arXiv subjects

Wonhyuk Lee

Publications and source records attributed to Wonhyuk Lee.

11 recordsLinked to original sources

arXiv2026

Efficient Quantum Modular Reduction: Crandall reduction and its Fault-tolerant resource analysis

Modular arithmetic is central to quantum algorithms for cryptographic problems, including Shor's algorithm and Grover-based cryptanalysis, with modular reduction contributing substantially to circuit cost. Pseudo-Mersenne moduli $q=2^n-c$ allow classical Crandall reduction to replace division with folding and constant arithmetic, providing a structural opportunity for more efficient quantum modular reduction than Barrett reduction. We translate this advantage into a reversible quantum setting by deriving explicit folding and normalization conditions for $2n$-bit inputs. To the best of our knowledge, this constitutes the first exact reversible quantum circuit formulation of Crandall reduction. Based on this formulation, we develop two variants: Crandall reduction-1 is designed to minimize execution cost through one-step normalization, whereas Crandall reduction-2 uses two-step normalization to support a wider range of $c$ with limited overhead. Logical resource estimates show that both variants require fewer qubits and lower T-count and T-depth than optimized folding Barrett reduction. At $n=10$, Crandall reduction-1 reduces both T-count and T-depth by approximately 46.9% relative to optimized folding Barrett reduction. Surface-code analysis further shows that, at $n=20$ under the Sparse Blossom decoder, the estimated runtimes of the two variants are 30.05 ms and 35.39 ms, respectively, compared with 53.77 ms for optimized folding Barrett reduction. These results demonstrate the practical value of exploiting modulus-specific arithmetic structure in fault-tolerant quantum circuit design.

quant-ph↗

arXiv2026

Guard Vector: Beyond English LLM Guardrails with Task-Vector Composition and Streaming-Aware Prefix SFT

We introduce Guard Vector, a safety task vector computed as the parameter difference between a guardrail model (Guard Model) and a same-architecture pretrained language model. Composing this vector with a target language model yields a Target Guard Model (TGM). We then adapt TGM with a streaming-aware approach that combines prefix-based training and evaluation with a classifier that produces a single-token output. With this composition alone, TGM improves classification quality over established Guard Models across standard safety suites and enables language extensibility to Chinese, Japanese, and Korean, requiring neither additional training nor target language labels for this composition step. It also demonstrates model portability across two widely used public guardrail backbones, Llama and Gemma. With prefix SFT (supervised fine-tuning), TGM preserves classification quality under streaming by aligning the behavior between prefix inputs and full-text inputs. The single-token output design increases throughput and reduces latency. Together, these components reduce data and compute requirements while promoting streaming-aware evaluation practices, thereby contributing to a more responsible AI ecosystem.

cs.CL↗

arXiv2026

Syndrome-as-Header: A Quantum Label-Switching Architecture via Uncorrectable Error Injection

Scaling quantum networks beyond point-to-point links requires packet forwarding that can tolerate control-plane timing uncertainty. Existing optical-burst switching (OBS) and quantum-wrapper-based packet switching (QW) rely on external classical headers whose jitter-prone processing must remain aligned with in-flight quantum payloads, creating guard-window or fiber-delay-line budget failures. This paper proposes Syndrome-as-Header (SAH), a quantum label-switching architecture that embeds routing labels into the syndrome structure of an encoded payload. The proposed scheme uses Uncorrectable Error Injection (UEI) to map flow labels to reference syndromes and builds a syndrome-space header codebook whose decoding regions remain distinguishable under correctable channel-induced syndrome deviations. Core routers extract the syndrome header, suppress the correctable residual channel-error component, and swap labels without measuring or decoding the logical payload. SAH supports FAST forwarding and VERIFICATION mode, the latter providing end-to-end consistency checking while retaining swappable labels. In NSFNet timing benchmarks with common per-hop quantum error correction (QEC), SAH eliminates the external-header/payload alignment condition, so timing uncertainty appears as memory residence and delivered latency rather than alignment-budget drops. With a $T_2$-based memory-residence penalty, the advantage depends on router memory coherence; for sufficiently long coherence, SAH maintains higher acceptance and throughput than OBS+QEC and QW+QEC under high jitter.

quant-ph↗

arXiv2026

Asymmetry-aided measurement-based quantum repeaters and distributed quantum computing with a decoder-free client

