🤖 AI Summary
Training binary neural networks (BNNs) is notoriously difficult, and joint hyperparameter and architecture search incurs prohibitive computational cost. Method: This paper proposes a quantum hypernetwork framework that— for the first time—unifies optimization of BNN weights, hyperparameters, and network architecture within quantum superposition states. It employs variational quantum circuits for end-to-end quantum machine learning, integrating quantum superposition encoding with measurement-driven optimization, and validates the approach via classical simulation. Contribution/Results: A critical finding is the existence of an optimal quantum circuit depth that significantly boosts the sampling probability of high-performance BNN configurations. Experiments on a 2D Gaussian dataset and a simplified MNIST task demonstrate the paradigm’s efficacy: it discovers high-performing binary models with substantially lower computational overhead than conventional combinatorial search, achieving both efficiency and high success probability. This work establishes a novel quantum-enhanced paradigm for lightweight BNN design.
📝 Abstract
Binary neural networks, i.e., neural networks whose parameters and activations are constrained to only two possible values, offer a compelling avenue for the deployment of deep learning models on energy- and memory-limited devices. However, their training, architectural design, and hyperparameter tuning remain challenging as these involve multiple computationally expensive combinatorial optimization problems. Here we introduce quantum hypernetworks as a mechanism to train binary neural networks on quantum computers, which unify the search over parameters, hyperparameters, and architectures in a single optimization loop. Through classical simulations, we demonstrate that of our approach effectively finds optimal parameters, hyperparameters and architectural choices with high probability on classification problems including a two-dimensional Gaussian dataset and a scaled-down version of the MNIST handwritten digits. We represent our quantum hypernetworks as variational quantum circuits, and find that an optimal circuit depth maximizes the probability of finding performant binary neural networks. Our unified approach provides an immense scope for other applications in the field of machine learning.
