Track the frontier, from condensed matter to cosmology.
Search papers, labs, and topics across Lattice.
This paper introduces a hybrid quantum neural network (QNN) framework combining variational quantum circuits and quantum annealing, implemented on a neuromorphic quantum computing platform, to forecast PM$_{2.5}$ air pollution. The QNN demonstrates superior resilience to high-entropy environmental data compared to classical MLPs and LSTMs, achieving state-of-the-art predictive fidelity when augmented with discrete wavelet transforms. The work establishes a benchmark for quantum utility by successfully scaling to 172 qubits and addressing a complex, non-linear real-world problem.
Quantum annealing and variational circuits on a 172-qubit neuromorphic platform achieve state-of-the-art air pollution nowcasting, outperforming classical models on raw, high-entropy data.
Air pollution poses a significant threat to public health, exacerbated by the rapid industrialization and urbanization of modern society. While traditional deep learning offers predictive capabilities, recent advancements in quantum machine learning (QML) present a compelling alternative. Here, we introduce a scalable hybrid framework that integrates variational quantum circuits with quantum annealing. Leveraging a neuromorphic quantum computing platform to transcend the qubit limitations of the NISQ era, we model particulate matter (PM$_{2.5}$) dynamics using high-density urban data from the UK. We demonstrate that our Quantum Neural Networks (QNNs) exhibit an intrinsic resilience to input volatility, significantly outperforming classical baselines (MLP and LSTM) on raw, high-entropy environmental data. Furthermore, we successfully scaled the model to 172 qubits, observing a synergistic enhancement when coupled with wavelet decomposition that achieves state-of-the-art predictive fidelity. These findings establish a critical benchmark for quantum utility, proving that scalable, neuromorphic quantum machine learning can now address complex, non-linear real-world problems.