arXiv subjects

Rei Sato

Publications and source records attributed to Rei Sato.

15 recordsLinked to original sources

arXiv2026

Quantum Circuits for Quantum Spatial Search on $d$-Dimensional Lattices

We propose an explicit quantum circuit for quantum spatial search based on discrete-time quantum walks on $d$-dimensional lattices. In this algorithm, the flip-flop shift operator moves the walker to a neighboring site along the selected spatial direction and reverses the corresponding direction label after the move. By encoding each pair of opposite directions so that they differ only in the least significant qubit of the coin register, we implement the shift using coin-controlled modular increment and decrement operations on the position registers, together with a single $X$ gate that reverses the direction label. We verify that the proposed circuits reproduce the theoretical dynamics on two- and three-dimensional periodic lattices. We further extend the circuit construction to systems with position-dependent shift rules, such as non-periodic boundaries, and validate this extension on a two-dimensional lattice. The resource analysis shows logarithmic growth of the circuit width. For the two-dimensional lattice, the synthesized depth is consistent with $O(\sqrt{N}(\log N)^{3/2})$, while the three-dimensional depth empirically follows an $O(\sqrt{N})$ dependence over the investigated range. Under a CX-gate depolarizing noise model, CX-optimized circuits exhibit improved noise robustness. These results provide a practical framework for implementing quantum spatial search on regular lattices and extending it to defective and other irregular lattice structures.

quant-ph↗

arXiv2026

Quantum Counterparty Credit Risk: A Study of Path-Dependent Derivatives

Estimating potential future exposure (PFE) for path-dependent derivatives, such as FX Target Redemption Forwards (TARFs), represents a formidable computational challenge due to the demand of nested Monte Carlo simulations. We present a hybrid quantum-classical framework that leverages Iterative Quantum Amplitude Estimation (IQAE) to address this via a reduced-order counterparty credit risk model. Our methodology maps the non-linear TARF payoff -- including cumulative gains and knock-out features -- into a quantum circuit via a two-step formulation, whereby a first-step percentile is computed classically and then used to condition quantum evaluation of subsequent exposure. We employ discretisation of the FX process and a linearised additive approximation of dynamics to enable implementation on current quantum platforms. Developed via the Classiq platform and validated on NVIDIA CUDA-Q and Amazon Braket SV1, our approach achieves relative errors of 1%-8% against classical benchmarks at the 97.5% and 99% confidence levels. While discretisation constraints and approximate monotonicity assumption may introduce bias and limit recovery of the full exposure distribution, our framework offers a tractable testbed for quantum acceleration. Scaling analysis suggests that $\sim$300 logical qubits could enable full 52-week exposure estimation, reducing sample complexity for tail-risk estimation via amplitude estimation at the cost of increased circuit depth.

quant-ph↗

arXiv2025

Coined Quantum Walks on Complex Networks for Quantum Computers

We propose a quantum circuit design for implementing coined quantum walks on complex networks. In complex networks, the coin and shift operators depend on the varying degrees of the nodes, which makes circuit construction more challenging than for regular networks. To address this issue, we use a dual-register encoding to enable a simplified shift operator and reduces the resource overhead. We implement the circuit using Qmod, a high-level quantum programming language, and evaluated the performance through numerical simulations on Erd\H{o}s-R\'enyi, Watts-Strogatz, and Barab\'asi-Albert models. The results show that the circuit depth scales as approximately $N^{1.9}$ regardless of the network topology. Furthermore, we execute the proposed circuits on the ibm\_torino superconducting quantum processor for Watts-Strogatz models with $N=4$ and $N=8$. The experiments show that hardware-aware optimization slightly improved the variation distance and Hellinger fidelity for the larger network, whereas connectivity constraints imposed overhead for the smaller one. These results indicate that while current NISQ devices are limited to small-scale validations, the polynomial scaling of our framework makes it suitable for larger-scale implementations in the fault-tolerant quantum computing era.

