Machine Learning For Dummies: Part 1
What is the one thing that ties Amazon’s Alexa, Google’s self-driving car and Netflix’s personalized recommendations together? We’ll give you a minute to think. Ready?
If you said that they all make use of Machine Learning algorithms, well done. You are a true Ravenclaw! (Or did you simply read the title of the article?)
Let’s Learn Machine Learning!
The advances in data processing using programming languages (like Python and R), along with access to vast amounts of numeric data, gave birth to the field of data science - the cradle of Machine Learning. Data scientists soon started developing computer models that could learn and adapt to newer data without human intervention (because we're lazy, right?). These computer models could improve from experience so we wouldn’t have to write new lines of code each time!
Put simply; any Machine Learning model learns and improves its operation over multiple iterations; the more iterations, the better it gets. This is the essence of Machine Learning. As impressive as that sounds, the greatest asset of Machine Learning, or ML in tech jargon, is that it reduces manual effort and eliminates any human flaws. Since a Machine Learning algorithm is based on numerous mathematical models, it can never be wrong, only less accurate. As Mr. Incredible once exclaimed, “Math is math!” and no one has the gall to argue with him, right? (except for Syndrome and The Underminer!)
The Humble Beginnings Of Machine Learning
How Does Machine Learning Actually Work?
We're sure that most of you must be wondering, how exactly does one go about teaching a machine? It sounds near impossible as we speak a completely different language than the only language (binary language) a computer can understand. It would be pretty cool if Robocop could use his Data Spike to interface with computers and upload the necessary data but alas, he's too busy catching criminals!
Here's how it's done - we let the machines teach themselves! Sounds like sci-fi, doesn't it? Instead of having a programmer spend hours writing a code, we allow the model to develop and improve independently. The computer model explores vast volumes of data and analyzes it to find patterns and trends. The more data it analyzes, the better it gets at finding such patterns.
Hence, ML has become instrumental in making predictions when humans can't manually go through the data to identify the trends. This technology has improved several areas of forecasting - from stock market analysis to weather predictions. No more carrying around an umbrella because the weather app said it might rain!
Imagine that we want to create a basic ML model that can count the number of cars and bikes appearing in a video. Let's call our model Chappie, cause why not? To begin, we would have to "teach" Chappie the difference between cars and bikes. The easiest way to do this is to show Chappie several images of cars and bikes - till it can accurately identify a bike from a car - be it a sports bike or a chopper.
This is the first step in Supervised Learning, a Machine Learning technique where the model is initially presented with labeled data called the training set. Once the model has analyzed the data from the training set, it can be served the data you actually want to process. To train Chappie, we would have to show him labeled images of cars and bikes. The more images it learns from, the higher its precision will be.
Another technique of Machine Learning, called Unsupervised Learning, takes only a test set as input. It identifies trends, patterns and commonalities in a dataset that has no previous training data available. Banks use such an Unsupervised Learning technique called Clustering to group customers (hence the name 'clustering'), who are more likely to enroll in a specific scheme to save on unnecessary marketing effort. Another clustering application is the spam mail filter, which analyzes a mail's content and sorts it either into your inbox or the spam folder with great accuracy. (No more emails saying that you’ve won $100,000 in the Powerball lottery!)
One more popular ML algorithm is the Reinforcement Learning model. It essentially assigns a score to every decision (think positive or negative), which helps the model reinforce certain decisions. During the initial phase, the model makes many mistakes, however, it improves through trial and error. Think of AI like the Terminator who uses extraordinarily complex ML algorithms. Save John Connor, the leader of the Human Resistance? Yes. Enable Skynet and destroy all of humanity? No. Had the Terminator used a Reinforcement Learning model, it would have been so much easier (spoiler alert!) to save John Connor from dying.
What’s Next For Machine Learning?
While Machine Learning is still spreading its influence to more niche industries, some have already integrated it into day-to-day processes. From NASA to NASDAQ, every sector can use ML to improve business functions, process automation and forecasting. If your firm generates data - be it sales reports or your employees' coffee consumption per month - Machine Learning can help you identify key trends and make accurate predictions for the future. We wouldn't want to run out of coffee now, do we?
However, ML can do much more than just make predictions. Join us in the next article of this two-part series where we will discuss the commercial applications of Machine Learning, where it is headed next and why top researchers and scientists have called for a ban on using Machine Learning — intrigued yet?
Like what you read? Head to the TechDogs homepage to find the latest tech content infused with drama and entertainment. We've got Articles, White Papers, Case Studies, Reports, Videos and Events - the complete lot to help you Know Your World of Technology.
Disclaimer - Reference to any specific product, software or entity does not constitute an endorsement or recommendation by TechDogs nor should any data or content published be relied upon. The views expressed by TechDogs’ members and guests are their own and their appearance on our site does not imply an endorsement of them or any entity they represent. Views and opinions expressed by TechDogs’ Authors are those of the Authors and do not necessarily reflect the view of TechDogs or any of its officials. All information / content found on TechDogs’ site may not necessarily be reviewed by individuals with the expertise to validate its completeness, accuracy and reliability.
Join The Discussion