About August Deer
August studied federated learning, which is a way to train machine learning algorithms that rely on users’ private data. His improvements to federated learning may help keep private data more secure and make machine learning systems more resilient to some types of malicious attacks.
On Multi-Round Privacy in Federated LearningView Poster
August Deer, 18, of Venice, helped make distributed learning more secure for his Regeneron Science Talent Search computer science project. Developers of popular tech devices like voice assistants and recommendation algorithms hope to one day employ a mathematical technique called federated learning. This will allow companies to improve services by utilizing the users’ private data without sharing it outright. Currently, even with federated learning, a malicious attacker might reverse engineer the process to acquire that data. August helped develop an improvement to federated learning to ensure that such attacks are made much more difficult. His key contribution was modifying an existing algorithm to work well even if a user’s data suddenly became unavailable. While his theoretical research still requires real-world testing, he believes it has the potential to protect people’s privacy while they use the web-based services they enjoy.
August attends Geffen Academy at UCLA in Los Angeles. He is active in theater and plays classical piano. He also takes part in the UCLA Olga Radko Endowed Math Circle where he teaches advanced math to middle schoolers. The son of Laurie and Jonathan Deer, August is planning a career in mathematical research.
Beyond the Project
August loves the collaborative spirit of his school’s theater program. It has allowed him to act, write and produce music. His next project is co-directing a tribute to Monty Python.
FUN FACTS: August is a junior member of the Academy of Magical Arts at the Magic Castle, an organization promoting the art of magic and training magicians. He is developing a 20-minute act.