Exponential Quantum Communication Advantage in Distributed Inference and Learning for Large-Scale Machine Learning Models

Distributed quantum circuits can achieve exponential communication advantages over classical methods for inference and gradient-based training of large parameterized machine learning models.