Feature selection, the task of identifying the most informative subset of attributes from a large feature pool, is an NP-hard problem central to Information Retrieval (IR) and Recommender Systems (RS). This paper presents Team SYZYGY’s submission to Task 1 of the QuantumCLEF 2026 Lab, formulating feature selection as a Quadratic Unconstrained Binary Optimization (QUBO) problem solved on a D-Wave quantum annealer. For Task 1A (IR), we introduce five information-theoretic QUBO formulations that organically balance feature relevance and redundancy without hard cardinality constraints, combining mutual information, conditional mutual information, interaction information, and two normalized weighted mutual information variants. On the MQ2007 benchmark, our best configurations reduce the feature count from 46 to just 21 to 24 while surpassing both the full feature and recursive feature elimination baselines. For Task 1B (RS), a simplified collaborative driven QUBO aligns content based similarity with real user behavior, with the simulated annealer successfully improving over the full feature baseline while the physical quantum annealer is hampered by embedding complications. Across both tasks, quantum hardware completes sampling an order of magnitude faster than simulated annealing, though run to run inconsistencies on the physical QPU remain a key open challenge.
Feature selection, the task of identifying the most informative subset of attributes from a large feature pool, is an NP-hard problem central to Information Retrieval (IR) and Recommender Systems (RS). This paper presents Team SYZYGY’s submission to Task 1 of the QuantumCLEF 2026 Lab, formulating feature selection as a Quadratic Unconstrained Binary Optimization (QUBO) problem solved on a D-Wave quantum annealer. For Task 1A (IR), we introduce five information-theoretic QUBO formulations that organically balance feature relevance and redundancy without hard cardinality constraints, combining mutual information, conditional mutual information, interaction information, and two normalized weighted mutual information variants. On the MQ2007 benchmark, our best configurations reduce the feature count from 46 to just 21 to 24 while surpassing both the full feature and recursive feature elimination baselines. For Task 1B (RS), a simplified collaborative driven QUBO aligns content based similarity with real user behavior, with the simulated annealer successfully improving over the full feature baseline while the physical quantum annealer is hampered by embedding complications. Across both tasks, quantum hardware completes sampling an order of magnitude faster than simulated annealing, though run to run inconsistencies on the physical QPU remain a key open challenge.
In this work, we tackle the feature selection problem for content-based recommender systems using Quadratic Unconstrained Binary Optimization (QUBO). Our approach, submitted as Team MALTO to the QuantumCLEF 2025 challenge, aims to improve the performance of an Item-Based K-Nearest Neighbors (Item-KNN) model by selecting a compact and informative subset of item features. We formulate a QUBO objective that combines feature relevance – estimated via Random Forest (RF) importance scores – and feature redundancy – captured through pairwise Pearson correlations. We compare our method against a collaborative-driven QUBO baseline and a random selection strategy. Experiments on the official QuantumCLEF dataset demonstrate that our relevanceaware strategy outperforms the other methods regarding recommendation quality, especially in low-dimensional feature regimes. Our results highlight the potential of combining machine learning and quantum optimization for effective feature selection in recommender systems
The emerging field of Quantum Computing (QC) is attracting considerable research interest due to its potential. It is in fact believed that QC could revolutionize the way we approach complex problems by significantly reducing the time required to solve them. Although QC is still in its early stages of development, certain problems can already be addressed using quantum computers, offering a glimpse into its capabilities. The goal of the QuantumCLEF lab is to raise awareness of QC and to design, develop, and evaluate new QC algorithms aimed at solving challenges typically encountered in the implementation of Information Retrieval (IR) and Recommender Systems (RS). Furthermore, the lab provides a valuable opportunity to engage with QC technologies, which are often difficult to access. In this work, we present an overview of the second edition of QuantumCLEF, a lab focused on applying Quantum Annealing (QA), a specific QC paradigm, to three tasks: Feature Selection for IR and RS systems, Instance Selection for IR systems, and Clustering for IR systems. A total of 44 teams registered for the lab, with 5 teams successfully submitting their runs in accordance with the lab guidelines. Given the novelty of the topics, participants were provided with extensive examples and comprehensive materials to help them understand how QA works and how to program quantum annealers.
