Poster Session 1

Paper ID: 30

Authors: Zhen Yang, Shan Jin, Yajie Hao, Guangwei Deng, Xiu-Hao Deng, Re-Bing Wu and Xiaoting Wang

Title: Pareto-Optimal Floquet Waveform Design for Dynamical Sweet Spots in Superconducting Qubits

Abstract: In superconducting systems, periodic flux modulation can create dynamical sweet spots (DSSs) that strongly suppress dephasing from low-frequency flux noise, but improving pure dephasing often comes with a competing penalty in energy relaxation. Here, we develop a waveform-agnostic Floquet framework in which the periodic flux drive is fully parameterized by trainable Fourier components and optimized in a multi-objective manner to expose the Pareto-optimal trade-off between $T_1$ and $T_\phi$ under realistic noise (combined $1/f$ flux noise and dielectric loss). For a fluxonium qubit with experimentally relevant parameters, the computed Pareto front contains operating points that improve $T_\phi$ by approximately 3-5 times compared with standard single-/two-tone DSS recipes, while maintaining $T_1$ in the few-hundred-microsecond regime. Beyond numerical optimization, we establish analytic upper bounds on achievable $T_1$ under arbitrary periodic modulation, showing that DSS operation cannot drive the relaxation rate arbitrarily close to zero. The Pareto set also reveals “double-DSS” operating bands that are first-order insensitive to both DC flux bias and AC amplitude fluctuations, providing robust experimental working regions. As a application, we design single- and two-qubit gates at these working points and simulate open-system dynamics, obtaining process fidelities of 99.9993% for a 10ns X-gate and 99.995% for a 28ns $\sqrt{iSWAP}$ gate. This provides a general route to coherence- and gate-aware DSS engineering via Pareto-optimal Floquet waveform design (with details in arXiv:2601.19209).

Paper ID: 58

Authors: Shuixin Xiao, Xiangyu Wang, Yuanlong Wang, Zhibo Hou, Jun Zhang, Ian R. Petersen, Wen-Zhe Yan, Hidehiro Yonezawa, Franco Nori, Guo-Yong Xiang and Daoyi Dong

Title: Unified formalism and adaptive algorithms for optimal quantum state, detector and process tomography

Abstract: Quantum tomography is a standard technique for characterizing, benchmarking and verifying quantum systems/devices and plays a vital role in advancing quantum technology and understanding the foundations of quantum mechanics. Achieving the highest possible tomography accuracy remains a central challenge. Here we unify the infidelity indices for quantum state, detector and process tomography in a single index $1-F(\hat S,S)$, where $S$ represents the true density matrix, positive operator-valued measure (POVM) element, or process matrix, and $\hat S$ is its estimator. We establish a sufficient and necessary condition for any tomography protocol to attain the optimal scaling $1-F= O(1/N) $ where $N$ is the number of state copies consumed, in contrast to the $O(1/\sqrt{N})$ worst-case scaling of static methods. Guided by this result, we propose adaptive algorithms with provably optimal infidelity scalings for state, detector, and process tomography. Numerical simulations and quantum optical experiments validate the proposed methods, with our experiments reaching, for the first time, the optimal infidelity scaling in ancilla-assisted process tomography. This work has potential applications in quantum device calibration, quantum precision measurement, quantum metrology and quantum engineering.

Paper ID: 60

Authors: Guangpu Wu, Shibei Xue, Guofeng Zhang, Rebing Wu, Min Jiang and Ian Petersen

Title: Model Reduction for Augmented Model of Linear Non-Markovia Quantum Systems

Abstract: An augmented system model provides an effective way to model non-Markovian quantum systems, which is useful in ffltering and control for this class of systems. However, since a large number of ancillary quantum oscillators representing internal modes of a non-Markovian environment directly interact with the principal system in these models, the dimension of the augmented system may be very large causing signiffcant computational burden in designing fflters and controllers. In this context, this paper proposes an H2 model reduction method for the augmented model of linear non-Markovian quantum systems. We first establish necessary and sufffcient conditions for the physical realizability of the augmented model of linear non-Markovian quantum systems, which are more stringent than those for Markovian quantum systems. However, these physical realizability conditions of augmented system model pose non-convex constraints in the optimization problem of model reduction, which makes the problem different from the corresponding classical model reduction problem. To solve the problem, we derive necessary conditions for determining the input matrix in the reduced model, with which a theorem for designing the system matrix of the ancillary system in the reduced system is proved. Building on this, we convert the nonlinear equality constraints into inequality constraints so that a semideffnite programming algorithm can be developed to solve the optimization problem for model reduction. A numerical example of a two-mode linear quantum system driven by three internal modes of a non-Markovian environment validates the effectiveness of our method.

