AI-Based Computing and Efficient Systems (ACES) Lab

Research Areas

ACES Lab's current research addresses several aspects of secure and efficient computing, with a focus on robust machine learning under resource constraints, hardware and system security, AI-based optimization, intellectual property (IP) protection, as well as cryptographically secure privacy-preserving computing.

Our lab has made foundational contributions to all of these domains, including invention of the concepts of logic obfuscation/logic locking for both sequential and combinational logic that enabled the first active unique control of chips post-silicon, novel statistical methods for assessing chip security, the first automated co-design of data/algorithm/software/hardware for energy-efficient and compact deep learning, invention of co-design methods for creation of low-overhead robust and safe AI, the first trigger inversion methodology for detection of AI poisoning attacks, creation of the first method for watermarking/tracing deep learning models and generative output and attestation on trusted platforms, AI-based generic solvers for combinatorial optimization with constraints, creation of physical proofs of provenance, as well as novel co-design and optimization of algorithm and hardware/software, including co-design with cryptographic constructs for privacy-preserving computing.

Our work has linked seemingly disparate fields of logic synthesis and cryptographically secure computing requiring Boolean logic/vector computing, creating many new opportunities for researchers and leading to practical methods for managing function nonlinearities in the ciphertext domain.

Contact

Funding Acknowledgements