Machine learning has become a buzzword in recent years, but the term dates to 1959 when AI-pioneer Arthur Samuel described machine learning as giving “computers the ability to learn without being explicitly programmed.” Decades later, having witnessed the exponential growth of computing power and its transformational potential, information security expert and author, Daniel Miessler called machine learning the new statistics. “It’s nothing less than a foundational upgrade to our ability to learn about the world, which applies to nearly everything else we care about,” he wrote.
With advances in computer science and data digitization, the variety of machine learning applications—as well as the potential value—have grown exponentially. But what exactly does machine learning encompass and how can companies put big data to work to train machine learning algorithms?
Pattern recognition just one part of machine learning
A subfield of artificial intelligence (AI), machine learning enables computer systems to automatically learn and improve from experience. It falls into four classifications:
- * Supervised machine learning—These human-trained algorithms are given tagged ‘training’ data and an expected output. The result is a rules-based, decision-making process that enables the machine learning algorithm to forecast on new and existing data. Examples are regression analysis and predictive modeling.
- * Unsupervised machine learning—These self-taught algorithms are designed to gather, classify and categorize data. Subsequently, decisions are made on the data. Unsupervised machine learning, such as clustering, is used to solve very complex problems.
- * Semi-supervised machine learning—These algorithms fall somewhere in between supervised and unsupervised learning. They use both labelled and unlabelled data for training.
- * Reinforcement learning— This variety of machine learning takes cues from behavior psychology. Used for game theory and simulation optimization, reinforcement learning facilitates actions based on rewards for previous actions.
Regardless of the type, machine learning algorithms offer a distinct advantage given the volume and velocity of big data growth. They enable users to process high volumes of data more quickly than humans to recognize patterns, classify data and make correlations automatically.
Companies are racing to take advantage of machine learning
The problem-solving potential of machine learning—and deep learning which uses even larger data sets—makes it an attractive undertaking for organizations across numerous industries. “We found that much AI investment is currently dominated by large tech companies like Google, Amazon, Baidu and Microsoft,” writes Priceonomics, noting that 10 to 30 percent of non-tech companies are also exploring the potential of AI.
Machine learning is the big winner in the investment scene. According to research by McKinsey, machine learning accounts for 62 percent of AI investment, twice that of the next closest category. Where are machine learning algorithms proving their value already?
One industry that has embraced machine learning is Finance. Fortune notes, “Firms reported to be using A.I. for investment research include BlackRock, Fidelity, Invesco, Schroders, and T. Rowe Price.” The article further cites a rush to hire alternative data analysts in the past five years, quadrupling their numbers. Machine learning empowers financial services organizations in numerous ways:
- * Conducting sentiment analysis of news related to stocks
- * Identifying trading signals from a range of news, financial and legal data
- * Detecting fraud by spotting patterns and using predictive analytics
- * Using natural language processing to stay alert to and interpret new regulations
- * Automate compliance checks to mitigate money laundering risk
German lender Commerzbank, for example, reportedly plans to automate up to 80 percent of compliance checks by next year. Has your organization started down the path of machine learning?
