arXiv subjects

Junghyun Lee

Publications and source records attributed to Junghyun Lee.

At least 19 recordsLinked to original sources

arXiv2026

Hybrid dynamical decoupling and coherent driving for high-fidelity nuclear-spin control in diamond

Nitrogen-vacancy (NV) centers in diamond provide room-temperature electron-nuclear spin registers for quantum sensing and quantum information processing, with surrounding 13C nuclear spins serving as long-lived quantum memories. However, coherent control of large nuclear-spin registers is limited by finite electron-spin coherence and spectral addressability. Existing approaches follow two complementary strategies: dynamical-decoupling (DD) gates exploit filter-function resonances to realize selective conditional evolution but permit only discrete rotation angles, whereas dynamical-decoupling radio-frequency (DDrf) control restores continuous tunability at the cost of stringent hyperfine-geometry and RF-power requirements. Here, we introduce hybrid dynamical-decoupling and radio-frequency (H-DDrf) control, which preserves the DD-induced conditional evolution and employs a geometrically phase-matched RF drive to complete the target operation. This approach reduces both RF power and gate duration while maintaining high-fidelity control, thereby expanding the accessible 13C nuclear-spin register for room-temperature NV-based quantum memories and quantum processors.

quant-ph↗

arXiv2026

A First-principles Computational Framework for Quantum Decoherence in Complex Diamond Spin Environments

Quantum decoherence induced by defects remains a major limitation for solid-state quantum technologies, yet predicting decoherence in realistic materials remains computationally challenging. Complex defect populations are often approximated as homogeneous spin baths, obscuring the role of defect-specific electronic structure and spin dynamics. Here, we develop a predictive framework for decoherence in diamond by combining first-principles electronic-structure calculations, quantum many-body spin-bath simulations, and experimental validation. The framework incorporates defect-resolved spin Hamiltonians and heterogeneous spin baths containing multiple paramagnetic defect species. Using diamond nitrogen-vacancy ensembles as a model platform, we investigate mixed nitrogen-, vacancy-, and hydrogen-related defect environments. We show that decoherence depends not only on defect density but also on defect identity and bath composition, whose distinct electronic structures, hyperfine interactions, and spin dynamics produce different coherence behavior. Heterogeneous defect populations can either suppress or enhance decoherence, producing trends unexplained by homogeneous-bath models. Magnetic-field-dependent Hahn-echo measurements on samples with different defect concentrations validate the framework. The calculations reproduce the observed coherence times and stretched-exponential decay behavior across a broad magnetic-field range and identify vacancy-related defects as critical contributors beyond the conventionally assumed P1 spin bath. By linking atomistic defect properties to quantum coherence, our framework provides a predictive route for identifying hidden defect environments and optimizing decoherence in defect-based quantum materials.

quant-ph↗

arXiv2026

A Jointly Efficient and Optimal Algorithm for Heteroskedastic Generalized Linear Bandits with Adversarial Corruptions

We consider the problem of heteroskedastic generalized linear bandits (GLBs) with adversarial corruptions, which subsumes heteroskedastic linear bandits and logistic/Poisson bandits, in the presence of adversarial corruptions. We propose HCW-GLB-OMD, which consists of two components: an online mirror descent (OMD)-based estimator and Hessian-based confidence weights to achieve corruption robustness. This is computationally efficient in that it only requires ${O}(1)$ space and time complexity per iteration. Under the self-concordance assumption on the link function, we show a regret bound of $\tilde{O}\left( d \sqrt{\sum_t g(τ_t) \dotμ_{t,\star}} + d^2 g_{\max} κ+ d (g_{\max} + κ) C \right)$, where $\dotμ_{t,\star}$ is the slope of $μ$ around the optimal arm at time $t$, $g(τ_t)$'s are potentially exogenously time-varying dispersions (e.g., $g(τ_t) = σ_t^2$ for heteroskedastic linear bandits, $g(τ_t) = 1$ for Bernoulli and Poisson), $g_{\max} = \max_{t \in [T]} g(τ_t)$ is the maximum dispersion, and $C \geq 0$ is the total corruption budget of the adversary. We complement this with a lower bound of $\tildeΩ(d \sqrt{\sum_t g(τ_t) \dotμ_{t,\star}} + d C)$, unifying previous problem-specific lower bounds. Thus, our algorithm achieves, up to a $κ$-factor in the corruption term, instance-wise minimax optimality simultaneously across various instances of heteroskedastic GLBs with adversarial corruptions.

