Machine learning is so pervasive today that you probably use it dozens of times a day without knowing it
It’s never too late to learn. This saying is true not only for human beings. Machines can learn as well. While people use past experiences, theoretical knowledge, and continuous training to improve their skills or raise awareness, machines rely on data algorithms to perform their functions better without using explicit instructions.
What is machine learning?
This widely-used notion is an integral part of Artificial Intelligence (AI). Truly, an ability to learn is the first step towards intelligence. Therefore, machine learning (ML) is, basically, the process of granting a machine or model access to data and letting it learn for itself, without further human intrusion. ML is dynamic. It is able to modify itself when exposed to more data.
The term was initially used by Arthur Samuel, an American pioneer in the field of computer gaming and AI, back in the late 1950s while he was working on his Checkers-playing Program at IBM. This project was unique and appears to be the world’s first self-learning program. The idea behind it was that we shouldn’t have to teach computers, but rather, we could let them learn on their own. Samuel coined the term “machine learning” to describe the ability of computers to learn autonomously.
Many of the fundamental principles of modern machine learning predate computers. They come from mathematics and statistics. They have a pragmatic focus on results, which can be interpreted for predictive modeling.
The other part of the machine learning theory is based on biology. In 1943, neurophysiologist Warren McCulloch and mathematician Walter Pitts joined their efforts to create the first neural network model. This was a computing system inspired by, but not completely identical to, the biological neural networks that constitute animal and human brains. In other words, the scientists tried to create a copy of our brain system using an electrical circuit.
It worked well for computer science that started developing rapidly. Neural networks brought optimization of many processes, keeping on measuring the error and modifying their parameters until they couldn’t achieve any less error.
Where is it used?
Machine learning is so pervasive today that you probably use it dozens of times a day without knowing it. During the past decades, ML has given us self-driving cars, speech recognition, effective web search, and a vastly improved understanding of the human genome.
- Natural language processing. Siri, Alexa, Cortana, Google Assistant, and other similar services rely on language data processing to recognize speech patterns. They use synthesis, allowing them to understand or pronounce words they have never encountered before. Google Translate is also based on a set of machine learning algorithms that update the dictionary automatically learning from users’ input, like new words and syntax.
JPMorgan Chase launched a Contract Intelligence (COiN) platform that leverages natural language processing to review legal documents and extract essential data from them. Therefore, instead of wasting 360,000 labor hours on manual review of dozens of thousands of commercial credit agreements, they can manage this routine task in a few hours.
- Gaming. DeepMind Technologies, a British company acquired by Google in 2014, gained prominence when it developed a neural network that could learn to play video games by analyzing the behavior of pixels on a screen. Developed by DeepMind researchers, AlphaGo neural network beat the best world’s professionals in the hard game of Go in 2016-2017. The first similar case was registered in 1997 when the IBM computer named Deep Blue beat world chess champion, Garry Kasparov.
- Customized recommendations. Netflix, Amazon, and Facebook display every individual user with tons of recommended content based on their search activity, likes, and previous behavior. Machines match sellers with buyers, movies with prospective viewers, photos with people who want to see them — all of which have improved our lives and online experiences significantly.
- Finance. Today, machine learning plays an integral role in many phases of the financial ecosystem, such as approving loans, assessing credit scores, managing assets, or estimating financial risks.
Robo-advisors (companies such as Betterment, Wealthfront, and others) are ML algorithms built to optimize users’ financial portfolios adapting them to their personal goals and risk tolerances.
Wells Fargo uses an AI-driven chatbot through the Facebook Messenger platform to communicate with users and provide assistance with passwords and accounts. Many banks and fintech startups across the world appreciate chatbots too.
- Algorithmic trading. It involves the use of complex AI systems to make extremely fast trading decisions. While human financiers cannot predict patterns of real-time trading behavior, machine learning algorithms can — and they respond to changes in the market much faster than humans. Most hedge funds and financial institutions do not openly disclose their AI approaches to trading, but it is believed that machine learning and deep learning are playing an increasingly important role in this sector.
- Fraud detection. Using machine learning, systems can detect unique activities or behaviors (“anomalies”) and flag them for security teams. Banks, e-commerce platforms, and financial services providers can use this technology to monitor thousands of transaction parameters for every account in real-time. Adyen, Payoneer, Paypal, Stripe, and Skrill are some notable fintech companies that invest heavily in security machine learning.
The algorithm examines each action a cardholder takes and assesses if an attempted activity is characteristic of that particular user. If the system identifies suspicious account behavior, it can request additional identification from the user to validate the transaction. Employees may contact the account holder on the phone, via email or messengers, or even block the transaction altogether if there is a very high probability of fraud.
- Autonomous vehicles. Cars that don’t require a human operator originated from a set of machine learning algorithms that enabled cars to learn how to drive safely and effectively. Tesla, Waymo, Uber, General Motors, and many more companies are testing their self-driving vehicles aimed at different goals and projects. Even the US Postal Service has cooperated with the self-driving trucking company TuSimple to haul mail during their trial period. TuSimple will carry the mail on five round trips between the USPS’s Phoenix, Arizona, and Dallas, Texas distribution centers, which is a stretch of more than 1,000 miles.
Where you can learn it?
Some of the best online courses for machine learning are:
- Udemy – a vast online learning platform – offers Data Science and Machine Learning Bootcamp with R along with other popular courses.
- ProjectPro offers a hands-on approach to mastering machine learning and data science through 150+ solved end-to-end deployable machine learning and data science projects. They also provide 2000+ FREE data science code examples that can help one master the foundations of basic data science and machine learning concepts.
- EdX offers online Machine Learning courses from Microsoft, IBM, MIT, and other top universities and institutions around the world. Machine Learning with Python: A Practical Introduction is a basic free course.
- Learning from Data (Introductory Machine Learning Course) is offered by the California Institute of Technology through Class Central. It is also free of charge.