quant-ph↗

arXiv2025

Quantum walk search based edge detection of images

Quantum walk has emerged as an essential tool for searching marked vertices on various graphs. Recent advances in the discrete-time quantum walk search algorithm have enabled it to effectively handle multiple marked vertices, expanding its range of applications further. In this article, we propose a novel application of this advanced quantum walk search algorithm for the edge detection of images\textemdash a critical task in digital image processing. Given the probabilistic nature of quantum computing, obtaining measurement result with a high success probability is essential alongside faster computation time. Our quantum walk search algorithm demonstrates a high success probability in detecting the image edges compared to the existing quantum edge detection methods and outperforms classical edge detection methods with a quadratically faster speed. A small Qiskit circuit implementation of our method using a one-dimensional quantum walk search has been executed in Qiskit's $qasm\_simulator$ and $ibm\_sydney(fake)$ device.

quant-ph↗

arXiv2025

A Provable Approach for End-to-End Safe Reinforcement Learning

A longstanding goal in safe reinforcement learning (RL) is a method to ensure the safety of a policy throughout the entire process, from learning to operation. However, existing safe RL paradigms inherently struggle to achieve this objective. We propose a method, called Provably Lifetime Safe RL (PLS), that integrates offline safe RL with safe policy deployment to address this challenge. Our proposed method learns a policy offline using return-conditioned supervised learning and then deploys the resulting policy while cautiously optimizing a limited set of parameters, known as target returns, using Gaussian processes (GPs). Theoretically, we justify the use of GPs by analyzing the mathematical relationship between target and actual returns. We then prove that PLS finds near-optimal target returns while guaranteeing safety with high probability. Empirically, we demonstrate that PLS outperforms baselines both in safety and reward performance, thereby achieving the longstanding goal to obtain high rewards while ensuring the safety of a policy throughout the lifetime from learning to operation.

cs.LG↗

arXiv2025

Vulnerability Mitigation for Safety-Aligned Language Models via Debiasing

Safety alignment is an essential research topic for real-world AI applications. Despite the multifaceted nature of safety and trustworthiness in AI, current safety alignment methods often focus on a comprehensive notion of safety. By carefully assessing models from the existing safety-alignment methods, we found that, while they generally improved overall safety performance, they failed to ensure safety in specific categories. Our study first identified the difficulty of eliminating such vulnerabilities without sacrificing the model's helpfulness. We observed that, while smaller KL penalty parameters, increased training iterations, and dataset cleansing can enhance safety, they do not necessarily improve the trade-off between safety and helpfulness. We discovered that safety alignment could even induce undesired effects and result in a model that prefers generating negative tokens leading to rejective responses, regardless of the input context. To address this, we introduced a learning-free method, Token-level Safety-Debiased Inference (TSDI), to estimate and correct this bias during the generation process using randomly constructed prompts. Our experiments demonstrated that our method could enhance the model's helpfulness while maintaining safety, thus improving the trade-off Pareto-front.

cs.AI↗

arXiv2024

Circuit Implementation of Discrete-Time Quantum Walks on Complex Networks

In this paper, we propose a circuit design for implementing quantum walks on complex networks. Quantum walks are powerful tools for various graph-based applications such as spatial search, community detection, and node classification. Although many quantum-walk-based graph algorithms have been extensively studied, specific quantum circuits for implementing these algorithms have not yet been provided. To address this issue, we present a circuit design for implementing the discrete-time quantum walk on complex networks. We investigate the functionality of our circuit using the small-sized Watts-and-Strogatz model as the complex network model, comparing it with theoretical calculations. This work offers a new approach to constructing quantum circuits for implementing quantum walks on arbitrary complex networks.