Feature selection is crucial for enhancing the accuracy and efficiency of machine learning models. This work investigates the possibility of the usefulness of a quantum annealer to tackle the challenge of selecting features based on maximizing mutual information (MI) and conditional mutual information (CMI). Calculating the optimal feature set for maximum MI and CMI remains computationally intractable for large datasets on classical computers, even with approximation methods. This study employs a Mutual Information Quadratic Unconstrained Binary Optimization (MIQUBO) formulation, enabling its solution on a quantum annealer. Importantly, the study demonstrates the relevancy of this approach in identifying the best feature combinations that maximizes the MI and CMI. To showcase its real-world applicability, we apply MIQUBO to forecasting the price of used excavators. Our results demonstrate that using the MIQUBO approach there is an improvement in the prediction of machine learning models for datasets, with a smaller MI concentration on a subset of all features. • Problem Addressed: We tackle the computationally difficult problem of feature selection for machine learning models, specifically by maximizing mutual information (MI) and conditional mutual information (CMI) between features and the target variable. • Proposed Solution: We employ the adiabatic quantum computation paradigm to tackle this problem. The study utilizes a hybrid annealing approach, specifically the D-Wave Kerberos framework, which combines classical and quantum algorithms to solve the complex optimization problem. The feature selection task is reformulated as a Mutual Information Quadratic Unconstrained Binary Optimization (MIQUBO) problem, which can be solved on a quantum annealer. • Real-World Application: The MIQUBO method is applied to a practical, real-world scenario: forecasting the prices of used excavators. • Key Findings: The research demonstrates that using the MIQUBO approach leads to an improvement in the prediction performance of machine learning models. This is particularly true for datasets where the mutual information is less concentrated on a small subset of features.
Collaborative filtering models generally perform better than content-based filtering models and do not require careful feature engineering. However, in the cold-start scenario collaborative information may be scarce or even unavailable, whereas the content information may be abundant, but also noisy and expensive to acquire. Thus, selection of particular features that improve cold-start recommendations becomes an important and non-trivial task. In the recent approach by Nembrini et al., the feature selection is driven by the correlational compatibility between collaborative and content-based models. The problem is formulated as a Quadratic Unconstrained Binary Optimization (QUBO) which, due to its NP-hard complexity, is solved using Quantum Annealing on a quantum computer provided by D-Wave. Inspired by the reported results, we contend the idea that current quantum annealers are superior for this problem and instead focus on classical algorithms. In particular, we tackle QUBO via TTOpt, a recently proposed black-box optimizer based on tensor networks and multilinear algebra. We show the computational feasibility of this method for large problems with thousands of features, and empirically demonstrate that the solutions found are comparable to the ones obtained with D-Wave across all examined datasets.
The rapid explosion of medical image data poses a significant challenge to reliable disease Classification with high-dimensional feature spaces and intricate inter-feature correlations. In this paper, a new quantum annealing-inspired framework for optimal feature subset selection, specifically tailored to improve medical image Classification. In contrast to traditional filter, wrapper, or embedded techniques that tend to ignore redundancy and feature interactions, our solution poses feature selection as a Quadratic Unconstrained Binary Optimization (QUBO) problem and solves it effectively using D-Wave quantum annealer hardware. The presented approach makes three primary contributions: (i) a QUBO formulation that captures feature importance, redundancy suppression, and interaction synergy simultaneously, (ii) a hybrid quantum-classical pipeline with embedding, decomposition, and chain strength optimization for efficient execution on existing quantum hardware, and (iii) a comprehensive evaluation on three heterogeneous medical imaging datasets—diabetic retinopathy, breast cancer histopathology, and chest X-ray pneumonia. Experimental findings exhibit uniform improvements in Classification performance (up to 12.1%), drastic feature reduction (65–80%), and much reduced computational time over six state-of-the-art classical approaches. With the exposure of clinically relevant features and robust statistical significance across benchmarks, our approach places quantum annealing as an influential and scalable tool for future medical AI applications.
Feature selection is crucial for enhancing the accuracy and efficiency of machine learning (ML) models. This work investigates the utility of quantum annealing for the feature selection process in an ML-pipeline, used for maximizing the mutual information (MI) or conditional mutual information (CMI) of the underlying feature space. Calculating the optimal set of features that maximize the MI or CMI is computationally intractable for large datasets on classical computers, even with approximative methods. This study employs a Mutual Information Quadratic Unconstrained Binary Optimization (MIQUBO) formulation, enabling its solution on a quantum annealer. We demonstrate the capability of this approach to identify the best feature combinations that maximize the MI or CMI. To showcase its real-world applicability, we solve the MIQUBO problem to forecast the prices of used excavators. Our results demonstrate that for datasets with a small MI concentration the MIQUBO approach can provide a significant improvement over MI-only based approaches, dependent on the dimension of the selected feature space.