arXiv subjects

Bo Yang

Publications and source records attributed to Bo Yang.

At least 73 records · Page 4Linked to original sources

arXiv2026

Detecting Is Not Resolving: The Monitoring Control Gap in Retrieval Augmented LLMs

Retrieval-augmented LLMs are deployed for tasks where evidence quality determines action safety, yet evaluation protocols assume that single-turn robustness predicts robustness when evidence accumulates across turns. We show this assumption is fundamentally incorrect. Models exhibit a monitoring-control gap: they readily acknowledge contradictory evidence, yet this awareness fails to constrain their final recommendations - detecting epistemic conflict does not imply resolving it safely. Through a multi-turn document accumulation protocol across four model families (1.5B-32B parameters) and over 50,000 turn-level evaluations, we demonstrate that single-turn diagnostics systematically overestimate RAG safety, that contradiction acknowledgement is uncorrelated with safe resolution, a pattern corroborated by targeted human validation, and that no universal prompt fix exists. Converging mechanism evidence - hidden-state probing, attention analysis, and response-strategy taxonomy - points to action selection as the most plausible locus of the deficit: danger-relevant information is internally represented and receives enhanced attention during unsafe generation, yet fails to constrain output behavior. The gap between what models recognize and what they do must be measured and closed before retrieval-augmented systems can be trusted in high-stakes settings.

cs.AI↗

arXiv2026

FoundObj: Self-supervised Foundation Models as Rewards for Label-free 3D Object Segmentation

We address the challenging task of 3D object segmentation in complex scene point clouds without relying on any scene-level human annotations during training. Existing methods are typically constrained to identifying simple objects, primarily due to insufficient object priors in the learning process. In this paper, we present FoundObj, a novel framework featuring a superpoint-based object discovery agent that incrementally merges suitable neighboring superpoints, guided by our innovative semantic and geometric reward modules. These modules synergistically leverage semantic and geometric priors from self-supervised 2D/3D foundation models, providing complementary feedback to the object discovery agent and enabling robust identification of multi-class objects through reinforcement learning. Extensive experiments on diverse benchmarks demonstrate that our approach consistently outperforms existing baselines. Notably, our method exhibits strong generalization in zero-shot and long-tail scenarios, underscoring its potential for scalable, label-free 3D object segmentation.

cs.CV↗

arXiv2026

Beyond VLM-Based Rewards: Diffusion-Native Latent Reward Modeling

Preference optimization for diffusion and flow-matching models relies on reward functions that are both discriminatively robust and computationally efficient. Vision-Language Models (VLMs) have emerged as the primary reward provider, leveraging their rich multimodal priors to guide alignment. However, their computation and memory cost can be substantial, and optimizing a latent diffusion generator through a pixel-space reward introduces a domain mismatch that complicates alignment. In this paper, we propose DiNa-LRM, a diffusion-native latent reward model that formulates preference learning directly on noisy diffusion states. Our method introduces a noise-calibrated Thurstone likelihood with diffusion-noise-dependent uncertainty. DiNa-LRM leverages a pretrained latent diffusion backbone with a timestep-conditioned reward head, and supports inference-time noise ensembling, providing a diffusion-native mechanism for test-time scaling and robust rewarding. Across image alignment benchmarks, DiNa-LRM substantially outperforms existing diffusion-based reward baselines and achieves performance competitive with state-of-the-art VLMs at a fraction of the computational cost. In preference optimization, we demonstrate that DiNa-LRM improves preference optimization dynamics, enabling faster and more resource-efficient model alignment.

cs.CV↗

arXiv2026

$π$MPC: A Parallel-in-horizon and Construction-free NMPC Solver

The alternating direction method of multipliers (ADMM) has gained increasing popularity in embedded model predictive control (MPC) due to its code simplicity and pain-free parameter selection. However, existing ADMM solvers either target general quadratic programming (QP) problems or exploit sparse MPC formulations via Riccati recursions, which are inherently sequential and therefore difficult to parallelize for long prediction horizons. This technical note proposes a novel \textit{parallel-in-horizon} and \textit{construction-free} nonlinear MPC algorithm, termed $π$MPC, which combines a new variable-splitting scheme with a velocity-based system representation in the ADMM framework, enabling horizon-wise parallel execution while operating directly on system matrices without explicit MPC-to-QP construction. Numerical experiments and accompanying code are provided to validate the effectiveness of the proposed method.

