Rethink the Role of Deep Learning towards Large-scale Quantum Systems

Characterizing the ground state properties of quantum systems is fundamental to capturing their behavior but computationally challenging. Recent advances in AI have introduced novel approaches, with diverse machine learning (ML) and deep learning (DL) models proposed for this purpose. However, the necessity and specific role of DL models in these tasks remain unclear, as prior studies often employ varied or impractical quantum resources to construct datasets, resulting in unfair comparisons. To address this, we systematically benchmark DL models against traditional ML approaches across three families of Hamiltonian, scaling up to $127$ qubits in three crucial ground-state learning tasks while enforcing equivalent quantum resource usage. Our results reveal that ML models often achieve performance comparable to or even exceeding that of DL approaches across all tasks. Furthermore, a randomization test demonstrates that measurement input features have minimal impact on DL models' prediction performance. These findings challenge the necessity of current DL models in many quantum system learning scenarios and provide valuable insights into their effective utilization.

Understanding the fundamental behavior of quantum systems is essential, but it is also computationally challenging. To address this, researchers have turned to artificial intelligence (AI), including both advanced "deep learning" (DL) and simpler "traditional machine learning" (ML). However, it wasn't clear if the advanced DL methods were truly necessary or better, especially since previous comparisons were often unfair. This research conducted a fair head-to-head evaluation. By giving both DL and ML models the same quantum resource to learn from, the study found that simpler ML models often performed just as well, or even better, at predicting quantum system properties. While current simpler ML methods may be more effective for many quantum learning tasks, discovering DL methods that are well-matched to these tasks remains an important direction for future research.