Paper: arXiv 2306.12965
Authors: Sohum Thakkar, Skander Kazdaghli, Natansh Mathur, Iordanis Kerenidis, André J. Ferreira-Martins, Samurai Brito
Abstract
Quantum algorithms have the potential to enhance machine learning across a variety of domains and applications. In this work, we show how quantum machine learning can be used to improve financial forecasting. First, we use classical and quantum Determinantal Point Processes to enhance Random Forest models for churn prediction, improving precision by almost 6%. Second, we design quantum neural network architectures with orthogonal and compound layers for credit risk assessment, which match classical performance with significantly fewer parameters. Our results demonstrate that leveraging quantum ideas can effectively enhance the performance of machine learning, both today as quantum-inspired classical ML solutions, and even more in the future, with the advent of better quantum hardware.
Complexity vs Empirical Score
- Math Complexity: 7.5/10
- Empirical Rigor: 4.0/10
- Quadrant: Lab Rats — high math complexity, limited empirical testing
Why this score: The paper introduces advanced mathematical concepts like Determinantal Point Processes and quantum orthogonal neural networks, requiring significant theoretical background. While it includes real-world data and reported metrics (6% precision improvement), the implementation details are high-level, the quantum hardware performance is limited by noise, and the primary focus remains on theoretical quantum-inspired algorithms rather than fully realized, backtest-ready models.
Research Flowchart
flowchart TD A["Research Goal: Enhance Financial Forecasting<br>using Quantum Machine Learning"] --> B["Data: Financial Datasets<br>Churn Prediction & Credit Risk"] B --> C["Method 1: Quantum-Inspired DPP-Random Forest"] B --> D["Method 2: Quantum Neural Networks<br>Orthogonal & Compound Layers"] C --> E["Computational Process:<br>Classical/Quantum Determinantal Point Processes"] D --> F["Computational Process:<br>Quantum-Inspired Architecture Optimization"] E --> G["Outcome 1: Churn Prediction<br>Precision improved by ~6%"] F --> H["Outcome 2: Credit Risk Assessment<br>Matches classical performance<br>with significantly fewer parameters"] G --> I["Key Finding: Quantum ideas effectively<br>enhance ML via quantum-inspired<br>classical solutions and future hardware"] H --> I
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a DPP-enhanced Random Forest model for churn prediction and quantum neural network architectures with orthogonal and compound layers for credit risk assessment.
Source sentence
we use classical and quantum Determinantal Point Processes to enhance Random Forest models for churn prediction, improving precision by almost 6%. Second, we design quantum neural network architectures with orthogonal and compound layers for credit risk assessment, which match classical performance with significantly f
Assumes 3
- The data split into training and test sets is based on time period and assumed to have no significant covariate shift. data
Source sentence
split into training and test sets based on time period, with 130,000 datapoints being set aside as the test set and 174,000 datapoints used for training. The data was split in a way that did not produce any significant covariate shift between the train and test sets.
- Resource limitations require flagging only a relatively small number of customers with high confidence. execution
Source sentence
resource limitations make it necessary to flag a relatively small number of customers with high confidence.
- Quantum hardware errors increase with larger matrix dimensions, limiting performance of quantum DPP sampling. model
Source sentence
as the dimensions grow bigger, the samples from the quantum DPP circuits lead to worse classifier performance. This highlights the limitations of the available quantum hardware, which is prone to errors.
Data
- Proprietary churn dataset other proprietary
Source sentence
The primary datasets and model used in this study are proprietary to Itau and QC Ware and are not publicly available.
- OpenML classification datasets benchmark public
Source sentence
We further benchmarked our model on various classification datasets from OpenML.
Costs
Not applicable
Finds 4
- The DPP-Random Forest captures a slightly higher percentage of total withdrawals (61.42% to 62.77%). performance
Source sentence
% total withdrawals captured 61.42% 62.77%
- Quantum DPP sampling degrades classifier performance for larger matrix dimensions due to hardware errors. negative
Source sentence
as the dimensions grow bigger, the samples from the quantum DPP circuits lead to worse classifier performance. This highlights the limitations of the available quantum hardware, which is prone to errors.
- Quantum compound neural network models have far fewer parameters and achieve better accuracy and generalization than classical fully‑connected neural networks. performance
Source sentence
the quantum compound neural network models both have far fewer parameters and achieve better accuracy and generalization than classical fully-connected neural networks.
- Current quantum hardware is not powerful enough to provide real improvements; performance gains can be achieved today by turning quantum ideas into classical ML solutions. theoretical
Source sentence
In the present day, quantum hardware is not powerful enough to provide real improvements or conclusive large‑scale benchmarks. Performance enhancements can be achieved today by turning these quantum ideas into classical ML solutions run on GPUs.
Checked by others
Supported by
- Hybrid Quantum Neural Networks with Amplitude Encoding: Advancing Recovery Rate Predictions
Compare the findings
This paper: Current quantum hardware is not powerful enough to provide real improvements; performance gains can be achieved today by turning quantum ideas into classical ML solutions.
That paper: The model’s practical applicability is limited by the need for deep circuits on real quantum hardware, which are more prone to noise and decoherence.
- Quantum computer based Feature Selection in Machine Learning
Compare the findings
This paper: Current quantum hardware is not powerful enough to provide real improvements; performance gains can be achieved today by turning quantum ideas into classical ML solutions.
That paper: QAOA on actual quantum hardware achieves decent early results but does not yet match non‑quantum metaheuristics.
7 papers in the archive : PO-QA: A Framework for Portfolio Optimization using Quantum …
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.