math.OC↗

arXiv2026

Expectation Consistency Loss: Rethink Confidence Calibration under Covariate Shift

Confidence calibration for classification models is vital in safety-critical decision-making scenarios and has received extensive attention. General confidence calibration methods assume training and test data are independent and identically distributed, limiting their effectiveness under covariate shifts. Previous calibration methods under covariate shift struggle with class-wise or canonical calibrations and often rely on unstable importance weighting when density ratios are large or unbounded. Given the above limitations, this paper rethinks confidence calibration under covariate shifts. First, we derive a necessary and sufficient condition for confidence calibration under covariate shifts, named Expectation consistency condition, which reveals covariate shifts do not necessarily lead to uncalibrated confidence and provides a weaker condition for confidence calibration than global covariate distribution alignment. Then, utilizing Expectation consistency condition, this paper proposes an unsupervised domain adaptation loss to calibrate confidence of the target domain, named Expectation consistency loss (ECL), which is compatible with canonical calibration, class-wise calibration, and top-label calibration. Third, we prove that computing ECL loss has the same sample complexity as Expected Calibration Error (ECE) and provide a theoretically grounded mini-batch trainable scheme for ECL loss. Finally, we validate the effectiveness of our method on both simulated and real-world covariate shift datasets.

cs.LG↗

arXiv2026

Beyond Commutativity: Redesigning Trotter Decomposition via Local Symmetry

The product formula, commonly known as Trotter decomposition, is a central tool for digital quantum simulation, whose performance depends critically on how the Hamiltonian is partitioned into tractable blocks. Standard decompositions typically rely on direct commutativity among Hamiltonian terms in a chosen operator representation, which can lead to large residual errors and deep circuits for complex, practically relevant many-body quantum systems. We address this fundamental bottleneck by introducing a new decomposition principle that goes beyond commutativity, grouping Hamiltonian terms into local three-site clusters according to the underlying SU(2) symmetry of the local dynamics. We show that three-site generators fall into at most four SU(2)-symmetry classes, each admitting an effective two-qubit SU(4) representation with exact and efficient implementations. By reducing the number of clusters, this decomposition principle substantially suppresses commutator-induced errors and circuit overhead while preserving underlying physical structures that commutativity-based decompositions may violate. We demonstrate the proposed method on several physically relevant spin-lattice models, where the reduced cluster structure can even realise the second-order product formula without doubling the circuit depth, as would be required by conventional decompositions. Numerical simulations of a Kagome Heisenberg model with triangular spin-chirality interactions show that the proposed method reduces both state infidelity and average spin-chirality bias by more than three orders of magnitude compared with conventional decompositions, while using substantially fewer gates. These results establish local symmetry as a flexible and practical design principle for product-formula simulation, opening a route to more accurate and hardware-efficient simulations of broader classes of many-body systems.

quant-ph↗

arXiv2026

EvObj: Learning Evolving Object-centric Representations for 3D Instance Segmentation without Scene Supervision

We introduce EvObj for unsupervised 3D instance segmentation that bridges the geometric domain gap between synthetic pretraining data and real-world point clouds. Current methods suffer from structural discrepancies when transferring object priors from synthetic datasets (e.g., ShapeNet) to real scans (e.g., ScanNet), particularly due to morphological variations and occlusion artifacts. To address this, EvObj integrates two innovative modules: (1) An object discerning module that dynamically refines object candidates, enabling continuous adaptation of object priors to target domains; and (2) An object completion module that reconstructs partial geometries after discovering objects. We conduct extensive experiments on both real-world and synthetic datasets, demonstrating superior 3D object segmentation performance over all baselines while achieving state-of-the-art results.

cs.CV↗

arXiv2026

Value-Decomposed Reinforcement Learning Framework for Taxiway Routing with Hierarchical Conflict-Aware Observations