Distributed quantum computation needs to move logical qubits across lossy optical links, yet this transmission layer is usually designed separately from the computation it serves. We treat the two together by recognizing that a measurement-based quantum repeater is a two-dimensional code foliated along the transmission axis, so that the dominant channel loss is concentrated on the transmitted sector while the locally measured qubits are largely spared. Matching a code's distance to this structural asymmetry, we show that a rectangular Bacon-Shor subsystem code transmits a logical qubit markedly more efficiently than transmission-unaware encodings. Over continental distances, its cost-optimal repeater density is about an order of magnitude lower than that of a recent $[[48,6,8]]$ benchmark at comparable transmission rate, and roughly half that of a symmetric code of equal size. Moreover, we extend the framework to a central-to-client round trip in which a code-level, distance-preserving code switch joins the transmission legs to the client's computation, and joint decoding of the heterogeneous syndrome record at the central node lets distributed quantum computation proceed with a decoder-free client.

quant-ph↗

arXiv2026

KSAFE-MM: A Multimodal Safety Benchmark via Localized Contextualization for Korean Cultural Risks

Multimodal Large Language Models (MLLMs) exacerbate safety risks by introducing vulnerabilities across multiple modalities, such as language and vision. Current MLLM safety evaluation tools, however, suffer from major limitations: 1) English-centric dataset construction, and 2) a focus on generic risks that are not tied to local cultural contexts. This paper introduces KSAFE-MM, a benchmark for Korean multimodal safety evaluation that covers both general safety risks and culture-specific vulnerabilities. KSAFE-MM consists of two parts, KSAFE-MM-G and KSAFE-MM-C. KSAFE-MM-G evaluates globally shared risks in Korean contexts through linguistic contextualization, which transforms generic safety queries into contextually grounded multimodal samples. KSAFE-MM-C targets culture-dependent MLLM safety vulnerabilities using localized visual queries derived from real-world contexts. It pairs these visual queries with jailbreak-style textual queries to cover multimodal safety risks involving cultural visual cues and malicious textual intent. Together, these components provide a general-to-local construction pipeline for evaluating both globally shared safety risks and culture-specific vulnerabilities. We evaluate 12 state-of-the-art MLLMs on KSAFE-MM and reveal that models exhibit greater vulnerability to culturally grounded attacks than to generic ones. Notably, jailbreaking strategies substantially amplify attack success rates, with ProgramExecution yielding up to 74.2% ASR compared to 13.4% for standard queries. Furthermore, we identify a systematic trade-off between safety and over-refusal, where models achieving low ASR tend to exhibit excessive refusal behavior on benign queries. These findings highlight the urgent need for culturally grounded safety evaluation beyond English-centric benchmarks.

cs.CL↗

arXiv2026

Responsible AI Technical Report

KT developed a Responsible AI (RAI) assessment methodology and risk mitigation technologies to ensure the safety and reliability of AI services. By analyzing the Basic Act on AI implementation and global AI governance trends, we established a unique approach for regulatory compliance and systematically identify and manage all potential risk factors from AI development to operation. We present a reliable assessment methodology that systematically verifies model safety and robustness based on KT's AI risk taxonomy tailored to the domestic environment. We also provide practical tools for managing and mitigating identified AI risks. With the release of this report, we also release proprietary Guardrail : SafetyGuard that blocks harmful responses from AI models in real-time, supporting the enhancement of safety in the domestic AI development ecosystem. We also believe these research outcomes provide valuable insights for organizations seeking to develop Responsible AI.

cs.CL↗

arXiv2026

Designing generalized elegant Bell inequalities in higher dimensions from a Tsirelson bound

Elegant Bell inequality is well known for its distinctive property, being maximally violated by maximal entanglement, mutually unbiased bases, and symmetric informationally complete positive operator-valued measure elements. Despite its significance in quantum information theory demonstrated based on its unique violation feature, it remains the only known one with the characteristic. We present a method to construct Bell inequalities with violation feature analogous to elegant Bell inequality in higher local dimensions from a simple, analytic quantum bound. A Bell inequality with the generalized violation feature is derived in three dimension for the first time. It exhibits larger violation than existing Bell inequalities of similar classes, including the original elegant Bell inequality, while requiring arguably small number of measurements.

quant-ph↗

arXiv2025

Fault-tolerant and secure long-distance quantum communication via uncorrectable-error-injection

