One of the buzzwords in the financial services industry is “machine learning.” While even among experts in the field there is no single accepted definition of machine learning, in simplest terms, machine learning focuses on the use of technology (often referred to as artificial intelligence or cognitive computing), to find solutions through continuous learning. It’s software that gets better over time because of its ability to learn from experience.
For a textbook definition, let’s take the one from a class offered on Coursera. According to Andrew Ng, vice president and chief scientist at Baidu, co-founder of Coursera, and a professor at Stanford University, “Machine learning is the science of getting computers to act without being explicitly programmed.” To put this in better perspective, take email. As you receive email, you may mark certain messages as spam. Over time, your email client gets better at identifying which emails could be considered spam. The email program could then automatically put spam messages in a separate folder or simply flag them so that you can decide for yourself.
The IT Leader's Guide to Multicloud Readiness
This guide provides practical key insights and important factors to consider to make informed decisions in your multicloud journey.
Download the Guide
Within machine learning itself there are different types of algorithms that can be applied. Machine learning can leverage supervised learning, in which you train the computer to do something. Training an email client to decipher spam from legitimate messages is one example. Machine learning can also use unsupervised learning algorithms, in which the computer learns by itself, with no instruction. An example of this is having a computer take a customer database and grouping customers into different segments.
While these are the two most widely used machine learning algorithms, there’s another – called reinforcement learning – which provides positive rewards for good and correct actions, and negative rewards for bad and incorrect actions. Think of Netflix and the recommendations it provides. Those recommendations are based on several factors, including the ratings you yourself have provided, as well as your viewing history. The more you watch Netflix, the better the recommendations become. Amazon.com recommendations are another example of how reinforcement learning works. In this case, recommendations are based on several factors, including the feedback you have provided for purchased items.
While you may not realize it, there are other examples of machine learning that are part of our everyday lives. Google – and how it has mastered its ability to rank searches based on keywords – may be the best example. How Apple and Google recognize people in photos is another great example of machine learning.
In our new guide, we explore the basic tenets of anti-money laundering (AML) and know your customer (KYC), as well as how you can leverage the power of machine learning technologies to enhance your firm’s compliance programs. You can download it here.