Mar 2, 2018 | Atlanta, GA
Machine learning at Georgia Tech was in the spotlight recently as The Center for Machine Learning at Georgia Tech (ML@GT) hosted its spring seminar on Feb. 22 in the Klaus Advanced Computing Building.
Billed as a “day of discussions around machine learning,” more than 200 students and faculty from across campus registered for the daylong event.
“AI is like going to the moon. Data science is the rocket, but machine learning is the fuel that is propelling us forward,” said ML@GT Director Irfan Essa.
“The field crosses a wide variety of disciplines so Georgia Tech is an ideal setting to build a home for thought leaders and train the next generation in machine learning.”
Algorithms and bias
The day began with an informal discussion over breakfast with Essa and Charles Isbell, executive associate dean and professor in the College of Computing.
“Developers tend to walk around feeling objective because it’s the algorithm that is determining the answer,” Isbell said in response to a question relating to the ethical aspects of machine learning.
“However, they need to ensure no bias is being introduced. The algorithms they create need to be ‘inspectable’ and must be able to explain their answers.”
Following an update from the ML@GT leadership team about current research projects, recent achievements, and plans for the coming year, attendees were treated to lunch and a presentation from Princeton University’s Sanjeev Arora.
Arora, the Charles C. Fitzmorris Professor of Computer Science at Princeton, explored the mysteries of deep learning as he shared his thoughts on generative adversarial nets (GANs) and their efficacy in learning from relatively small data sets. Associate Professor Joelle Pineau from McGill University followed Arora. Her remarks circled back to ethical issues raised by artificial intelligence agents, such as chatbots.
ML@GT’s day of discussion wrapped up with another informal chat session and a reception in the Klaus Atrium.
ML@GT is an interdisciplinary research center launched in July 2016. A machine learning Ph.D. program was approved in June 2017. An inaugural class of approximately 15 students is scheduled to convene for the Fall 2017 semester.