arXiv subjects

Sanghyun Park

Publications and source records attributed to Sanghyun Park.

At least 19 recordsLinked to original sources

arXiv2026

DELTA: Deformable Elevation-Based Local Terrain Attention Encoder for Sparse-Terrain Quadrupedal Locomotion

Stable quadrupedal locomotion on sparse terrain requires selecting state-relevant terrain evidence for precise foot placement. Model-based foothold planners provide precise foothold selection but rely heavily on explicit model assumptions. Recent attention-based map encoding (AME) studies show that end-to-end reinforcement learning (RL) can learn implicit foothold guidance. However, the computational cost of dense AME encoding grows with map resolution, limiting its scalability to fine-grained sparse terrain. We propose DELTA, a Deformable Elevation-Based Local Terrain Attention encoder. DELTA predicts state-conditioned sampling locations, forms terrain evidence tokens from adaptive local elevation patches, and attends only to a fixed-size token set. With fixed sampling and patch settings, DELTA's encoder cost is independent of map resolution. Experiments show that DELTA achieves final traversal performance comparable to AME at the standard resolution while improving learning efficiency. This fixed encoder cost enables the use of higher-resolution terrain maps, improving traversal on fine-grained sparse terrain. DELTA also demonstrates strong generalization to unseen mixed evaluation courses composed of continuous and discrete terrain elements. Beyond simulation, DELTA demonstrates successful sim-to-real transfer on RAIBO2. Analysis of the learned sampling offsets and attention weights shows that DELTA samples steppable regions and attends to terrain evidence relevant to future touchdowns without foothold labels or attention supervision.

cs.RO↗

arXiv2026

GranQ: Efficient Channel-wise Quantization via Vectorized Pre-Scaling for Zero-Shot QAT

Zero-shot quantization (ZSQ) enables neural network compression without original training data, making it a promising solution for restricted data access scenarios. To compensate for the lack of data, recent ZSQ methods typically rely on synthetic inputs generated from the full-precision model. However, these synthetic inputs often lead to activation distortion, especially under low-bit settings. To mitigate this, existing methods typically employ per-channel scaling, but they still struggle due to the severe computational overhead during the accumulation process. To overcome this critical bottleneck, we propose GranQ, a novel activation quantization framework that introduces an efficient pre-scaling strategy. Unlike conventional channel-wise methods that repeatedly perform scaling operations during accumulation, GranQ applies scaling factors in a pre-scaling step through fully vectorized computation, eliminating runtime scaling overhead. This design enables GranQ to maintain fine-grained quantization accuracy while significantly reducing computational burden, particularly in low-bit quantization settings. Extensive experiments under quantization-aware training (QAT) settings demonstrate that GranQ consistently outperforms state-of-the-art ZSQ methods across CIFAR and ImageNet. In particular, our method achieves up to 5.45% higher accuracy in the 3-bit setting on CIFAR-100 and even surpasses the full-precision baseline on CIFAR-10.

cs.CV↗

arXiv2026

AsyncOPD: How Stale Can On-Policy Distillation Be?

On-policy distillation (OPD) trains a student on its own rollouts guided by teacher feedback and is becoming increasingly important for large language model (LLM) post-training. Like reinforcement learning (RL), however, OPD faces an on-policy systems bottleneck, as rollouts can dominate training time for reasoning workloads. Asynchronous training pipelines can alleviate this bottleneck by decoupling rollout generation from learner updates, but doing so introduces stale-policy data. While prior work has studied stale data in asynchronous RL, its effects in OPD remain underexplored. We present the first systematic study of staleness in asynchronous OPD, focusing on a practical setting where teacher feedback is implemented through local KL losses and full-vocabulary teacher logits are too expensive to store or transfer, necessitating finite teacher-score caches. We first show that KL direction changes the stale-data problem: teacher-weighted forward KL is more robust to stale rollouts, whereas student-weighted reverse KL is vulnerable. Second, for this vulnerable reverse-KL case, we study whether methods designed to stabilize asynchronous RL can mitigate OPD staleness. In our experiments, they do not improve over a simpler OPD-specific surrogate: recomputing the reverse-KL signal under the current student at learner time. Third, we analyze how finite teacher-score caches create a bias-variance tradeoff for sparse and sampled reverse-KL OPD estimators. This motivates multi-sample Monte Carlo (MC), which preserves MC correctability while reducing one-sample variance. Finally, we present and open-source AsyncOPD, a fully asynchronous OPD training pipeline built from these estimator choices. Experiments show that AsyncOPD improves training throughput by $1.6\times$ to $3.8\times$ over strict synchronous training while reaching comparable accuracy.

