Search papers, labs, and topics across Lattice.
This study addresses the expressivity-trainability paradox in Quantum Machine Learning (QML) by linking the algebraic dimensions of Parameterized Quantum Circuits (PQCs) to their optimization dynamics through Dynamical Lie Algebras (DLAs). It reveals that the vast capacity of PQCs leads to Barren Plateaus, where gradient landscapes become exponentially flat, resulting in quantum underfitting despite high training accuracy. By embedding group-theoretic geometric priors, the authors propose a "Trainability-by-Design" framework that balances expressivity with trainability, enhancing the optimization landscape for scalable quantum neural networks.
Unstructured quantum architectures risk quantum underfitting, but embedding geometric priors can transform barren plateaus into gradient-rich training landscapes.
As Quantum Machine Learning (QML) transitions toward practical implementation, the field faces a critical architectural bottleneck that challenges the fundamental assumptions of classical statistical learning theory. In classical deep learning, increasing model capacity typically risks overfitting. However, this study advances a counter-intuitive paradigm: unstructured contemporary QML architectures suffer from a profound state of quantum underfitting, driven by the "expressivity-trainability paradox." We demonstrate that the vast Hilbert space capacity of Parameterized Quantum Circuits (PQCs)-traditionally chased as the source of quantum advantage is the direct mathematical cause of Barren Plateaus (BPs), where gradient landscapes become exponentially flat. By synthesizing recent breakthroughs in Dynamical Lie Algebras (DLAs) and Geometric QML, we establish a comprehensive framework linking the algebraic dimension of circuit generators to their optimization dynamics. Furthermore, we empirically validate this framework on a non-linear binary classification task, illuminating a uniquely quantum manifestation of the bias-variance tradeoff: while unstructured architectures achieve near-perfect training accuracy via unscalable parameterization (quantum overfitting), embedding group-theoretic geometric priors acts as a structural regularizer. By restricting the DLA growth to a polynomial regime, our symmetry-preserving approach sacrifices raw memorization capacity to guarantee scalable, gradient-rich training landscapes, offering a robust roadmap for "Trainability-by-Design" in scalable quantum neural networks.