Paper ID: 61

Authors: Guangpu Wu, Shibei Xue, Yuting Zhu, Guofeng Zhang and Ian Petersen

Title: Optimal filtering for a giant cavity in waveguide QED systems

Abstract: In waveguide QED systems, a giant cavity can be engineered to interact with quantum fields by multiple distant coupling points so that its non-Markovian dynamics are quite different from traditional quantum optical cavity systems. Towards feedback control this system, this paper designs an optimal filter for the giant cavity systems to estimate its state evolution under continuous quantum measurements. Firstly, the Langevin equation in the Heisenberg picture are derived, which is a linear continuous-time system with both states and inputs delays resulting from the unconventional distant couplings. Compared to existing modeling approaches, this formulation effectively preserves the nonlocal coupling and multiple delay dynamic characteristics inherent in the original system. In particular, the presence of coupling and propagation delays leads to noncommutativity among the system operators at different times, which prevents the direct application of existing quantum filtering methods. To address this issue, an optimal filter is designed, in which the delayed-state covariance matrices are computed. By iteratively evaluating the delayed-state covariance over successive time intervals, the resulting optimal fflter can be implemented in an interval-wise forward recursion algorithm. Finally, numerical simulations are conducted to evaluate the tracking performance of the proposed optimal filter for the giant cavity. By comparing between the evolutions of Wigner functions of coherent and cat states and the filter, the effectiveness of the optimal filter is validated.

Paper ID: 63

Authors: Tommaso Grigoletto, Francesco Ticozzi and Lorenza Viola

Title: Exact Model Reduction for Continuous-Time Open Quantum Dynamics

Abstract: We consider finite-dimensional many-body quantum systems described by time-independent Hamiltonians and Markovian master equations, and present a systematic method for constructing smaller-dimensional, reduced models that exactly reproduce the time evolution of a set of initial conditions or observables of interest. Our approach exploits Krylov operator spaces and their extension to operator algebras, and may be used to obtain reduced linear models of minimal dimension, well-suited for simulation on classical computers, or reduced quantum models that preserve the structural constraints of physically admissible quantum dynamics, as required for simulation on quantum computers. Notably, we prove that the reduced quantum-dynamical generator is still in Lindblad form. By introducing a new type of observable-dependent symmetries, we show that our method provides a non-trivial generalization of techniques that leverage symmetries, unlocking new reduction opportunities. We quantitatively benchmark our method on paradigmatic open many-body systems of relevance to condensed-matter and quantum-information physics. In particular, we demonstrate how our reduced models can quantitatively describe decoherence dynamics in central-spin systems coupled to structured environments, magnetization transport in boundary-driven dissipative spin chains, and unwanted error dynamics on information encoded in a noiseless quantum code.

Paper ID: 65

Authors: Nico Meyer, Christopher Mutschler, Andreas Maier and Daniel Scherer

Title: ML-Driven Quantum Error Correction: Noise-Tailored Codes and Logical Gates for Early Fault Tolerance

Abstract: We show how machine learning can design quantum error-correcting codes and their logical gates to match the actual noise seen on hardware. The approach discovers non-additive codes tailored to structured noise and combines them via concatenation to amplify protection. In simulated and hardware tests, these learned codes reduce overhead by up to two orders of magnitude under structured noise and achieve beyond break-even performance on IBM and IQM devices. The approach also learns fault-tolerant logical gate sets for both non-additive and additive codes, enforcing transversality when needed, and is validated by rediscovering operations for standard stabilizer codes. Finally, code and gate constructions are co-designed, yielding noise-tailored codes that admit restricted transversal operations. As a concrete outcome, the approach yields variational codes that enable early fault-tolerant implementations of IQP circuits, a class believed hard to sample from. Together, this positions ML-driven QEC as a practical route to early fault tolerance, unifying code discovery, concatenation, and logical gate learning in one toolkit.

