Welcome to the era of machine learning where machines are becoming independent, and humans are becoming more dependent (on the machines). Care to venture? Let's see what history has for us.
Since the beginning of time, humans tend to excel in games of their own. And why not? They designed these games, organized them, participated in them, and so won them as well. Until one day, the course of history was forever changed by something Ã¢â¬â other than humans.
1.1 Garry Kasparov vs. Deep Blue one unforgettable chess match
1.2 IBM Watson makes a record on Jeopardy!
1.3 World stands still AlphaGo vs. Lee Sedol
1.4 AlphaStar's new milestone in the real-time strategy game Starcraft
This journey takes us back to the late 90s (1996-1997) when the then chess grandmaster Garry Kasparov faced an opponent, unlike any humans. And so, his opponent was an AI and machine learning algorithm-driven chess-playing computer named Deep Blue, manufactured by IBM. Their battles over the two years led to some unprecedented results which the grandmaster was not ready for, and nor was everyone. In short, Garry Kasparov caved in and accepted his defeat by resigned mid-game for the first time in his career. This was the pathway for machine dominance in the human world. (HISTORY, 2019)
Back then, Deep Blue relied mainly on Artificial Intelligence and less on Machine Learning. But later, IBM created Watson in 2005, which was an ML system that further spread the horizon of AI. In 2011, IBM Watson took part in the famous American (quiz) game show named Jeopardy! to play against the record-holder Ken Jennings and another contestant named Brad Rutter. If there was any doubt, who would win, then let it be clear that Watson won by a fine margin than its opponents and earned $77,147 from the show.
Fast pacing towards 2016, another formidable force that came into being was AlphaGo. This enhanced program used machine learning to assess the game of Go (Asian board game). It made its move from self-learning and self-improvement. Now, in the game of Go, participants could find themselves up to around 2*10^170 number of states. And every decision they would take led to multiple states, which led to more states. So, AlphaGo played with itself in some of the states and predicted the outcome based on it. Later, on March 19, 2016, they tested the program, and to everyone's surprise, AlphaGo defeated the legendary South Korean Go player Lee Sedol.
After the introduction of AlphaGo, they built another program that could win Chess, Go and Shogi (Japanese chess). They named it AlphaZero (2017), and it outperformed and defeated AlphaGo quite easily. Later in 2019, they transformed AlphaGo into AlphaStar to defeat anyone in the real-time strategy game Starcraft. In January, AlphaStar defeated the two top players in the world and became a grandmaster in this game. (Walker, 2020)
Machine learning is an extended branch in the field of Artificial Intelligence. So, in order to understand ML, we need to get a brief idea about AI. AI surrounds a wide range of areas in computer science. Its primary purpose serves to build smart machines that can perform tasks that usually require human intelligence. So, in short, AI works like a human mind, learns from experience, and solves problems according to the percepts it receives from its environment.
Before getting into any debates, artificial intelligence uses different approaches to solve problems, and one of its most successful strategies in machine learning. So, we need ML to achieve artificial intelligence. Now similarly, machine learning receives percepts from its surroundings, but it is not limited to perform tasks according to those prior percepts. It grows beyond its given codes and performs freely like a human brain through imitating the cognitive process of learning. To sum it up, ML analyses data makes predictions using statistical machine learning algorithms and goes further to perform tasks beyond its initial programming. It is like a living program that keeps on evolving on its own.
ML has slowly found its way to make an impact on our daily lives Ã¢â¬â directly, sometimes indirectly as well. Usually, when we search things online, how often do we see recommendations of similar things popping in front of our screen? Every time! For example, when we are searching on websites like Amazon, eBay, and alike, it is safe to say that for the next few days, we will be getting similar products from these websites. Well, this is part of the recommendation system powered by ML, and it is operating in real-time to make the search experience better than ever.
Here are a few types of machine learning applications used to help the system become more assertive:
Machine learning has made its presence felt in the tech industry, and the flourishing results have set a pathway to bring innovation to other sectors as well.
Here are the implications:
A storehouse for streaming old and new movies and series is not the only thing that Netflix offers. Instead, it is their recommendation system that allows them to keep their viewers from getting up from bed or couch. This is possible because of ML as it uses the streaming history and preferences of the users to predict individually Ã¢â¬â in real-time.
Have you noticed anything different about Twitter? Previously, Twitter used to show the most recent tweets on the user's timeline. This took some of the not-so-recent but yet sweet posts away from the users. It was a major bummer, and Tweeter seemed to notice it. A significant change came into being by using ml to prioritize tweets based on a user-specific relevance scorecard (focusing on their likes and dislikes with popular accounts). (Gottsegen, 2019)
After going through the above ML applications and implementations in almost every industry, if not all, it is impossible to disagree that this branch of AI will take over. Already, it has turned the tables for the tech industry by providing solutions to complex and time-consuming problems more efficiently, effectively, and in real-time. Starting with Facebook's 'people you may know' list to Apple's Siri answering funny questions like 'Do you have a pet? Yes, I have an Angry Bird.', these are all possible because of a learning machine. The best part about ml is that the world is yet to explore the limitless possibilities it encompasses fully. While humans can only do so much due to ml, they can now do so much more. (Vinayak, 2020)
Machine Learning is generally divided into three broad segments. They have supervised learning, unsupervised learning, and reinforcement learning. Each segment serves a different purpose and brings specific results from the data it uses.
In simple words, supervised learning is a process where the system tries to learn from previous results. Labeled data (i.e., known data) is fed to the system with their correct outcome, and the machine learning algorithms are trained accordingly to develop a mapping pattern. It uses methods like classification, regression, and prediction to predict values for unlabeled data. Supervised learning is commonly used for indicating the weather, credit card fraud cases, and so on.
In unsupervised learning, the system is not trained with labeled data and correct outcomes. The system has to find a pattern within the dataset and segregate similar data from the dataset to form different groups. This method is popularly used in market segmentation, feature selection, finding outlier in a dataset, and so on.
This particular learning follows a trial and error process where the system predicts the outcome and the agent (the decision-maker) decides whether it is satisfactory or not. Here system accepts an outcome only when the agent chooses to. Until then, it keeps on looking for possible outcomes that may yield the best reward. Reinforcement learning machine learning is used in robotics, recommendation system, and so on.
ML requires a few prerequisites to understand this field of computer science. So, if anyone is interested to learn more about ML and excel in it, these are the following requirements:
Basic knowledge of programming language, such as Python
Data science methodology and tools
Sound knowledge of statistical analysis and probability
SQL for Data Science and Data wrangling
Basics knowledge of Artificial Neural Networks, Big Data and Deep Learning
The above-mentioned requirements are a must to learn Machine Learning.
But if anyone wants to start from the beginning and work their way onto the top, then there are many online courses that offer boot camps for beginners. These courses also come with highly valued certificates paving the way in the job market and career.