quant-ph↗

arXiv2024

QWalkVec: Node Embedding by Quantum Walk

In this paper, we propose QWalkVec, a quantum walk-based node embedding method. A quantum walk is a quantum version of a random walk that demonstrates a faster propagation than a random walk on a graph. We focus on the fact that the effect of the depth-first search process is dominant when a quantum walk with a superposition state is applied to graphs. Simply using a quantum walk with its superposition state leads to insufficient performance since balancing the depth-first and breadth-first search processes is essential in node classification tasks. To overcome this disadvantage, we formulate novel coin operators that determine the movement of a quantum walker to its neighboring nodes. They enable QWalkVec to integrate the depth-first search and breadth-first search processes by prioritizing node sampling. We evaluate the effectiveness of QWalkVec in node classification tasks conducted on four small-sized real datasets. As a result, we demonstrate that the performance of QWalkVec is superior to that of the existing methods on several datasets. Our code will be available at \url{https://github.com/ReiSato18/QWalkVec}.

quant-ph↗

arXiv2024

Two-Step Quantum Search Algorithm for Solving Traveling Salesman Problems

Quantum search algorithms, such as Grover's algorithm, are anticipated to efficiently solve constrained combinatorial optimization problems. However, applying these algorithms to the traveling salesman problem (TSP) on a quantum circuit presents a significant challenge. Existing quantum search algorithms for the TSP typically assume that an initial state -- an equal superposition of all feasible solutions satisfying the problem's constraints -- is pre-prepared. The query complexity of preparing this state using brute-force methods scales exponentially with the factorial growth of feasible solutions, creating a significant hurdle in designing quantum circuits for large-scale TSPs. To address this issue, we propose a two-step quantum search (TSQS) algorithm that employs two sets of operators. In the first step, all the feasible solutions are amplified into their equal superposition state. In the second step, the optimal solution state is amplified from this superposition state. The TSQS algorithm demonstrates greater efficiency compared to conventional search algorithms that employ a single oracle operator for finding a solution within the encoded space. Encoded in the higher-order unconstrained binary optimization (HOBO) representation, our approach significantly reduces the qubit requirements. This enables efficient initial state preparation through a unified circuit design, offering a quadratic speedup in solving the TSP without prior knowledge of feasible solutions.

quant-ph↗

arXiv2024

Stepwise Alignment for Constrained Language Model Policy Optimization

Safety and trustworthiness are indispensable requirements for real-world applications of AI systems using large language models (LLMs). This paper formulates human value alignment as an optimization problem of the language model policy to maximize reward under a safety constraint, and then proposes an algorithm, Stepwise Alignment for Constrained Policy Optimization (SACPO). One key idea behind SACPO, supported by theory, is that the optimal policy incorporating reward and safety can be directly obtained from a reward-aligned policy. Building on this key idea, SACPO aligns LLMs step-wise with each metric while leveraging simple yet powerful alignment algorithms such as direct preference optimization (DPO). SACPO offers several advantages, including simplicity, stability, computational efficiency, and flexibility of algorithms and datasets. Under mild assumptions, our theoretical analysis provides the upper bounds on optimality and safety constraint violation. Our experimental results show that SACPO can fine-tune Alpaca-7B better than the state-of-the-art method in terms of both helpfulness and harmlessness.

cs.LG↗

arXiv2024

Universal scaling hypothesis of quantum spatial search in complex networks

Since quantum spatial searches on complex networks have a strong network dependence, the question arises whether the universal perspective exists in this quantum algorithm for complex networks. Here, we uncover the universal scaling laws of the quantum spatial search on complex networks such as small-world and scale-free networks. The average path length, a key quantity in the complex network science, is useful to expose this universal feature, where the collapse plot can be generated for the optimal time, the maximal finding probability and the optimal hopping parameter. Based on the path integral method, we also clarify that the probability amplitude in the continuous-time quantum walk can be determined by the path length distribution. Our results demonstrate a new link between the quantum physics and the complex networks.

quant-ph↗

arXiv2023

Few-Shot Image-to-Semantics Translation for Policy Transfer in Reinforcement Learning