cs.LG↗

arXiv2026

Provably Efficient Regularized Online RLHF with Generalized Bilinear Preferences

We consider the problem of regularized best-response max-regret minimization in online RLHF under general preferences and bandit feedback. While various regularizers are utilized to robustify alignment, known polylogarithmic regret guarantees remain heavily specific to KL. To investigate whether such fast rates extend beyond KL, we adopt the Generalized Bilinear Preference Model (GBPM) -- capturing intransitive preferences over $d$-dimensional item-wise features via a rank-$2r$ skew-symmetric matrix -- to isolate the impact of generic regularization. Crucially, under GBPM, we prove that the dual gap of any greedy policy is bounded by the squared estimation error, derived using \emph{only} strong convexity and skew-symmetry. Under a feature coverage assumption, we establish a \emph{generic} polylogarithmic regret of $\tilde{\mathcal{O}}(ηd^4 C_{\min}^{-1} (\log T)^2 \wedge d^2 C_{\min}^{-1/2} \sqrt{T})$ with Greedy Sampling, and a dimension-wise improved regret (for well-conditioned arm-sets) of $\tilde{\mathcal{O}}(C_{\min}^{-2} \sqrt{ηr T} \wedge r^{1/3} C_{\min}^{-4/3} T^{2/3})$ with Explore-Then-Commit, where $η^{-1}$ is the regularization coefficient, $T$ is the time horizon, and $C_{\min}$ is an arm-set dependent quantity. This demonstrates that ``fast'' regrets are not KL-specific, but rather a fundamental consequence of generic strongly convex geometry.

cs.LG↗

arXiv2026

Ordering Matters: Rank-Aware Selective Fusion for Blended Emotion Recognition

Blended emotion recognition is challenging because emotions are often expressed as mixtures of subtle and overlapping multimodal cues rather than a single dominant signal. We propose a rank-aware multi-encoder framework that selectively combines complementary representations from diverse pre-extracted video and audio encoders. Our method projects heterogeneous encoder features into a shared latent space, estimates sample-wise encoder importance through an attention-based gating module, and fuses only the top-n most informative encoders. To better model blended emotions, we decouple prediction into presence and salience heads and align them through probability-level fusion. We further incorporate feature-level unsupervised domain adaptation without pseudo-labeling to improve robustness under distribution shift. Experiments on the BlEmoRE challenge show that the proposed framework outperforms strong individual encoders and naïve multi-encoder fusion baselines. Our final system ranked 2nd in the competition, supporting the effectiveness of rank-aware selective fusion for fine-grained blended emotion recognition.

cs.CV↗

arXiv2026

Instance-Optimal Estimation with Multiple LLM Judges on a Budget

Evaluating large language models increasingly relies on LLM-as-a-judge protocols, but such evaluations remain costly: different judges have different prices and reliabilities, and the difficulty of each prompt-response pair can vary substantially. This raises a basic allocation question: under a fixed budget, how should one distribute evaluation queries across heterogeneous judges and instances to obtain the most accurate score estimates? We formalize this question as *budgeted heteroskedastic multi-judge estimation*. Given $K$ prompt-response pairs, $J$ judges with known costs, and unknown query-judge variances, the goal is to estimate a bounded score vector while minimizing an $\ell_p$-error. Our first contribution is to analyze the inverse-variance weighted estimator (IVWE) and to derive the oracle allocation that minimizes its error rate. Since this allocation depends on the unknown variances, we then address the practical unknown-variance setting by proposing EST-IVWE, an adaptive algorithm that constructs and leverages *optimistically biased* variance estimates to stabilize the empirical allocation. We prove that EST-IVWE matches the oracle IVWE rate up to lower-order terms in the budget. Our second and central theoretical contribution is a matching *local* minimax lower bound, which establishes the instance-optimality of the proposed algorithms. A key technical insight is that Fano-type high-probability arguments are too coarse for this problem: their packing construction loses the local variance structure that governs the optimal allocation. We instead use an Assouad-type in-expectation argument, based on local perturbations, which preserves this structure and yields the sharp allocation-dependent lower bound. Finally, we numerically validate the superiority of our approach over naïve uniform allocation on synthetic and HelpSteer2 datasets.

