Quantum Feature Engineering for Credit Default Prediction: When and Why IQP Circuits Help Linear Classifiers (arXiv:2609.10505)

Forum topic · 小凯 · 2026-09-11

Summary

A new arXiv paper (2609.10505) by Finkelstein, Legziel Levy, Yakhini, and Cohen investigates whether Instantaneous Quantum Polynomial-time (IQP) circuits can serve as feature engineers for credit default prediction, a tabular classification task where small F1 gains directly reduce financial exposure. The method selects n of the dataset's 23 financial attributes per client, encodes each as a rotation angle in an n-qubit circuit, and extracts 2n expectation values as new features; an n-qubit IQP circuit runs in constant depth while its exact classical simulation scales exponentially. On the UCI Default of Credit Card Clients dataset with five-fold cross-validation, appending 16 IQP features (n = 8 qubits) to Logistic Regression raises F1 from 0.462 to 0.517 (+0.055, p < 0.0001). Kernel PCA, the strongest unsupervised classical non-linear baseline, reaches only 0.493 at the same feature budget, a gap that survives Benjamini-Hochberg correction across 12 tests (p = 0.00007). No other classifier tested (Random Forest, SVM, XGBoost, k-NN) benefits, suggesting a linear-expressivity mechanism rather than generic improvement. Feature selection matters too: Random Forest importance-guided selection achieves F1 = 0.523, while maximally uncorrelated features drop to 0.496, indicating the circuit amplifies informative structure rather than creating it.

Paper Overview

Field: Machine Learning Authors: Menachem Finkelstein, Diana Legziel Levy, Zohar Yakhini, Sarel Cohen arXiv: 2609.10505

Abstract

Credit default prediction is a tabular classification problem in which modest gains in F1 translate directly into reduced financial exposure. We ask whether Instantaneous Quantum Polynomial-time (IQP) circuits can produce features that improve a classifier over both its raw classical baseline and Kernel PCA - the strongest unsupervised classical non-linear alternative - at an equal feature budget. The dataset provides 23 financial attributes per client; for an n-qubit circuit we select n of them, encode each as a rotation angle, and read 2n expectation values back out as new features. The motivation for using a quantum circuit is computational: an n-qubit IQP circuit runs in constant depth and encodes feature correlations in a 2^n-dimensional Hilbert space, whereas classical simulation of its exact output statistics scales exponentially in n. Using the UCI Default of Credit Card Clients dataset and five-fold cross-validation, we find that appending 16 IQP features (n = 8 qubits) to a Logistic Regression model raises F1 from 0.462 to 0.517 (+0.055, p < 0.0001). Kernel PCA, the next-best method, reaches only 0.493 at the same feature count; the gap survives Benjamini-Hochberg correction across 12 tests (p = 0.00007). No other classifier - Random Forest, SVM, XGBoost, or k-NN - benefits, which points to a linear-expressivity mechanism rather than a generic improvement. We also show that how the 8 input features are chosen matters: Random Forest importance-guided selection reaches F1 = 0.523, while encoding maximally uncorrelated features drops it to 0.496, demonstrating that the circuit amplifies informative structure rather than creating it from scratch.

Key Takeaways

Appending 16 IQP-circuit features (n = 8 qubits) to Logistic Regression improves F1 from 0.462 to 0.517 (+0.055, p < 0.0001) on the UCI Default of Credit Card Clients dataset.
Kernel PCA, the strongest unsupervised classical non-linear baseline, only reaches F1 = 0.493 at the same feature budget; the gap survives Benjamini-Hochberg correction across 12 tests (p = 0.00007).
Only linear classifiers benefit - Random Forest, SVM, XGBoost, and k-NN show no gains - pointing to a linear-expressivity mechanism rather than a universal quantum advantage.
Feature selection is critical: importance-guided selection yields F1 = 0.523, while maximally uncorrelated features drop performance to 0.496.
The quantum motivation is computational: constant-depth n-qubit IQP circuits encode correlations in a 2^n-dimensional Hilbert space that is exponentially costly to simulate classically.