cs.LG↗

arXiv2026

GenZ-LIO: Generalizable LiDAR-Inertial Odometry Beyond Confined--Open Boundaries

For field robotic missions such as inspection, search-and-rescue, and exploration, light detection and ranging (LiDAR)-inertial odometry (LIO) can serve as a core component of autonomy by providing localization and mapping in GNSS-denied or unstructured environments. However, transitions between confined and open spaces, which are commonly encountered in field deployments, can induce substantial changes in scan density and local geometric structure, thereby reducing the robustness and computational efficiency of LIO. To address these issues, we present GenZ-LIO, a generalizable LIO framework designed to adapt to variations in spatial scale across confined and open environments. GenZ-LIO comprises three components: (i) scale-aware adaptive voxelization for regulating scan downsampling across spatial scale changes, (ii) hybrid-metric state update for combining point-to-plane and point-to-point residuals under varying geometric structure, and (iii) voxel-pruned correspondence search for efficient point-to-point matching. We conduct a comprehensive evaluation using 42 sequences from nine public datasets and our newly collected NarrowWide dataset to analyze LIO performance under spatial scale variations across diverse field scenarios. Across the evaluated sequences, GenZ-LIO maintains stable odometry estimation without divergence, indicating practical robustness under the tested field conditions. The source code and collected dataset will be made publicly available upon publication.

cs.RO↗

arXiv2026

GeneralThinker: Domain-General Reasoning through Likelihood-Guided Answer-Conditioned Optimization

Reinforcement learning with verifiable rewards improves language model reasoning, but its reliance on domain-specific verifiers, sparse outcome rewards, and coarse-grained credit assignment limits its applicability. We introduce GeneralThinker, an on-policy framework that reformulates reasoning supervision as dense answer-conditioned optimization, enabling response-level evaluation and token-level credit assignment without domain-specific verifiers. GeneralThinker evaluates generated reasoning trajectories using the likelihood of the ground-truth answer and derives token-wise compatibility signals for fine-grained credit assignment. To stabilize optimization, it constrains token-level updates through clipping and direction-preserving modulation. Across 11 benchmarks spanning mathematics, STEM, and general reasoning, GeneralThinker achieves the best average performance. Further analyses show that uncontrolled token-level modulation can destabilize training, whereas controlled modulation makes fine-grained credit assignment consistently effective.

cs.CL↗

arXiv2026

SpiralThinker: Latent Reasoning through an Iterative Process with Text-Latent Interleaving

Recent advances in large reasoning models have been driven by reinforcement learning and test-time scaling, accompanied by growing interest in latent rather than purely textual reasoning. However, existing latent reasoning methods lack mechanisms to ensure stable reasoning dynamics in latent space and a systematic way to interleave implicit and explicit reasoning. We introduce SpiralThinker, a stabilized iterative latent reasoning framework that performs iterative updates over latent representations while interleaving latent and textual reasoning steps. At its core, it combines a progressive alignment objective that explicitly regulates latent representations across iterations with structured annotations for text-latent interleaving, thereby stabilizing latent updates and maintaining coherence with textual reasoning. Across mathematical, logical, and commonsense reasoning tasks, SpiralThinker achieves state-of-the-art performance among latent reasoning baselines. Further analysis shows that both iteration and alignment are essential, that the optimal numbers of latent tokens and iterations vary by dataset, and that proper alignment is crucial for effective iterative latent reasoning. Overall, SpiralThinker bridges iterative computation and latent reasoning, demonstrating that aligned iterative updates can reliably steer reasoning in the latent space.

cs.CL↗

arXiv2026

ALIVE-LIO: Degeneracy-Aware Learning of Inertial Velocity for Enhancing ESKF-Based LiDAR-Inertial Odometry

Odometry estimation using light detection and ranging (LiDAR) and an inertial measurement unit (IMU), known as LiDAR-inertial odometry (LIO), often suffers from performance degradation in degenerate environments, such as long corridors or single-wall scenarios with narrow field-of-view LiDAR. To address this limitation, we propose ALIVE-LIO, a degeneracy-aware LiDAR-inertial odometry framework that explicitly enhances state estimation in degenerate directions. The key contribution of ALIVE-LIO is the strategic integration of a deep neural network into a classical error-state Kalman filter (ESKF) to compensate for the loss of LiDAR observability. Specifically, ALIVE-LIO employs a neural network to predict the body-frame velocity and selectively fuses this prediction into the ESKF only when degeneracy is detected, providing effective state updates along degenerate directions. This design enables ALIVE-LIO to utilize the probabilistic structure and consistency of the ESKF while benefiting from learning-based motion estimation. The proposed method was evaluated on publicly available datasets exhibiting degeneracy, as well as on our own collected data. Experimental results demonstrate that ALIVE-LIO substantially reduces pose drift in degenerate environments, yielding the most competitive results in 22 out of 32 sequences. The implementation of ALIVE-LIO will be publicly available.