cs.LG↗

arXiv2026

GL-LowPopArt: A Nearly Instance-Wise Minimax-Optimal Estimator for Generalized Low-Rank Trace Regression

We present `GL-LowPopArt`, a novel Catoni-style estimator for generalized low-rank trace regression. Building on `LowPopArt` (Jang et al., 2024), it employs a two-stage approach: nuclear norm regularization followed by matrix Catoni estimation. We establish state-of-the-art estimation error bounds, surpassing existing guarantees (Fan et al., 2019; Kang et al., 2022), and reveal a novel experimental design objective, $\mathrm{GL}(π)$. The key technical challenge is controlling bias from the nonlinear inverse link function, which we address with our two-stage approach. We prove a *local minimax lower bound*, showing that our `GL-LowPopArt` enjoys instance-wise optimality up to the condition number of the ground-truth Hessian. Our method immediately achieves an improved Frobenius error guarantee for generalized linear matrix completion. We also introduce a new problem setting called **bilinear dueling bandits**, a contextualized version of dueling bandits with a general preference model. Using an explore-then-commit approach with `GL-LowPopArt', we show an improved Borda regret bound over naïve vectorization (Wu et al., 2024).

stat.ML↗

arXiv2026

Random Is Hard to Beat: Active Selection in online DPO with Modern LLMs

Modern LLMs inherit strong priors from web-scale pretraining, which can limit the headroom of post-training data-selection strategies. While Active Preference Learning (APL) seeks to optimize query efficiency in online Direct Preference Optimization (DPO), the inherent richness of on-policy candidate pools often renders simple Random sampling a surprisingly formidable baseline. We evaluate uncertainty-based APL against Random across harmlessness, helpfulness, and instruction-following settings, utilizing both reward models and LLM-as-a-judge proxies. We find that APL yields negligible improvements in proxy win-rates compared to Random. Crucially, we observe a dissociation where win-rate improves even as general capability -- measured by standard benchmarks -- degrades. APL fails to mitigate this capability collapse or reduce variance significantly better than random sampling. Our findings suggest that in the regime of strong pre-trained priors, the computational overhead of active selection is difficult to justify against the ``cheap diversity'' provided by simple random samples. Our code is available at https://github.com/BootsofLagrangian/random-vs-apl.

cs.LG↗

arXiv2026

Near-Optimal Clustering in Mixture of Markov Chains

We study the problem of clustering $T$ trajectories of length $H$, each generated by one of K unknown ergodic Markov chains over a finite state space of size $S$. We derive an instance-dependent, high-probability lower bound on the clustering error rate, governed by the stationary-weighted KL divergence between transition kernels. We then propose a two-stage algorithm: Stage I applies spectral clustering via a new injective Euclidean embedding for ergodic Markov chains, a contribution of independent interest enabling sharp concentration results; Stage II refines clusters with a single likelihood-based reassignment step. We prove that our algorithm achieves near-optimal clustering error with high probability under reasonable requirements on $T$ and $H$. Preliminary experiments support our approach, and we conclude with discussions of its limitations and extensions.

stat.ML↗

arXiv2026

When to Ensemble: Identifying Token-Level Points for Stable and Fast LLM Ensembling

Ensembling Large Language Models (LLMs) has gained attention as a promising approach to surpass the performance of individual models by leveraging their complementary strengths. In particular, aggregating models' next-token probability distributions to select the next token has been shown to be effective in various tasks. However, while successful for short-form answers, its application to long-form generation remains underexplored. In this paper, we show that using existing ensemble methods in long-form generation requires a careful choice of ensembling positions, since the standard practice of ensembling at every token often degrades performance. We identify two key factors for determining the ensembling positions: tokenization mismatch across models and consensus in their next-token probability distributions. Based on this, we propose SAFE, (Stable And Fast LLM Ensembling), a framework that selectively ensembles by jointly considering these factors. To further improve stability, we apply a probability sharpening strategy when the ensemble distribution becomes overly smooth, enabling the selection of more confident tokens during ensembling. Our experiments on diverse benchmarks, including MATH500 and BBH, demonstrate that SAFE outperforms existing methods in both accuracy and efficiency, with gains achieved even when ensembling fewer than 1% of tokens.