Paper ID: 69

Authors: Tobias Kiermeyer, Thomas Heydenreich, Leo Van Damme, Sebastian Hohenemser, Florian Marquardt and Steffen Glaser

Title: Uncovering Latent Structures in Robust Pulse Design: A Physics-Informed Neural Network Approach for Efficient Optimal Control

Abstract: Optimal control of quantum systems requires tailored pulses that maintain high fidelity under realistic experimental imperfections. Standard gradient-based methods such as GRAPE solve each optimization instance independently, discarding information between runs and requiring costly reinitialization when system parameters change. Physics-informed neural networks (PINNs) offer an alternative by embedding the system Hamiltonian directly into the training loss, enabling a single network to learn control strategies across a large parameter space — unlike supervised approaches, no pre-computed training data from any optimal control solver is required. We apply this framework to generate robust, constant-amplitude single-qubit rotation gates across a parameter space spanning pulse duration, rotation angle, frequency detuning, and amplitude inhomogeneity — the latter two representing the dominant experimental imperfections. Trained on a single consumer GPU for 5 days, the network produces a pulse in milliseconds at inference, generating a full library of 100,000 pulses in under 9 seconds—compared to multi-seed GRAPE requiring over 6 days on a high-end server CPU (128 cores)—with neither method consistently outperforming the other in fidelity across the entire parameter space. Structural analysis reveals that both methods discover structured phase profiles with distinct symmetry properties, confirming that these solutions are a genuine feature of the control landscape rather than an artifact of either method. While the network discovers these structures as a natural consequence of generalization across the parameter space, we independently identified similar phase profiles in GRAPE solutions through rotation-angle-dependent symmetry arguments and multiple runs with different initial seeds. More broadly, the framework is highly adaptable, with Hamiltonians, system size, and output representation being straightforward to modify.

Paper ID: 70

Authors: Stijn De Backer, Luis E.C. Rocha, Jan Ryckebusch and Koen Schoors

Title: Characterizing long-term financial return distributions through discrete-time quantum walks

Abstract: With the prospect of advanced quantum computation technologies, we investigate the discrete-time quantum walk as a framework for modeling the evolution of financial asset prices. In particular, it offers a promising approach to model long-term financial return distributions. For large time scales of the order of two trading years, the anticipated Gaussian behavior of the returns often does not emerge, and their distributions often exhibit a high level of asymmetry and bimodality. These features are inadequately captured by the majority of classical models to address financial time series and return distributions. We use a model based on the discrete-time quantum walk to fit empirical long-term return distributions displaying asymmetry and bimodality. The quantum walk distinguishes itself from a classical diffusion process by the occurrence of interference effects, which allows for the generation of bimodal and asymmetric probability distributions. By capturing the broader trends and patterns that emerge over extended periods, the quantum walk model complements traditional short-term models and offers opportunities to more accurately describe the probabilistic structure underlying long-term financial decisions. Integrating these results into option pricing and risk assessment can significantly improve the evaluation of long-term financial uncertainty.

References: De Backer, S., Rocha, L. E., Ryckebusch, J., & Schoors, K. (2025). On the potential of quantum walks for modeling financial return distributions. Physica A: Statistical Mechanics and its Applications, 657, 130215. De Backer, S., Rocha, L. E., Ryckebusch, J., & Schoors, K. (2026). Characterizing asymmetric and bimodal long-term financial return distributions through quantum walks. The European Physical Journal B, in press. arXiv:2505.13019.

Paper ID: 75

Authors: Henrik Glavind Clausen

Title: A Computationally Efficient Filter for Parameter Estimation in Continuously Monitored Quantum Systems

Abstract: In this work, we consider the problem of estimating unknown static or time-varying parameters in quantum systems subject to continuous monitoring. Using the framework of quantum filtering and stochastic master equations (SMEs), which cover repeated measurements in discrete-time as well as both diffusive and jump measurements in continuous-time, we formulate the problem as a joint state-parameter quantum filtering problem from which we derive an approximate filter based on a Gaussian approximation of the true filtering density for the parameter. The resulting recursive algorithm, relying on propagation of the associated sensitivity equations of the SME, is computationally efficient compared to state-of-the-art methods of similar generality and requires only an initial prior distribution and, optionally, a dynamic model for the possibly time-varying parameter. We demonstrate the performance of the proposed filter through simulations of a spin chain subject to sequential projective measurements, a two-level system subject to continuous homodyne detection, and a $\Lambda$-type system subject to photon counting. Compared against a fine-grained discretization of the exact filter, the simulations show that the proposed filter generally approximates the true filtering density well in all three examples, both in the case of time-varying parameters and multiple static parameters, with the covariance providing a meaningful quantification of uncertainty in the estimates.