Taxiway routing and on-surface conflict avoidance are coupled safety-critical decision problems in airport surface operations. Existing planning and optimization methods are often limited by online computational cost, while reinforcement learning methods may struggle to represent downstream traffic conflicts and balance multiple objectives. This paper presents Conflict-aware Taxiway Routing (CaTR), a reinforcement learning framework for real-time multi-aircraft taxiway routing. CaTR constructs a grid-based airport surface environment with action masking, introduces a hierarchical foresight traffic representation to encode current and downstream conflict-related traffic conditions, and adopts a value-decomposed reinforcement learning strategy to prioritize sparse but safety-critical objectives. Experiments are conducted on a realistic environment based on Changsha Huanghua International Airport under multiple traffic density levels. Results show that CaTR achieves better safety--efficiency trade-offs than representative planning, optimization, and reinforcement learning baselines while maintaining practical runtime.

cs.AI↗

arXiv2026

Uncertainty-Aware Trip Purpose Inference from GPS Trajectories via POI Semantic Zones and Pareto Calibration

Large-scale GPS trajectory data offer rich observations of human mobility, yet assigning trip purposes to detected stops remains challenging due to the absence of individual-level ground truth, spatial uncertainty from GPS noise and incomplete points of interest (POIs) coverage, and fundamental behavioral differences across trip purposes. We propose a weakly supervised framework integrating neighborhood-level POI semantic zones with distance-weighted spatial likelihoods, differentiated inference strategies for mandatory and non-mandatory activities, and a multi-phase Pareto optimization that jointly minimizes distributional divergence from household travel survey statistics and maximizes inference reliability without requiring annotated labels. Evaluated on over 81 million staypoints in Los Angeles, the framework reduces activity type frequency Jensen-Shannon distance (JSD) by 23%, start time JSD by 48%, and duration JSD by 12% respectively relative to a comparable baseline. The proposed approach provides a scalable and uncertainty-aware path from raw GPS trajectories to semantically annotated mobility data for travel demand modeling and transportation policy analysis.

cs.AI↗

arXiv2026

Hierarchical Learning for IRS-Assisted MEC Systems with Rate-Splitting Multiple Access

Intelligent reflecting surface (IRS)-assisted mobile edge computing (MEC) systems have shown notable improvements in efficiency, such as reduced latency, higher data rates, and better energy efficiency. However, the resource competition among users will lead to uneven allocation, increased latency, and lower throughput. Fortunately, the rate-splitting multiple access (RSMA) technique has emerged as a promising solution for managing interference and optimizing resource allocation in MEC systems. This paper studies an IRS-assisted MEC system with RSMA, aiming to jointly optimize the passive beamforming of the IRS, the active beamforming of the base station, the task offloading allocation, the transmit power of users, the ratios of public and private information allocation, and the decoding order of the RSMA to minimize the average delay from a novel uplink transmission perspective. Since the formulated problem is non-convex and the optimization variables are highly coupled, we propose a hierarchical deep reinforcement learning-based algorithm to optimize both continuous and discrete variables of the problem. Additionally, to better extract channel features, we design a novel network architecture within the policy and evaluation networks of the proposed algorithm, combining convolutional neural networks and densely connected convolutional network for feature extraction. Simulation results indicate that the proposed algorithm not only exhibits excellent convergence performance but also outperforms various benchmarks.

eess.SP↗

arXiv2026

Global well-posedness and scattering for defocusing energy-critical inhomogeneous NLS in dimensions $d\ge 3$

We study the defocusing energy-critical inhomogeneous nonlinear Schrödinger equation \[ i\partial_tu+Δu=|x|^{-b}|u|^{\frac{4-2b}{d-2}}u, \qquad (t,x)\in\R\times\R^d, \] with initial data $u_0\in\dot H_x^1(\R^d)$, where $d\ge 3$ and $0<b<\min\{2,\frac d2\}$. We prove global well-posedness and scattering for arbitrary non-radial data. The main difficulties are that, when $d\ge 6$, the derivative of the critical nonlinearity is only Hölder continuous, so the short-time perturbation argument cannot be closed in $\dot S^1$, and that the singular coefficient $|x|^{-b}$ breaks translation symmetry. To handle these issues, we exploit the weak-space structure $|x|^{-b}\in L^{\frac{d}{b},\infty}(\R^d)$, introduce exotic Strichartz norms, and prove a long-time stability theorem for the general energy-critical inhomogeneous nonlinear Schrödinger equation. We also show that profiles escaping to spatial infinity are asymptotically linear because of the decay of $|x|^{-b}$. Consequently, almost periodic solutions are compact modulo scaling only, with neither spatial nor frequency center parameters. Combined with the concentration--compactness argument of Kenig--Merle [\emph{Invent. Math.} \textbf{166} (2006), 645--675], this yields the main theorem.

