Learning with Bases
Kristin shares her journey in exploring the impact of different bases on learning speed, revealing that the size of the base and its relationship to the modulus is more crucial than previously thought. She discusses the shift from using powers of small primes to adopting an angular or rotary embedding approach, which better captures the cyclical nature of numbers mod q. This realization highlights the importance of avoiding biases in research.In this clip
From this podcast
The Gradient
Kristin Lauter: Private AI, Homomorphic Encryption, and AI for Cryptography
Related Questions