cs.RO↗

arXiv2026

Let Triggers Control: Frequency-Aware Dropout for Effective Token Control

Text-to-image models such as Stable Diffusion have achieved unprecedented levels of high-fidelity visual synthesis. As these models advance, personalization of generative models -- commonly facilitated through Low-Rank Adaptation (LoRA) with a dedicated trigger token -- has become a significant area of research. Previous works have naively assumed that fine-tuning with a single trigger token to represent new concepts. However, this often results in poor controllability, where the trigger token alone fails to reliably evoke the intended concept. We attribute this issue to the frequent co-occurrence of the trigger token with the surrounding context during fine-tuning, which entangles their representations and compromises the token's semantic distinctiveness. To disentangle this, we propose Frequency-Aware Dropout (FAD) -- a novel regularization technique that improves prompt controllability without adding new parameters. FAD consists of two key components: co-occurrence analysis and curriculum-inspired scheduling. Qualitative and quantitative analyses across token-based diffusion models (SD~1.5 and SDXL) and natural language--driven backbones (FLUX and Qwen-Image) demonstrate consistent gains in prompt fidelity, stylistic precision, and user-perceived quality. Our method provides a simple yet effective dropout strategy that enhances controllability and personalization in text-to-image generation. Notably, it achieves these improvements without introducing additional parameters or architectural modifications, making it readily applicable to existing models with minimal computational overhead.

cs.CV↗

arXiv2026

Benchmarking Visual Feature Representations for LiDAR-Inertial-Visual Odometry Under Challenging Conditions

Accurate localization in autonomous driving is critical for successful missions including environmental mapping and survivor searches. In visually challenging environments, including low-light conditions, overexposure, illumination changes, and high parallax, the performance of conventional visual odometry methods significantly degrade undermining robust robotic navigation. Researchers have recently proposed LiDAR-inertial-visual odometry (LIVO) frameworks, that integrate LiDAR, IMU, and camera sensors, to address these challenges. This paper extends the FAST-LIVO2-based framework by introducing a hybrid approach that integrates direct photometric methods with descriptor-based feature matching. For the descriptor-based feature matching, this work proposes pairs of ORB with the Hamming distance, SuperPoint with SuperGlue, SuperPoint with LightGlue, and XFeat with the mutual nearest neighbor. The proposed configurations are benchmarked by accuracy, computational cost, and feature tracking stability, enabling a quantitative comparison of the adaptability and applicability of visual descriptors. The experimental results reveal that the proposed hybrid approach outperforms the conventional sparse-direct method. Although the sparse-direct method often fails to converge in regions where photometric inconsistency arises due to illumination changes, the proposed approach still maintains robust performance under the same conditions. Furthermore, the hybrid approach with learning-based descriptors enables robust and reliable visual state estimation across challenging environments.

cs.RO↗

arXiv2026

ROFT-VINS: Robust Feature Tracking-based Visual-Inertial State Estimation for Harsh Environment

SLAM (Simultaneous Localization and Mapping) and Odometry are important systems for estimating the position of mobile devices, such as robots and cars, utilizing one or more sensors. Particularly in camera-based SLAM or Odometry, effectively tracking visual features is important as it significantly impacts system performance. In this paper, we propose a method that leverages deep learning to robustly track visual features in monocular camera images. This method operates reliably even in textureless environments and situations with rapid lighting changes. Additionally, we evaluate the performance of our proposed method by integrating it into VINS-Fusion (Monocular-Inertial), a commonly used Visual-Inertial Odometry (VIO) system.

cs.RO↗

arXiv2026

Motion-Specific Battery Health Assessment for Quadrotors Using High-Fidelity Battery Models

