show them pictures of both a dog and cat until they recognize specific patterns that can differentiate the two. Once they find these specific patterns, the child knows the difference between a dog and a cat.

In machine learning, the child is analogous to the computer and the pictures are the data the computer is given.

Implementing an ML model is much harder than how it was described above as there are many other variables to consider such as learning rate, loss function, etc,  that can dictate the accuracy of your results.

We understand that Machine Learning may be intimidating for those that don’t have a background in Data Science . With terminology such as linear regression, logistic regression, training and stochastic gradient decent (what?!), It’s easy for someone to be confused very quickly. However, you have to be familiar with these words (and many more) to understand the basis of ML. Thus we would like to give you some recommendations to get started:

  1. Andrew Ng’s Machine Learning course on Coursera

    This 11-week online course is a great introduction for anyone interested in getting into machine learning/deep learning. Although it is an 11-week course, you can skip through the weeks, or start wherever you’d like. Knowing the programming language is not required. This course goes through the intuition of many of the machine learning models and introduces the math that is involved. After doing the readings or listening to the videos for a specific week, you can apply what you learned in an exercise for that week.

  2. Google’s Machine Learning crash course

    This course is also a great introduction to machine learning, but requires the reader to know basic linear algebra and to be proficient in basic-level python coding. It takes you through the concepts and allows you to practice what you learned in an environment on your browser with no local set-up. You will also learn some of the basic and high-level implementations of some machine learning models with TensorFlow’s APIs.