Matthew Hale

Achieving an impressive 88.17% accuracy on CIFAR10 under differential privacy, this framework redefines the balance between data privacy and model performance.

Asynchronous trajectory estimation can slash communication needs by nearly 97% while boosting accuracy in multi-robot systems.

Existing differential privacy methods struggle with symbolic trajectory data, but this new mechanism slashes error by up to 55% on real-world data.

Achieve differentially private Markov chain modeling with minimal impact on accuracy, showing less than 2% error in the stationary distribution under typical privacy settings.