cs.CL↗

arXiv2026

Mesoscopic Spin Coherence in a Disordered Dark Electron Spin Ensemble

Harnessing dipolar spin environments as controllable quantum resources is a central challenge in solid-state quantum technologies. Here, we report the observation of a coherent mesoscopic spin state in a disordered ensemble of substitutional nitrogen (P1) centers in diamond. An iterative Hartmann-Hahn protocol transfers polarization from dense nitrogen-vacancy (NV) centers to a P1 ensemble, yielding a 740-fold enhancement over room-temperature thermal equilibrium as revealed by differential readout. The resulting mesoscopic P1 spin ensemble exhibits collective Rabi oscillations and long-lived spin-lock and Hahn-echo coherences. We identify a crossover in the saturation polarization arising from the competition between coherent driving and local disorder, providing a quantitative measure of the system's intrinsic disorder. These results establish a foundation for utilizing dark electron spin ensembles as robust resources for quantum sensing and quantum many-body simulation.

quant-ph↗

arXiv2026

Learning to Reason in LLMs by Expectation Maximization

Large language models (LLMs) solve reasoning problems by first generating a rationale and then answering. We formalize reasoning as a latent variable model and derive a reward-based filtered expectation-maximization (FEM) objective for learning to reason. This view connects EM and modern reward-based optimization, and shows that the main challenge lies in designing a sampling distribution of rationales that justify correct answers. We instantiate and compare three sampling schemes: rejection sampling with a budget, self-taught reasoner (STaR), and prompt posterior sampling (PPS), which only keeps the rationalization stage of STaR that conditions on the correct answer in the prompt. We experiment with LLM-as-a-judge calibration and summarization from feedback tasks, where conditioning on the correct answer provides a strong guidance for generating rationales. Our experiments show the efficacy of PPS over other sampling schemes, and that the sampling scheme can have a significant impact on performance.

cs.LG↗

arXiv2025

Optimized Layerwise Approximation for Efficient Private Inference on Fully Homomorphic Encryption

Recent studies have explored the deployment of privacy-preserving deep neural networks utilizing homomorphic encryption (HE), especially for private inference (PI). Many works have attempted the approximation-aware training (AAT) approach in PI, changing the activation functions of a model to low-degree polynomials that are easier to compute on HE by allowing model retraining. However, due to constraints in the training environment, it is often necessary to consider post-training approximation (PTA), using the pre-trained parameters of the existing plaintext model without retraining. Existing PTA studies have uniformly approximated the activation function in all layers to a high degree to mitigate accuracy loss from approximation, leading to significant time consumption. This study proposes an optimized layerwise approximation (OLA), a systematic framework that optimizes both accuracy loss and time consumption by using different approximation polynomials for each layer in the PTA scenario. For efficient approximation, we reflect the layerwise impact on the classification accuracy by considering the actual input distribution of each activation function while constructing the optimization problem. Additionally, we provide a dynamic programming technique to solve the optimization problem and achieve the optimized layerwise degrees in polynomial time. As a result, the OLA method reduces inference times for the ResNet-20 model and the ResNet-32 model by 3.02 times and 2.82 times, respectively, compared to prior state-of-the-art implementations employing uniform degree polynomials. Furthermore, we successfully classified CIFAR-10 by replacing the GELU function in the ConvNeXt model with only 3-degree polynomials using the proposed method, without modifying the backbone model.

cs.CR↗

arXiv2025

AdaSTaR: Adaptive Data Sampling for Training Self-Taught Reasoners

Self-Taught Reasoners (STaR), synonymously known as Rejection sampling Fine-Tuning (RFT), is an integral part of the training pipeline of self-improving reasoning Language Models (LMs). The self-improving mechanism often employs random observation (data) sampling. However, this results in trained observation imbalance; inefficiently over-training on solved examples while under-training on challenging ones. In response, we introduce Adaptive STaR (AdaSTaR), a novel algorithm that rectifies this by integrating two adaptive sampling principles: (1) Adaptive Sampling for Diversity: promoting balanced training across observations, and (2) Adaptive Sampling for Curriculum: dynamically adjusting data difficulty to match the model's evolving strength. Across six benchmarks, AdaSTaR achieves best test accuracy in all instances (6/6) and reduces training FLOPs by an average of 58.6% against an extensive list of baselines. These improvements in performance and efficiency generalize to different pre-trained LMs and larger models, paving the way for more efficient and effective self-improving LMs.

cs.LG↗

arXiv2025

Coherent control of solid-state defect spins via patterned boron-doped diamond circuit