math.AP↗

arXiv2026

MCAT: Scaling Many-to-Many Speech-to-Text Translation with MLLMs to 70 Languages

Multimodal Large Language Models (MLLMs) have achieved great success in Speech-to-Text Translation (S2TT) tasks. However, current research is constrained by two key challenges: language coverage and efficiency. Most of the popular S2TT datasets are substantially English-centric, which restricts the scaling-up of MLLMs' many-to-many translation capabilities. Moreover, the inference speed of MLLMs degrades dramatically when the speech is converted into long sequences (e.g., 750 tokens). To address these limitations, we propose a Multilingual Cost-effective Accelerated Speech-to-Text Translator (MCAT) framework, which includes two innovations. First, a language scaling method that leverages curriculum learning and a data balancing strategy is introduced to extend the language coverage supported by MLLMs to 70 languages and achieve mutual translation among these languages. Second, an optimized speech adapter module is designed to reduce the length of the speech sequence to only 30 tokens. Extensive experiments were conducted on MLLMs of different scales (9B and 27B). The experimental results demonstrate that MCAT not only surpasses state-of-the-art end-to-end models on the FLEURS dataset across 70x69 directions but also enhances inference efficiency. The code and models are released at https://github.com/yxduir/m2m-70.

cs.CL↗

arXiv2026

MCGA: A Multi-task Classical Chinese Literary Genre Audio Corpus

With the rapid advancement of Multimodal Large Language Models (MLLMs), their potential has gained significant attention in Chinese Classical Studies (CCS). While existing research primarily focuses on text and visual modalities, the audio corpus within this domain remains largely underexplored. To bridge this gap, we introduce the Multi-task Classical Chinese Literary Genre Audio Corpus (MCGA), a 119-hour corpus comprising 22,000 audio samples. It encompasses a diverse range of literary genres across six tasks: Automatic Speech Recognition (ASR), Speech-to-Text Translation (S2TT), Speech Emotion Captioning (SEC), Spoken Question Answering (SQA), Speech Understanding (SU), and Speech Reasoning (SR). Through the evaluation of ten MLLMs, our experimental results demonstrate that current MLLMs still face substantial challenges on the MCGA test set. Furthermore, we introduce a domain-specific metric for SEC and a metric to measure the consistency between speech and text capabilities. We release MCGA to the public to facilitate the development of more robust MLLMs. MCGA Corpus: https://github.com/yxduir/MCGA

cs.CL↗

arXiv2026

Real-Time Coordinated Operation of Off-Grid Wind Powered Multi-Electrolyzer Systems Considering Thermal Dynamics and HTO Safety

Coordinated operation of alkaline water electrolysis (AWE) systems with multiple electrolyzers under fluctuating renewable power input is challenging due to varying power availability and dynamic safety constraints. Moreover, the conventional separation between optimization and control may result in inconsistent decisions across timescales. To address these issues, this paper proposes a two-layer coordinated operation method integrating feedback optimization (FO) with a projection-based safety layer. The FO layer generates real-time reference inputs to improve renewable energy utilization, while the safety layer corrects these inputs to ensure compliance with operational and safety constraints. To explicitly address the safety constraints arising from the inertial dynamics of AWE systems, discrete-time control barrier function theory is incorporated into the safety layer, thereby enhancing safety assurance and online computational tractability. Theoretical analysis establishes the feasibility and effectiveness of the proposed method. Case studies based on annual wind generation data show that the proposed method achieves high energy utilization, maintains safe operation, and demonstrates online applicability, scalability, and robustness.

eess.SY↗

arXiv2026

Synthesizing Safety in Infinite-Horizon Optimal Control for Disturbed High-Relative-Degree Systems via Barrier-Regulating Auxiliary Variables

