The new Center for Secure and Trusted Machine Learning will be part of the university’s Viterbi School of Engineering, USC and Amazon announced Jan. 28.
Researchers from both the university and the company will work alongside one another at the center, which will support projects focused on privacy protection within machine learning systems. The center will also offer annual fellowships to doctoral students at the university.
“Creating such mutually beneficial partnerships between academia and industry will help guide the development of a field as important and as rapidly changing as machine learning and artificial intelligence,” Viterbi School Dean Yannis C. Yortsos said in a statement.
Prem Natarajan, vice president of Amazon’s Alexa AI, said “customer trust” was a top priority for the company, and that research into privacy protection would en-sure the company would be able to maintain “the highest possible standards of security and privacy when handling customer data.”
Amazon came under fire from consumer privacy advocates in 2019 when Bloomberg News reported that the company was actively transcribing audio recordings of customers using its voice-activated Alexa system.
The company later announced user features that would allow customers to automatically delete recordings picked up by the artificial intelligence system.
“We are delighted to bring together top talent at Amazon and USC in a joint mission to drive groundbreaking advances in privacy and security preserving machine learning," Natarajan said in a statement.
Amazon and USC say the center will host an annual symposium on artificial intelligence, as well as workshops and recruiting events for current and prospective students.
The center will be overseen by Salman Avestimehr, director of Viterbi’s Information Theory and Machine Learning research lab. Avestimehr said the center would provide an “exciting opportunity” for “university-industry collaboration” in the study of key security issues around the development of machine learning and artificial intelligence.
For reprint and licensing requests for this article, CLICK HERE.