Paper ID: 14

Authors: Alan Daleth Hernandez Barreto, Maria Fernanda Velasco Campos, Karol Yenaro Morales Escobar and Jose Luis Perez Estudillo

Title: Nanosecond-Scale Neuromorphic Processing Kernels for Quantum Error Correction: A Hardware-Software Co-Design

Abstract: Real-time Quantum Error Correction (QEC) imposes strict timing constraints that challenge the integration of classical decoding algorithms into scalable control electronics. While approaches such as Minimum Weight Perfect Matching (MWPM) achieve high accuracy, their hardware realization often entails non-deterministic latency and significant resource consumption. This paper proposes a hardware-software co-design of a neuromorphic processing kernel based on Spiking Convolutional Neural Networks (ConvSNNs) for real-time QEC. By leveraging a "temporal folding" architecture, the spatiotemporal syndrome volume is compressed into a 2.5D feature map, bypassing the memory overhead of standard 3D convolutions. We introduce an optimized hardware implementation of Leaky Integrate-and- Fire (LIF) neurons utilizing a subtractive reset mechanism and static pre-activation batch normalization, achieving zero-cost stabilization in hardware. Furthermore, a hardware-aware calibration strategy is implemented to adapt neuronal firing thresholds to the code distance. This calibration optimizes the trade-off between noise suppression and signal sensitivity, reducing network switching activity to approximately 18% for enhanced energy efficiency. For a code distance of d = 3, the synthesized kernel utilizes 181 LUTs (< 2% of the device) [3], offering a deterministic and compact solution suitable for first-line decoding in cryogenic quantum control systems.

Paper ID: 20

Authors: Amitav Krishna

Title: Frobenius Normalization Enables Stable Training for Quantum State Denoising

Abstract: Quantum computers offer asymptotic speedups for problems like molecular simulation and integer factorization, but noise limits their near-term utility. Quantum Error Correction can address this, but the number of extra qubits required makes it infeasible for near-term use while the cost per qubit remains high. Quantum Error Mitigation is a nearterm alternative, but only recovers expectation values, not the full quantum state needed for extracting accurate results from simulations run on quantum hardware. For these, we need quantum state reconstruction. Quantum state tomography estimates a state by measuring the system across many bases, but these measurements are inherently noisy; quantum state reconstruction recovers a clean density matrix from this noisy data. Neural networks have shown promise here, but previous work has only been demonstrated up to 5 qubits. We find that Frobenius normalization (scaling inputs to unit purity) removes this bottleneck, enabling scaling to 8-qubit systems while maintaining a large improvement in fidelity.

Paper ID: 21

Authors: Ece Yurtseven

Title: Parallel Multi-Circuit Quantum Feature Fusion in Hybrid Quantum-Classical Convolutional Neural Networks for Breast Tumor Classification

Abstract: Quantum machine learning has emerged as a promising approach to improve feature extraction and classification tasks in high-dimensional data domains such as medical imaging. In this work, we present a hybrid Quantum-Classical Convolutional Neural Network (QCNN) architecture designed for the binary classification of the BreastMNIST dataset, a standardized benchmark for distinguishing between benign and malignant breast tumors. Our architecture integrates classical convolutional feature extraction with two distinct quantum circuits: an amplitude-encoding variational quantum circuit (VQC) and an angle-encoding VQC circuit with circular entanglement, both implemented on four qubits. These circuits generate quantum feature embeddings that are fused with classical features to form a joint feature space, which is subsequently processed by a fully connected classifier. To ensure fairness, the hybrid QCNN is parameter-matched against a baseline classical CNN, allowing us to isolate the contribution of quantum layers. Both models are trained under identical conditions using the Adam optimizer and binary cross-entropy loss. Experimental evaluation in five independent runs demonstrates that the hybrid QCNN achieves statistically significant improvements in classification accuracy compared to the classical CNN, as validated by a onesided Wilcoxon signed rank test (p = 0.03125) and supported by large effect size of Cohen’s d = 2.14. Our results indicate that hybrid QCNN architectures can leverage entanglement and quantum feature fusion to enhance medical image classification tasks. This work establishes a statistical validation framework for assessing hybrid quantum models in biomedical applications and highlights pathways for scaling to larger datasets and deployment on near-term quantum hardware.

