DISTINGUISHED DAVIDSONIANS

Phillipe Loher Leads Local Tech Group in Helping Machines Learn

by | Oct 12, 2018 | Bottom Right Red Box, Distinguished Davidsonians, News

 

With just a few minutes of instruction, even kids can learn to play tic-tac-toe. But teaching even a powerful computer to play is another matter entirely. “It’s not as easy as you might think,” said Phillipe Loher, the founder and leader of a growing community of people in Davidson who are collaborating to teach each other new technologies. The group is free for anyone to join, learn, and participate, and now has more than 600 members from throughout the region.

Loher is particularly interested in a subfield of artificial intelligence called “machine learning,” a technology that exploded in popularity around 2011 and mimics the way humans learn. By using data science techniques and computer algorithms, these systems learn from data—lots of data—Big Data.

For example, Loher said, “My small children can easily tell what’s a dog and what’s a cat. But if I ask them how they know which is which, they have no idea. They just know. Machine learning identifies which is which by feeding in huge numbers of cat and dog images, instead of looking at a set of rules about which is which.”

Machine learning is changing just about every industry, from helping doctors fight cancer, implementing facial recognition, enabling self-driving cars, making robots smarter, and making better cooking recipes.

Using machine learning, Loher and other group members were able to teach a computer to play tic-tac-toe with minimal human input. “The amazing thing is that we never taught the computer anything,” said Loher. “We never told the computer how to use a mouse, or which symbol was an X and which was an O, or when it was time for each symbol to take a turn. We just turned it on and the machine started teaching itself the game.”

A native of south Charlotte, Loher grew up as an avid student of math and science. He earned a computer science degree from N.C. State in 2002, then moved north for 13 years. He first worked in Boston as a software engineer and manager for IBM. But he grew increasingly interested in applying his technical skills to an understanding of how the human body works. He got that opportunity in 2011 and has recently been appointed director of machine learning in the computational medicine center at Thomas Jefferson University in Philadelphia.

“We do a lot of fun things at Thomas Jefferson,” he said. He and his research team conduct experiments to reverse-engineer cellular function. They create and discover new molecules, and they manipulate RNA to try to determine its functions and applicability to diseases and cures. Their ground-breaking human genome research has uncovered new categories of molecules, as well as a myriad of previously unsuspected RNA regulators. Since 2012 he has authored or co-authored over two dozen articles in medical journals about their work.

Loher moved his family to Davidson three years ago to be closer to family and be part of a community that cherishes learning on many levels. But he continues his work at Thomas Jefferson remotely.

Soon after establishing his family in Davidson, Loher began investigating the idea of developing a local MakerSpace. He had been associated in Boston with a fabulous facility that gave members access to electron microscopes, 3-D printers, wood shops, metal working shops, leatherworking spaces, pottery shops and more.

But what Loher appreciated more than the facilities was the way the MakerSpace brought people together. Patrons would collaborate informally, and routinely describe their projects in more formal presentations. “You literally had rocket scientists working beside blue collar workers,” he recalled. Loher taught a class on a program that, among other capabilities, vastly speeded up internet searches. “It was exciting to be part of that knowledge eco system,” he said.

As a start to creating a similar facility in Davidson, he purchased and posted a Meet Up announcement for people interested in collaborating on tech projects and machine learning. Attendance at the weekly meetings, which were held inside or outside Summit Coffee, was minimal at first. But the group got a big boost when the Davidson Public Library agreed to let the machine learners use its community room every Tuesday evening. Having a guaranteed, comfortable space rapidly increased attendance, and with increased attendance came more collaboration with other tech communities.

The group’s primary activities are hosting machine learning-related presentations and working together on data science competitions and projects. Members pride themselves on being a community where they network, share knowledge, and help each other.

Loher met people at MakerSpace Charlotte and worked with them on development of a machine learning scale model car that could be controlled from a cell phone. Just for fun, they also programmed it to say “Whee! I’m dizzy!” when it spun around three times! There are conscious efforts to make sure projects and ideas are accessible and appealing to new learners who joined the group. For instance, the group developed an algorithm that could create a portrait of itself in the style of famous artists like Picasso or Monet. “Our mission is to get people together to work on interesting things and figure out how to make things better,” said Loher.

He has also led AI discussions for the Charlotte Chamber of Commerce, co-presented with Google on AI at the downtown UNC Charlotte Data Science Initiative building, and partnered with Davidson College and UNCC on cancer-related projects that use AI. In addition to hosting regular group events, the machine learning group has also been a frequent contributor to the NC Science Festival, STEAM events, and helping inspire young children in the sciences. As of last December, the machine learning group formally joined the local nonprofit group, LaunchLKN.org, a collaborative community of entrepreneurs, mentors, and community leaders dedicated to sharing knowledge and motivating growth. The association with LaunchLKN came with meeting space accessibility to Davidson College’s Hurt Hub.

The machine learning group has recently begun a whimsical initiative in collaboration with Davidson College, members of MakerSpace Charlotte, and UNCC students and professors—the development of an autonomous boat. The project is underway through the efforts of several subgroups—artificial intelligence, hull design, renewable energy, application for business, and system integration. The final product will be a boat that delivers ice cream sandwiches to boats on Lake Norman via an air cannon!

To get involved with that effort, or just contact the group, visit their website or visit LaunchLKN.

Bill Giduz

Bill Giduz was the son who followed his father’s footsteps into journalism. He has been involved his entire life with news and photography in schools he attended and jobs he’s held. He believes now that he’s got a few good years left to devote to The News of Davidson.

Support Your Community News