Rajaram's project also involved examining an open-source model that can read text from images, which he found to be vulnerable to specific adversarial attacks. He was able to isolate and remove the model's text reading capability, thereby protecting it against specific failures. Rajaram, who plans to continue studying computer science at MIT in the fall, beat out over 2,000 competitors to win the prestigious award.
Key takeaways:
- Achyuta Rajaram won the Regeneron Science Talent Search $250,000 top prize for his work on making machine learning more efficient and safer.
- Rajaram's project focused on improving the speed and accuracy of computer vision models, potentially making AI smarter and faster.
- He built an algorithm to automate the process of finding specific circuits in a computer's brain, leading to a smarter, faster model.
- Rajaram plans to continue studying computer science at MIT in the fall and encourages others to be curious about everything.