Quantum networks aim to facilitate the fault-tolerant and secure transmission of quantum states across distant devices. The widely adopted quantum teleportation scheme requires multiple rounds of entanglement swapping and purification, leading to significant resource overhead and operational complexity. In this study, we propose a novel fault-tolerant and secure quantum communication scheme based on uncorrectable error injection. Our method exploits a quantum state encoding scheme based on quantum error correction codes, which strategically introduces uncorrectable errors to enhance security. It eliminates the need for entanglement distribution while reducing resource requirements. The injected errors protect against eavesdropping by preventing unauthorized parties from retrieving meaningful information. Security analysis shows that as the data length and encoded message size increase, information leakage becomes negligible relative to the size of the total message. Comparative performance analysis with existing approaches indicates that our method reduces transmission overhead while maintaining comparable fidelity in low-error regimes. These findings suggest that the proposed method offers a scalable and practical alternative for secure long-distance quantum communication, distributed quantum computing, and future quantum internet applications.

quant-ph↗

arXiv2025

Structural encoding with classical codes for computational-basis bit-flip correction in the early fault-tolerant regime

Achieving reliable performance on early fault-tolerant quantum hardware will depend on protocols that manage noise without incurring prohibitive overhead. We propose a novel framework that integrates quantum computation with the functionality of classical error correction. In this approach, quantum computation is performed within the codeword subspace defined by a classical error correction code. The correction of various types of errors that manifest as bit flips is carried out based on the final measurement outcomes. The approach leverages the asymmetric structure of many key algorithms, where problem-defining diagonal operators (e.g., oracles) are paired with fixed non-diagonal operators (e.g., diffusion operators). The proposed encoding maps computational basis states to classical codewords. This approach commutes with diagonal operators, obviating their overhead and confining the main computational cost to simpler non-diagonal components. Noisy simulations corroborate this analysis, demonstrating that the proposed scheme serves as a viable protocol-level layer for enhancing performance in the early fault-tolerant regime.

quant-ph↗

arXiv2025

Enhanced Extrapolation-Based Quantum Error Mitigation Using Repetitive Structure in Quantum Algorithms

Quantum error mitigation is a crucial technique for suppressing errors especially in noisy intermediate-scale quantum devices, enabling more reliable quantum computation without the overhead of full error correction. Zero-Noise Extrapolation (ZNE), which we mainly consider in this work, is one of prominent quantum error mitigation methods. For algorithms with deep circuits - such as iterative quantum algorithms involving multiple oracle calls - ZNE's effectiveness is significantly degraded under high noise. Extrapolation based on such low-fidelity data often yields inaccurate estimates and requires substantial overhead. In this study, we propose a lightweight, extrapolation-based error mitigation framework tailored for structured quantum algorithms composed of repeating operational blocks. The proposed method characterizes the error of the repeated core operational block, rather than the full algorithm, using shallow circuits. Extrapolation is used to estimate the block fidelity, followed by a reconstruction of the mitigated success probability. We validate our method via simulations of the 6-qubit Grover's algorithm on IBM's Aer simulator, then further evaluating it on the real 127-qubit IBM Quantum system based on Eagle r3 under a physical noise environment. Our results, particularly those from Aer simulator, demonstrate that the core block's error follows a highly consistent exponential decay. This allows our technique to achieve robust error mitigation, overcoming the limitations of conventional ZNE which is often compromised by statistically unreliable data from near-random behavior under heavy noise. In low-noise conditions, our method approaches theoretical success probability, outperforms ZNE. In high-noise conditions, ZNE fails to mitigate errors due to overfitting of its extrapolation data, whereas our method achieves over a 20% higher success probability.

quant-ph↗

arXiv2025

Error correctable efficient quantum homomorphic encryption using Calderbank-Shor-Steane codes

The integration of quantum error correction codes and homomorphic encryption schemes is essential for achieving fault-tolerant secure cloud quantum computing. However, owing to the significant overheads associated with these schemes, their efficiency is paramount. In this study, we develop an efficient quantum homomorphic encryption scheme based on quantum error correction codes that uses a single encoding process to simultaneously perform encryption and encoding. By using a longer quantum error correction code, both the security and error-correction capabilities of the scheme are improved. Through comprehensive evaluations, we demonstrate that the proposed scheme is more secure than the conventional permutation-key-based QHE scheme when the number of maximally mixed states is not more than twice the length of the quantum error-correction code. The proposed scheme offers a more secure and efficient approach to quantum cloud computing, potentially paving the way for more practical and scalable quantum cryptographic protocols.

quant-ph↗