Determine the generality of the observed scaling behavior

Determine whether the observed non-monotonic circuit-size, accuracy, and generalization behavior of Q-FLAIR's adaptive growth mechanism generalizes to other high-dimensional inputs and to Q-FLAIR's quantum support-vector-machine variant.

Background

The experiments use a single task: full-resolution 784-pixel MNIST 3-versus-5 classification. The authors report that circuit size and test accuracy do not vary monotonically with training-set size, but they do not establish whether this behavior is specific to that dataset and classifier setting. They explicitly identify its applicability to other high-dimensional inputs and to Q-FLAIR's QSVM variant as unknown.

References

We do not know whether these observations generalize to other high-dimensional inputs, or to Q-FLAIR's QSVM variant.

— Toward Joint Optimization of Circuit Depth and Training Data Size in Adaptively Grown Quantum Classifiers  (2610.12428 - Nowmi et al., 8 Oct 2026) in Section 6, Limitations