Paper ID: 24

Authors: Mahdi Esmaeili, Katayoun Emadzadeh, M. Hossein Eskandari, Muhamad. B Barfar, Saeed Hajihosseini, Ahmad Salmanogli, Hesam Zandi and Abolfazl Eskandari

Title: Co-Design of 2-qubit coupled with cavity bus with QGHNN for High Separate Fidelity and Maximum Grover Algorithm Accuracy

Abstract: A hybrid physics–machine-learning framework is introduced for the joint optimization of separate fidelity and Grover algorithm performance in superconducting quantum processors. The method combines a Hamiltonian-based open-quantum-system model of a two-qubit architecture coupled through a bus resonator with a QGHNN. Readout performance is evaluated via the Lindblad master-equation simulation of the dispersive measurement process, in which state-dependent resonator dynamics are mapped to realistic I/Q distributions and classified to estimate single-shot measurement fidelity. In parallel, algorithmic performance is assessed by simulating a two-qubit Grover search circuit in the presence of dissipation and dephasing, using effective cross-Kerr interactions derived from the underlying circuit parameters. A physically constrained dataset is constructed by sampling of some critical variables relevant to system. The QGHNN is trained to learn the nonlinear relationship between device parameters and the dual objectives of readout (separation) fidelity and Grover success probability. Gradient-based optimization is then carried out directly in parameter space using the trained network, yielding operating points that simultaneously improve measurement distinguishability and Grover algorithm accuracy. Simulation results demonstrate that the readout separation fidelity exceeds 99.4%, while the Grover success probability reaches approximately 86%, and this is comparable with the accuracy achieved by IBM Q5 Tenerife processor. Finally, the optimized parameters are used to implement the quantum circuit layout in Qiskit Metal, and their validity is further confirmed through EPR-based certification.

Paper ID: 26

Authors: Javier Vecino Peñas, Ana Fernández-Vilas, Rebeca Pilar Díaz-Redondo and Sergio Gándara Gándara

Title: Purification Strategy Optimization for Entanglement Routing in Quantum Networks

Abstract: Quantum networks rely on the efficient distribution of entanglement to enable long-distance quantum communication and information processing. A key challenge in these networks is the design of routing protocols capable of maintaining highquality entanglement in the presence of noise, decoherence, and imperfect operations, which progressively degrade the fidelity of entangled states through entanglement swapping. Entanglement purification provides an effective mechanism to mitigate this degradation at the cost of additional resources. In this work, we study purification-aware quantum routing and formulate the problem of selecting optimal purification strategies as an optimization task. By employing dynamic programming techniques, we identify strategies that optimally balance resource consumption and end-to-end fidelity, demonstrating the effectiveness of our approach across different scenarios.

Paper ID: 27

Authors: Vinicius Marchioli, Mattia Boggio, Deborah Volpe and Carlo Novara

Title: Nonlinear Model Predictive Control for Spacecraft Rendezvous via Quantum Optimization

Abstract: We present a quantum-optimization–based Nonlinear Model Predictive Control (NMPC) framework and its application to autonomous spacecraft rendezvous. The validity of the proposed approach is proved on a realistic proximityoperations scenario in which a chaser spacecraft must rendezvous with a target while strictly satisfying Keep-Out Zone (KOZ) and Line-Of-Sight (LOS) constraints. By introducing a polynomial prediction model and a new strategy for handling nonlinear constraints based on dynamic penalty functions, the NMPC problem is written as a Quadratic Unconstrained Binary Optimization (QUBO) problem, making it suitable for execution on quantum and quantum-inspired solvers. Simulation results obtained on a high-fidelity spacecraft dynamics model show that the proposed method enables safe and efficient rendezvous trajectories while exhibiting good scalability with increasing problem size.

