Before I dive into the details of a topic, I like to first define key terms that I’ll encounter. With that mindset, my initial question when starting out with machine learning is, “What is machine learning anyways?” Intuitively, we can infer its definition by splitting it into its two root words, machine and learning. From this, our initial definition for the term could be the ability of a machine to learn. But what does this mean exactly?

IBM describes machine learning as “a branch of artificial intelligence and computer science which focuses on the use of data and algorithms to imitate the way that humans learn, gradually improving its accuracy.” Although similar to our initial definition, this statement enlightens us with some additional insights.

It states that machine learning is a branch of artificial intelligence, which means that these two have a parent-child relationship. This is different from how I used to perceive it as these terms are commonly used interchangeably. In a separate article, IBM defines artificial intelligence as “technology that enables computers and digital devices to learn, read, write, talk, see, create, play, analyze, make recommendations, and do other things humans do.” With this definition, it is more clear to me now where their difference lies. Artificial intelligence describes the study of how to make machines behave like humans. On the other hand, machine learning only covers a subset of that, which is how machines can learn like humans.

Going back to IBM’s definition of machine learning, another point I’d like to focus on is that it focuses on data and algorithms to achieve this goal. From this, we can conclude there are two parts of the machine learning process, one related to data and one related to algorithms. But what do you do in these two parts? Let us look at another definition from a Google course to further our understanding. In the Intro to Machine Learning Google Developer course, it defines machine learning as “the process of training a piece of software, called a model, to make useful predictions or generate content from data.” Here we can see the purpose of each of the two parts. In the first part, we have to gather data relevant to what we want the machine to learn. The second part then focuses on running this data through the model to train it using algorithms. The outcome of this process is a model that the machine can use to create predictions or generate content. This is the goal of the machine learning process.

After understanding the flow of the process, I can see similarities between how we humans learn and how machines “learn”. When we are born, we start without any knowledge about the world. We then gain a better understanding of how the world works through our insights and experiences. We then adjust our previously held beliefs and behaviors in accordance to these. This feedback loop is then repeated again and again until we get the most accurate representation of the world we believe in. Similarly, machines start without any understanding of the problem we wish to solve. In the classic approach of programming, we would think of a solution to a problem and then translate this to code for the machine to process and execute. In machine learning on the other hand, we feed the data to the machine and allow it to come up with its own solution by understanding the data. The data the machine is trained on is analogous to the experiences we humans learn from.

After looking at various definitions, I know have a clearer understanding of what machine learning is. I hope I was able to share that understanding with you as well.

Now, I plan to begin learning the actual machine learning content. I will do this by taking the Machine Learning Specialization in Coursera. Based on my research, this seems to be the best recommended course for beginners. Information in my next posts will probably come from there but I will also try to vary my sources to have a better understanding overall.

Sources:

https://www.ibm.com/topics/machine-learning

https://www.ibm.com/topics/artificial-intelligence

https://developers.google.com/machine-learning/intro-to-ml/what-is-ml