Secure Computation with RAMs - Dr. Mariana Raykova, Yale University

Explore secure computation techniques for protecting private data in this comprehensive lecture by Dr. Mariana Raykova from Yale University. Delve into the challenges of hiding memory access patterns in secure computation and discover two innovative approaches addressing this issue. Learn about a new oblivious RAM (ORAM) construction that modifies the classical square-root ORAM to improve efficiency for smaller database sizes. Examine a hybrid encrypted search scheme that reduces leakage in Boolean query solutions by combining ORAM techniques with efficient search index structures. Gain insights into the latest developments in cryptography and security, including secure computation, verifiable computation, and obfuscation. Follow Dr. Raykova's research journey in developing cryptographic techniques that balance data privacy protection with usability and verifiability.

Intro
How to Compute with Private Data?
MPC Applications
What can we compute securely?
Circuits imply linear work
Random Access
Oblivious RAM GOʻ96
Binary Search
ORAM Constructions
Secure Computation with RAMS
Revisiting Square Root ORAM
MPC Bottlenecks
Basic Construction
Dummy Lookups
Sampling Radom Element
Creating Position Map
Inverse permutation
Evaluation
Per-Access Crossover Points
Access time
Initialization cost
Benchmarks
Overview