Quadrotor endurance is ultimately limited by battery behavior, yet most energy aware planning treats the battery as a simple energy reservoir and overlooks how flight motions induce dynamic current loads that accelerate battery degradation. This work presents an end to end framework for motion aware battery health assessment in quadrotors. We first design a wide range current sensing module to capture motion specific current profiles during real flights, preserving transient features. In parallel, a high fidelity battery model is calibrated using reference performance tests and a metaheuristic based on a degradation coupled electrochemical model.By simulating measured flight loads in the calibrated model, we systematically resolve how different flight motions translate into degradation modes loss of lithium inventory and loss of active material as well as internal side reactions. The results demonstrate that even when two flight profiles consume the same average energy, their transient load structures can drive different degradation pathways, emphasizing the need for motion-aware battery management that balances efficiency with battery degradation.

cs.RO↗

arXiv2026

LatentTune: Efficient Tuning of High Dimensional Database Parameters via Latent Representation Learning

As data volumes continue to grow, optimizing database performance has become increasingly critical, making the implementation of effective tuning methods essential. Among various approaches, database parameter tuning has proven to be a highly effective means of enhancing performance. Recent studies have shown that machine learning techniques can successfully optimize database parameters, leading to significant performance improvements. However, existing methods still face several limitations. First, they require substantial time to generate large training datasets. Second, to cope with the challenges of highdimensional optimization, they typically optimize only a subset of parameters rather than the full configuration space. Third, they often rely on information from similar workloads instead of directly leveraging information from the target workload. To address these limitations, we propose LatentTune, a novel approach that differs fundamentally from traditional methods. To reduce the time required for data generation, LatentTune incorporates a data augmentation strategy. Furthermore, it constructs a latent space that compresses information from all database parameters, enabling the optimization of the full configuration space. In addition, LatentTune integrates external metric information into the latent space, allowing for precise tuning tailored to the actual target workload. Experimental results demonstrate that LatentTune outperforms baseline models across four workloads on MySQL and RocksDB, achieving up to 1332% improvement for RocksDB and 11.82% throughput gain with 46.01% latency reduction for MySQL.

cs.DB↗

arXiv2026

LitE-SQL: A Lightweight and Efficient Text-to-SQL Framework with Vector-based Schema Linking and Execution-Guided Self-Correction

The Text-to-SQL task translates natural language questions into SQL queries, enabling intuitive database interaction for non-experts. While recent methods leveraging Large Language Models (LLMs) achieve strong performance, their reliance on proprietary models raise concerns about deployment feasibility and data privacy. In this work, we introduce LitE-SQL, a Lightweight and Efficient framework with two components: (i) a Schema Retriever that performs efficient schema linking using a vector database of pre-computed schema embeddings, optimized with a hard-negative supervised contrastive objective to distinguish semantically similar but functionally irrelevant columns, and (ii) a SQL Generator fine-tuned in two stages-supervised fine-tuning followed by execution-guided reinforcement-enabling execution-guided self-correction without multi-candidate sampling, which is commonly required by prior LLM-based approaches. On BIRD, LitE-SQL achieves 72.10% execution accuracy, and on Spider 1.0 it reaches 88.45%, demonstrating comparable or superior performance to LLM-based methods despite using 2x to 30x fewer parameters. Our findings demonstrate that high-quality Text-to-SQL generation is feasible with lightweight models, offering a practical solution for privacy-sensitive and resource-constrained settings.

cs.CL↗

arXiv2025

Vertical NAND in a Ferroelectric-driven Paradigm Shift

Over the past decades, the relentless scaling and mass production of flash memory have underpinned the data-centric era. Yet charge-trap-based 3D NAND flash is now constrained by intrinsic physical and architectural limits, including reliability degradation at the device level, high operating power at the array level, and vertical scaling saturation at the system level. These bottlenecks hinder further advances in storage density and energy efficiency required by memory-centric computing. This Perspective outlines how coupling ferroelectric polarization with charge trapping can reconfigure the foundations of flash memory. In these hybrid architectures, polarization offers an energy-efficient pathway for charge modulation through enhanced Fowler-Nordheim tunneling, while trapped charges reinforce polarization-driven states to ensure stability. Such synergistic dynamics enable low-voltage operation and integration beyond one thousand layers without compromising process compatibility. We discuss the material, device, and architectural transitions required to realize this hybrid technology and chart future research directions to overcome the remaining scaling bottlenecks. Hybrid ferroelectric NAND extends conventional flash toward a scalable and energy-efficient platform, marking a paradigm shift for next-generation non-volatile memory.