Monolithic integration, which refers to the incorporation of all device functionalities within a single material, shows significant potential for creating scalable solid-state quantum devices. This study demonstrated the coherent control of nitrogen-vacancy (NV) spins using an electronic circuit monolithically integrated within diamond: a patterned, conductive boron-doped diamond (BDD) microwave waveguide. First, we validated the high-frequency performance of the circuit by characterizing its impedance up to the microwave range, confirming its capability for efficient microwave transmission. Then, using this monolithically integrated BDD--NV hybrid system, we performed optically detected magnetic resonance and observed noticeable Rabi oscillations driven by the metallic BDD circuit. Importantly, we verified that the BDD antenna has a minimal detrimental impact on the NV spins; microwave-induced heating is negligible under both pulsed and continuous driving, and the spin relaxation time ($T_1$) remains unperturbed. This approach paves the way for a new class of compact, robust, and versatile quantum platforms suitable for sensing and information processing in various environments.

physics.app-ph↗

arXiv2025

FlickerFusion: Intra-trajectory Domain Generalizing Multi-Agent RL

Multi-agent reinforcement learning has demonstrated significant potential in addressing complex cooperative tasks across various real-world applications. However, existing MARL approaches often rely on the restrictive assumption that the number of entities (e.g., agents, obstacles) remains constant between training and inference. This overlooks scenarios where entities are dynamically removed or added during the inference trajectory -- a common occurrence in real-world environments like search and rescue missions and dynamic combat situations. In this paper, we tackle the challenge of intra-trajectory dynamic entity composition under zero-shot out-of-domain (OOD) generalization, where such dynamic changes cannot be anticipated beforehand. Our empirical studies reveal that existing MARL methods suffer significant performance degradation and increased uncertainty in these scenarios. In response, we propose FlickerFusion, a novel OOD generalization method that acts as a universally applicable augmentation technique for MARL backbone methods. FlickerFusion stochastically drops out parts of the observation space, emulating being in-domain when inferenced OOD. The results show that FlickerFusion not only achieves superior inference rewards but also uniquely reduces uncertainty vis-à-vis the backbone, compared to existing methods. Benchmarks, implementations, and model weights are organized and open-sourced at flickerfusion305.github.io, accompanied by ample demo video renderings.

cs.LG↗

arXiv2025

Probability-Flow ODE in Infinite-Dimensional Function Spaces

Recent advances in infinite-dimensional diffusion models have demonstrated their effectiveness and scalability in function generation tasks where the underlying structure is inherently infinite-dimensional. To accelerate inference in such models, we derive, for the first time, an analog of the probability-flow ODE (PF-ODE) in infinite-dimensional function spaces. Leveraging this newly formulated PF-ODE, we reduce the number of function evaluations while maintaining sample quality in function generation tasks, including applications to PDEs.

cs.LG↗

arXiv2025

Quantum decoherence of nitrogen-vacancy spin ensembles in a nitrogen spin bath in diamond under dynamical decoupling

The negatively charged nitrogen-vacancy (NV) center in diamond has emerged as a leading qubit platform for quantum technology applications. One of the key challenges for NV-based quantum applications is building an accurate model to predict its decoherence properties and their quantum nature. In this study, we combine theory and experiment to investigate NV decoherence dynamics in the presence of nitrogen donor (P1 center) baths, which is one of the dominant decoherence sources in diamond. We employ a cluster-correlation expansion (CCE) method to compute the NV decoherence under the Hahn-echo (HE) and Carr-Purcell-Meiboom-Gill (CPMG) pulse sequences at various P1 concentrations from 1ppm to 300 ppm. We show that the coherence time (T2) increases with the number of pi pulses applied, indicating that the NV spin is decoupled from the P1 bath. Notably, we find that T2 scales quadratically as a function of the pulse number, on a logarithmic scale, as opposed to the linear scaling predicted by widely accepted semi-classical theories in the literature. In our experiment, we measure the CPMG signal for two diamond samples with high P1 concentrations of 0.8ppm and 13ppm. We demonstrate that the T2 scaling is indeed quadratic, thus confirming our theoretical predictions. Our results show that the quantum bath model combined with the CCE method can accurately capture the quantum nature of the P1-driven NV decoherence. Our study opens a new avenue for developing a complete noise model that could be used to optimize the performance of NV-based quantum devices.

quant-ph↗