Your Secrets Are Safe with Julia - A Compiler for Secure Computation
Strange Loop Conference via YouTube
Overview
Explore the world of Homomorphic Encryption (HE) in this 39-minute conference talk from Strange Loop. Dive into Ramparts, a compiler that translates Julia code into a form suitable for HE computation, enabling secure data processing without decryption. Learn how this innovative approach allows programmers and data scientists to leverage HE without compromising their familiar tools. Discover the challenges of HE's unique computation model and how Ramparts addresses them. Examine practical applications in image processing, including image sharpening techniques. Understand the role of symbolic execution, bulk manipulation, and multivalued expressions in this context. Gain insights into the architecture, limitations, and lessons learned from developing Ramparts. Explore the potential of open-source tools and Julia's dynamic features in advancing secure computation.
Syllabus
Introduction
Takeaways
Homomorphic Encryption
Rhythmatic Circuits
Bad Performance
Image Processing
Image Sharpening
Symbolic Execution
Bulk Manipulation
Multivalued Expressions
SymbolicExecution
Water
Open Source
Julia Dynamic Features
Using Crucible
Architecture
Questions
What we learned
No free lunches
Large circuits
Limitations
Grackle
Taught by
Strange Loop Conference