We investigate policy transfer using image-to-semantics translation to mitigate learning difficulties in vision-based robotics control agents. This problem assumes two environments: a simulator environment with semantics, that is, low-dimensional and essential information, as the state space, and a real-world environment with images as the state space. By learning mapping from images to semantics, we can transfer a policy, pre-trained in the simulator, to the real world, thereby eliminating real-world on-policy agent interactions to learn, which are costly and risky. In addition, using image-to-semantics mapping is advantageous in terms of the computational efficiency to train the policy and the interpretability of the obtained policy over other types of sim-to-real transfer strategies. To tackle the main difficulty in learning image-to-semantics mapping, namely the human annotation cost for producing a training dataset, we propose two techniques: pair augmentation with the transition function in the simulator environment and active learning. We observed a reduction in the annotation cost without a decline in the performance of the transfer, and the proposed approach outperformed the existing approach without annotation.

cs.LG↗

arXiv2023

Max-Min Off-Policy Actor-Critic Method Focusing on Worst-Case Robustness to Model Misspecification

In the field of reinforcement learning, because of the high cost and risk of policy training in the real world, policies are trained in a simulation environment and transferred to the corresponding real-world environment. However, the simulation environment does not perfectly mimic the real-world environment, lead to model misspecification. Multiple studies report significant deterioration of policy performance in a real-world environment. In this study, we focus on scenarios involving a simulation environment with uncertainty parameters and the set of their possible values, called the uncertainty parameter set. The aim is to optimize the worst-case performance on the uncertainty parameter set to guarantee the performance in the corresponding real-world environment. To obtain a policy for the optimization, we propose an off-policy actor-critic approach called the Max-Min Twin Delayed Deep Deterministic Policy Gradient algorithm (M2TD3), which solves a max-min optimization problem using a simultaneous gradient ascent descent approach. Experiments in multi-joint dynamics with contact (MuJoCo) environments show that the proposed method exhibited a worst-case performance superior to several baseline approaches.

cs.LG↗

arXiv2021

AdvantageNAS: Efficient Neural Architecture Search with Credit Assignment

Neural architecture search (NAS) is an approach for automatically designing a neural network architecture without human effort or expert knowledge. However, the high computational cost of NAS limits its use in commercial applications. Two recent NAS paradigms, namely one-shot and sparse propagation, which reduce the time and space complexities, respectively, provide clues for solving this problem. In this paper, we propose a novel search strategy for one-shot and sparse propagation NAS, namely AdvantageNAS, which further reduces the time complexity of NAS by reducing the number of search iterations. AdvantageNAS is a gradient-based approach that improves the search efficiency by introducing credit assignment in gradient estimation for architecture updates. Experiments on the NAS-Bench-201 and PTB dataset show that AdvantageNAS discovers an architecture with higher performance under a limited time budget compared to existing sparse propagation NAS. To further reveal the reliabilities of AdvantageNAS, we investigate it theoretically and find that it monotonically improves the expected loss and thus converges.

cs.LG↗

arXiv2020

Scaling Hypothesis of Spatial Search on Fractal Lattice Using Quantum Walk

We investigate a quantum spatial search problem on fractal lattices, such as Sierpinski carpets and Menger sponges. In earlier numerical studies of the Sierpinski gasket, the Sierpinski tetrahedron, and the Sierpinski carpet, conjectures have been proposed for the scaling of a quantum spatial search problem finding a specific target, which is given in terms of the characteristic quantities of a fractal geometry. We find that our simulation results for extended Sierpinski carpets and Menger sponges support the conjecture for the ${\it optimal}$ number of the oracle calls, where the exponent is given by $1/2$ for $d_{\rm s} > 2$ and the inverse of the spectral dimension $d_{\rm s}$ for $d_{\rm s} < 2$. We also propose a scaling hypothesis for the ${\it effective}$ number of the oracle calls defined by the ratio of the ${\it optimal}$ number of oracle calls to a square root of the maximum finding probability. The form of the scaling hypothesis for extended Sierpinski carpets is very similar but slightly different from the earlier conjecture for the Sierpinski gasket, the Sierpinski tetrahedron, and the conventional Sierpinski carpet.

quant-ph↗