Machine Learning for the Average Joe

When you hear the term "Machine Learning", do you have sudden bouts of panic and fears of a future run by Terminators? Well fear not, because I'm going to prove that Machine Learning (ML), is actually very simple on a fundamental level. Of course, that's not to say that ML does not progress to levels of complexity that will make your brain hurt, but let's cover the basics first. By definition, Machine Learning "is the scientific study of algorithms and statistical models that computer systems use in order to perform a specific task effectively without using explicit instructions, relying on patterns and inference instead". 

Essentially, we're providing the computer with Statistical Models, (which I'll cover in another post),  in order to achieve a task through repetition and patterns. For example, if one decides to create an app/program for facial recognition (quite popular nowadays), the ML's algorithm would have to be "trained" with thousands of practice data; in order to identify patterns in facial features so that it would eventually be able to distinguish whether a given image is a face or not. There are 3 main types of Machine Learning: Supervised, Unsupervised, and Reinforcement Learning. Rather than copying and pasting the definition of each of them from Wikipedia, I'm going to describe them using key points. 

Supervised Learning (SL): you already have "Labeled Data" & are trying to predict a label based off of known features. The labeled data simply means to column names of the table containing the samples/observations of the data. The SL algorithm receives a set of inputs along with the corresponding correct outputs, and the algorithm learns by comparing it's actual output with correct outputs in order to find errors. Then it modifies the model accordingly to be more accurate for future "testing" data. SL is commonly used in applications where historical data predicts likely future events.

Unsupervised Learning (UL): You have Unlabeled Data and are trying to group together similar data points based on their features. In these cases, the system is not told the "Right answer" and the algorithm must figure out what it is being shown. In other words, the goal is to explore the data & find some structure within. Popular UL techniques include: K-Means Clustering, and Singular Value decomposition (don't worry, I will cover these in more depth in a later post).

Reinforcement Learning: the algorithm learns to perform an action from experience, or "trial & error". There are 3 components of Reinforcement Learning; 1) the agent (learner/decision maker), 2)the environment (everything the agent interacts with), and 3) actions (what the agent can do). The main objective here is for the agent to choose actions that maximize the expected reward over a given amount of time.

See? That wasn't too excruciatingly painful right? Hopefully this description has given you a better understanding of the basic types of Machine Learning. In my next post, I will demonstrate how to use various ML models using Python, which is an excellent programming language that can be used for Data Science.

Thanks for reading!