Paper ID: 29

Authors: Mathieu Garrigues, Victor Onofre, Wesley Coelho and S Acheche

Title: A Scalable Heuristic for Molecular Docking on Neutral-Atom Quantum Processors

Abstract: Molecular docking is a critical computational method in drug discovery used to predict the binding conformation and orientation of a ligand within a protein’s binding site. Mapping this challenge onto a graph-based problem, specifically the Maximum Weighted Independent Set (MWIS) problem, allows it to be addressed by specialized hardware such as neutralatom quantum processors. However, a significant bottleneck has been the size mismatch between biologically relevant molecular systems and the limited capacity of near-term quantum devices. In this work, we overcome this scaling limitation by the use of a divide-and-conquer heuristic introduced in Cazals et al [1]. This algorithm decomposes a single, intractable graph instance into smaller sub-problems that can be solved sequentially on a neutral-atom quantum emulator, incurring only a linear computational overhead. We benchmark this approach on 10 real-world protein-ligand complexes, including 9 from the Astex Diverse Set, with graphs ranging from 225 to 585 vertices. The quantum heuristic consistently outperforms a greedy baseline and achieves the provably optimal solution on a 540-node instance (TACE-AS). We further assess the biological relevance of the reconstructed poses via the fraction of native contacts, and benchmark the full workflow on a standard dataset of diverse protein-ligand complexes. Our work establishes a scalable blueprint for applying quantum optimization to molecular docking, while identifying concrete directions for improving both the algorithmic strategy and the underlying graph model.

Paper ID: 33

Authors: Ying-Yi Hong, Dylan Josh Lopez, Mark Angelo Purio, Gerard Francesco Apolinario, Yun-Yuan Wang, Eddie Huang and Simon See

Title: GPU-Based Hybrid Quantum Computing for EV Charging Station Decision Planning–A Case Study

Abstract: Electric vehicle charging station (EVCS) planning increasingly relies on data analytics pipelines. Yet current approaches remain purely classical and do not exploit emerging hybrid quantum methods for coupled forecasting and optimization. This paper proposes a graphics processing unit (GPU)-based hybrid quantum computing pipeline that integrates quantum-enhanced forecasting and decision models into an end-to-end planning workflow. For forecasting, a hybrid quantum neural network augments a long short-term memory network forecaster with a quantum head, while GPU-parallelized parameter-shift gradients and batched inference reduce training time by up to 76.9% relative to a non-parallel baseline without sacrificing test-set accuracy. For decision optimization, a station decommissioning problem is formulated as a mixed quadratic integer program, converted to a scaled quadratic unconstrained binary optimization model, and then solved via both classical annealing and quantum approximate optimization algorithm using a toolchain spanning CVXPY, autoQUBO/OpenJij, Qamomile, Qiskit, and CUDA‑Q/X, with scaled model runs matching the classical optimal profit and configuration while benefiting from GPU-accelerated simulation. At the 15-qubit scale studied under noiseless simulation, hybrid quantum components match but do not exceed classical baselines; the evidence supports their use as weighted heads or optimization backends co-designed with GPU parallelization and classical solvers, with quantum advantage and noise robustness deferred to larger-scale and QPU-based future work.

Paper ID: 11

Authors: Volker Reers and Marc Maußner

Title: A Threat-Model–Driven Robustness Benchmark for Quantum Machine Learning Under Device Noise and Deployment Shift

Abstract: Quantum machine learning (QML) systems executed on noisy intermediate-scale quantum (NISQ) hardware are vulnerable to both adversarial perturbations and backenddependent noise effects. Existing robustness evaluations are typically performed under fixed execution conditions and rarely consider deployment shift between training and inference backends. This work proposes a threat-model–driven benchmark framework for evaluating QML robustness under adversarial attacks, stochastic device noise, and backenddependent deployment variability. Using FGSM and PGD attacks, we analyze robustness scaling under varying training budgets, encoding strategies, and fake quantum backends. Experimental results demonstrate substantial backenddependent robustness variability and reduced robustness transfer across heterogeneous backend environments. The findings highlight the importance of backend-aware robustness evaluation for realistic NISQ-era QML deployments.