physics.app-ph↗

arXiv2025

AnyAnomaly: Zero-Shot Customizable Video Anomaly Detection with LVLM

Video anomaly detection (VAD) is crucial for video analysis and surveillance in computer vision. However, existing VAD models rely on learned normal patterns, which makes them difficult to apply to diverse environments. Consequently, users should retrain models or develop separate AI models for new environments, which requires expertise in machine learning, high-performance hardware, and extensive data collection, limiting the practical usability of VAD. To address these challenges, this study proposes customizable video anomaly detection (C-VAD) technique and the AnyAnomaly model. C-VAD considers user-defined text as an abnormal event and detects frames containing a specified event in a video. We effectively implemented AnyAnomaly using a context-aware visual question answering without fine-tuning the large vision language model. To validate the effectiveness of the proposed model, we constructed C-VAD datasets and demonstrated the superiority of AnyAnomaly. Furthermore, our approach showed competitive results on VAD benchmarks, achieving state-of-the-art performance on UBnormal and UCF-Crime and surpassing other methods in generalization across all datasets. Our code is available online at github.com/SkiddieAhn/Paper-AnyAnomaly.

cs.CV↗

arXiv2025

Relation-Aware Bayesian Optimization of DBMS Configurations Guided by Affinity Scores

Database Management Systems (DBMSs) are fundamental for managing large-scale and heterogeneous data, and their performance is critically influenced by configuration parameters. Effective tuning of these parameters is essential for adapting to diverse workloads and maximizing throughput while minimizing latency. Recent research has focused on automated configuration optimization using machine learning; however, existing approaches still exhibit several key limitations. Most tuning frameworks disregard the dependencies among parameters, assuming that each operates independently. This simplification prevents optimizers from leveraging relational effects across parameters, limiting their capacity to capture performancesensitive interactions. Moreover, to reduce the complexity of the high-dimensional search space, prior work often selects only the top few parameters for optimization, overlooking others that contribute meaningfully to performance. Bayesian Optimization (BO), the most common method for automatic tuning, is also constrained by its reliance on surrogate models, which can lead to unstable predictions and inefficient exploration. To overcome these limitations, we propose RelTune, a novel framework that represents parameter dependencies as a Relational Graph and learns GNN-based latent embeddings that encode performancerelevant semantics. RelTune further introduces Hybrid-Score-Guided Bayesian Optimization (HBO), which combines surrogate predictions with an Affinity Score measuring proximity to previously high-performing configurations. Experimental results on multiple DBMSs and workloads demonstrate that RelTune achieves faster convergence and higher optimization efficiency than conventional BO-based methods, achieving state-of-the-art performance across all evaluated scenarios.

cs.LG↗

arXiv2025

Thinking with Many Minds: Using Large Language Models for Multi-Perspective Problem-Solving

Complex problem-solving requires cognitive flexibility--the capacity to entertain multiple perspectives while preserving their distinctiveness. This flexibility replicates the "wisdom of crowds" within a single individual, allowing them to "think with many minds." While mental simulation enables imagined deliberation, cognitive constraints limit its effectiveness. We propose synthetic deliberation, a Large Language Model (LLM)-based method that simulates discourse between agents embodying diverse perspectives, as a solution. Using a custom GPT-based model, we showcase its benefits: concurrent processing of multiple viewpoints without cognitive degradation, parallel exploration of perspectives, and precise control over viewpoint synthesis. By externalizing the deliberative process and distributing cognitive labor between parallel search and integration, synthetic deliberation transcends mental simulation's limitations. This approach shows promise for strategic planning, policymaking, and conflict resolution.

cs.CL↗

arXiv2025

Controllable 3D Molecular Generation for Structure-Based Drug Design Through Bayesian Flow Networks and Gradient Integration

Recent advances in Structure-based Drug Design (SBDD) have leveraged generative models for 3D molecular generation, predominantly evaluating model performance by binding affinity to target proteins. However, practical drug discovery necessitates high binding affinity along with synthetic feasibility and selectivity, critical properties that were largely neglected in previous evaluations. To address this gap, we identify fundamental limitations of conventional diffusion-based generative models in effectively guiding molecule generation toward these diverse pharmacological properties. We propose CByG, a novel framework extending Bayesian Flow Network into a gradient-based conditional generative model that robustly integrates property-specific guidance. Additionally, we introduce a comprehensive evaluation scheme incorporating practical benchmarks for binding affinity, synthetic feasibility, and selectivity, overcoming the limitations of conventional evaluation methods. Extensive experiments demonstrate that our proposed CByG framework significantly outperforms baseline models across multiple essential evaluation criteria, highlighting its effectiveness and practicality for real-world drug discovery applications.

cs.LG↗