Feature Story

Hate making decisions? Ask today's Magic 8 ball—the algorithm

by Stephanie Chan

Hate making decisions? Ask today's Magic 8 ball—the algorithm

Machine learning and AI are helping us decide everything from clothes to cloud user activity.

Algorithms are helping us make decisions every single day—they’re the seemingly magical models that can bring certain ads to the top of our social media feeds or pinpoint which shows you might enjoy on Netflix. But algorithms are perhaps simpler than they seem. They are a formula for processing information or performing a task. At its very basic, an algorithm can be a dinner recipe—at its most complex, algorithms can sort through tons of information for security purposes. In today's world, where algorithms are used in practically every technology, the humble algorithm works behind-the-scenes to determine what we see online. Combined with data, algorithms are the building blocks for machine learning and AI.

Beyond movie recommendations and social media ads, algorithms are helping recruiters hire potential employees and even the most flattering outfit choices. Online styling service Stitch Fix recently hired more than 100 data scientists to build a set of algorithms to determine the size, silhouette, and style of each clothing item. These are then used to see which clothes “go” or “match” with other pieces of clothing. This feature, called “Shop Your Looks”, suggests these algorithmically-created outfits for its clients. But Stitch Fix’s algorithms are working even harder to determine things like “latent style”, or the types of items clients actually like, and “latent size”, or clothes that best fit the nuances of different bodies.

See also: How AI and machine learning will improve intent-based networking 

Algorithms and AI are entering the world of sports, as well. English soccer team Leatherhead FC is working with IBM and its question answering computer system Watson to improve player and team performance. This AI project analyzes players pre and post-match, watching previous performances and upcoming events. Watson is able to do this in two ways—Watson Discovery looks at matches and social feeds to get information about opponents, and Watson Assistant provides key data points and video clips from the games. ZDNet writes that Watson is being used in other sports, like tennis, golf, and basketball.

See also: Why AI and 5G make a good team

Cisco also uses algorithms to examine data infiltration trends, a necessary move as more data and applications move to the cloud. In the Cisco 2018 Annual Cybersecurity Report, the company used a machine learning algorithm to look at trends for 150,000 users in 34 countries. The algorithm was then able to flag deviations from the norm for each user—it flagged 0.5 percent of users for suspicious downloads, a whopping 3.9 million documents from corporate cloud systems. An increase in downloads can show a possible threat, and discovering these abnormalities in a timely manner is crucial for defenders to investigate the threat quickly.

Although it can seem like machine learning algorithms have the wizardry of Merlin or a magic 8-ball, Cisco’s Han Yang says that it is not all rocket science. To learn why and to read more about Cisco’s work with machine learning, check here.

###

We welcome the re-use, republication, and distribution of "The Network" content. Please credit us with the following information: Used with the permission of http://thenetwork.cisco.com/.