Paper ID: 83

Authors: Ananya Kulkarni, Muhammad Saeed and Kanak Pandit

Title: Quantum-Enhanced Generative Adversarial Networks for Industrial Surface Defect Synthesis

Abstract: Industrial surface defect detection plays a crucial role in modern manufacturing; however, real-world datasets are often limited and imbalanced, making it challenging to train robust deep learning models. Generative modeling offers a practical solution by synthesizing realistic defect samples for data augmentation in data-constrained environments. This work investigates hybrid quantum–classical generative adversarial networks (GANs) for industrial surface defect synthesis. Parameterized quantum circuits are integrated into the latent mapping stage of conditional and style-based GAN architectures to enhance feature representation and improve generative performance. A unified experimental framework is developed using the NEU Surface Defect Dataset, with evaluation based on Fréchet Inception Distance (FID), precision and recall, per-class label fidelity, Grad-CAM-based interpretability, and downstream classification using real–generated data mixtures. Experimental results indicate that classical StyleGAN achieves strong global performance in terms of visual fidelity. In contrast, quantum-conditioned models demonstrate improved class-wise semantic consistency and enhanced generative diversity, reflected by lower FID scores and improved precision–recall characteristics. For lower-capacity Conditional GANs (CGANs), quantum conditioning yields noticeable improvements in generative quality and classification performance. In downstream evaluation, incorporating 10% synthetic data yields a classification accuracy of 99.42%, demonstrating the practical utility of the generated data. These findings suggest that quantum conditioning provides targeted improvements rather than universal gains, particularly for structurally complex defects and limited-capacity generators. However, the benefits are influenced by factors such as circuit design, qubit limitations, and simulation overhead, highlighting current scalability challenges. Overall, hybrid quantum–classical models offer a promising complementary approach for enhancing synthetic data quality and supporting intelligent industrial inspection under data-constrained conditions.

Paper ID: 88

Authors: Filip Turoboś and Piotr Koneczny

Title: A Two-Stage MILP-to-QUBO Reformulation for University Timetabling

Abstract: University timetabling is an NP-complete combinatorial optimisation problem traditionally solved via mixed-integer linear programming (MILP). As problem size grows --courses, lecturers, student groups, rooms, and multi-day horizons -- traditional exact MILP solvers face various problems related to scalability. Quantum and hybrid quantum-classical optimisers offer a promising alternative, but they usually require the problem to be cast as a Quadratic Unconstrained Binary Optimisation (QUBO) instance. We present a detailed, constraint-by-constraint transition of a pair of MILP models for university course timetabling into their QUBO counterparts. The source models reduce the initial complexity by formulating the problem as a two-stage pipeline: a time-allocation stage that assigns courses to lecturers and timeslots, and a room-assignment stage that maps the resulting schedule onto physical and virtual rooms according to previously solved time slotting. Each constraint family is mapped to a specific QUBO penalty pattern, with attention to limiting auxiliary variable overhead. We discuss several reduction approaches, simultaneously trying to keep a clear energy separation between feasible and infeasible energy levels. The poster visualises the MILP-to-QUBO pipeline for the university timetabling problem on illustrative examples and discusses implications for hybrid quantum-classical scaling. We discuss practical trade-offs between alternative constraint encodings, including auxiliary variable usage, penalty scaling, and their impact on solution feasibility and penalty interactions.

Paper ID: 89

Authors: Boris Baudel, Rebbecca Ty Thien and Nina Amini

Title: Finite-Horizon Directional Observability for Optical Phase Estimation in Cavity-Mode Systems

Abstract: We study optical phase estimation in a cavity-mode system, with particular emphasis on how the system design shapes the transfer of phase information to the homodyne measurement output. To quantify this effect, we introduce a finite-horizon directional observability metric for the phase channel, derived from the observability Gramian of the underlying model. This metric characterises the visibility phase over a finite time interval and yields a corresponding directional estimation bound. Simulation works indicate that, within fixed regimes, the proposed metric serves as a useful guide for selecting parameter configurations that yield a lower phase-estimation error. We then investigated an augmented sensing architecture in which an auxiliary mode is coupled to the cavity prior to detection. Numerical results show that such preprocessing can improve estimation in a certain manner, either by enhancing directional phase visibility or by reshaping the effective stochastic burden seen by the estimator. These results support a broader design viewpoint for quantum cavity sensing in which the observability structure and hidden stochastic phase dynamics must be assessed jointly.