Quantum architecture search (QAS) has emerged to automate the design of high-performance quantum circuits under specific tasks and hardware constraints. We propose a noise-aware quantum architecture search (NA-QAS) framework based on variational quantum circuit design. By incorporating a noise model into the training of parameterized quantum circuits (PQCs) , the proposed framework identifies the noise-robust architectures. We introduce a hybrid Hamiltonian
ε -greedy strategy to optimize evaluation costs and circumvent local optima. Furthermore, an enhanced variable-depth NSGA-II algorithm is employed to navigate the vast search space, enabling an automated trade-off between architectural expressibility and quantum hardware overhead. The effectiveness of the framework is validated through binary classification and iris multi-classification tasks under a noisy condition. Compared to existing approaches, our framework can search for quantum architectures with superior performance and greater resource efficiency under a noisy condition.
Existing quantum architecture search (QAS) methods typically treat quantum data embedding and circuit structures as independent components, leading to inefficient search processes and limited performance. We propose EAQAS, an embedding-aware QAS framework that explicitly models the joint representation of embedding operations and quantum gates using a Hierarchical Variational Graph Autoencoder (H-VGAE) with bi-directional cross-attention fusion. The framework integrates RepCap-based filtering for search space pruning, dual-branch encoders for modality-specific representation learning, and Bayesian optimization in the learned latent space. On binary classification tasks using MNIST and WDBC, EAQAS achieves accuracies of 99.28% and 97.01%, respectively. RepCap filtering at moderate thresholds (
τ=0.3 –0.5) accelerates search by 1.43–1.74× while fully preserving accuracy. Ablation studies confirm that cross-modal reconstruction is the most critical component, with its removal causing a 5.16% average performance drop. This work demonstrates that embedding-aware representation learning can enhance automated quantum circuit design under NISQ-compatible constraints.