Optimal stabilization of safety-critical nonlinear systems requires balancing long-term performance and strict safety constraints. Existing quadratic-programming-based control barrier function (CBF) safety filters are point-wise and may exhibit myopic behavior and local trapping when the safeguarding action conflicts with the nominal optimal control. This paper develops a safety-aware infinite-horizon optimal control framework by embedding a barrier-Lyapunov function (BLF)-based safeguarding action into the system dynamics and introducing a barrier-regulating auxiliary variable, thereby reformulating the original constrained problem as an unconstrained one on an extended state space. To mitigate local trapping, we introduce an adaptive alignment-conditioned tangential excitation orthogonal to the safety direction, with activation adaptively modulated by the degree of directional alignment between the nominal and safeguarding controllers, and incorporate it as an admissible $\mathcal{L}2$ disturbance in an $H\infty$ formulation. For high-relative-degree systems under disturbances, we further augment the recursive high-order safe-set construction with barrier compensation terms to obtain a high-order BLF and formulate an adversarial disturbance attenuation problem, which is approximately solved via safe-exploration-enhanced online critic learning. Simulations demonstrate reduced local trapping, improved safety--performance trade-offs, and safe operation under disturbances.

eess.SY↗

arXiv2026

PhysInOne: Visual Physics Learning and Reasoning in One Suite

We present PhysInOne, a large-scale synthetic dataset addressing the critical scarcity of physically-grounded training data for AI systems. Unlike existing datasets limited to merely hundreds or thousands of examples, PhysInOne provides 2 million videos across 153,810 dynamic 3D scenes, covering 71 basic physical phenomena in mechanics, optics, fluid dynamics, and magnetism. Distinct from previous works, our scenes feature multiobject interactions against complex backgrounds, with comprehensive ground-truth annotations including 3D geometry, semantics, dynamic motion, physical properties, and text descriptions. We demonstrate PhysInOne's efficacy across four emerging applications: physics-aware video generation, long-/short-term future frame prediction, physical property estimation, and motion transfer. Experiments show that fine-tuning foundation models on PhysInOne significantly enhances physical plausibility, while also exposing critical gaps in modeling complex physical dynamics and estimating intrinsic properties. As the largest dataset of its kind, orders of magnitude beyond prior works, PhysInOne establishes a new benchmark for advancing physics-grounded world models in generation, simulation, and embodied AI.

cs.CV↗

arXiv2026

RHVI-FDD: A Hierarchical Decoupling Framework for Low-Light Image Enhancement

Low-light images often suffer from severe noise, detail loss, and color distortion, which hinder downstream multimedia analysis and retrieval tasks. The degradation in low-light images is complex: luminance and chrominance are coupled, while within the chrominance, noise and details are deeply entangled, preventing existing methods from simultaneously correcting color distortion, suppressing noise, and preserving fine details. To tackle the above challenges, we propose a novel hierarchical decoupling framework (RHVI-FDD). At the macro level, we introduce the RHVI transform, which mitigates the estimation bias caused by input noise and enables robust luminance-chrominance decoupling. At the micro level, we design a Frequency-Domain Decoupling (FDD) module with three branches for further feature separation. Using the Discrete Cosine Transform, we decompose chrominance features into low, mid, and high-frequency bands that predominantly represent global tone, local details, and noise components, which are then processed by tailored expert networks in a divide-and-conquer manner and fused via an adaptive gating module for content-aware fusion. Extensive experiments on multiple low-light datasets demonstrate that our method consistently outperforms existing state-of-the-art approaches in both objective metrics and subjective visual quality.

cs.CV↗

arXiv2026

Emergence and transition of incompressible phases in decorated Landau levels

A single Landau level (LL) dressed with periodic electrostatic potentials can realize a plethora of interacting topological phases where the Hall conductivity generally does not equal to the LL filling factor. Their physics can be captured by a new family of flat topological bands: decorated Landau levels (dLL) from imposing an electrostatic delta potential lattice within a single LL. With $p/q$ magnetic fluxes per unit cell, there are $q$ dispersive bands and $p-q$ zero energy bands forming the dLL. When the electrostatic potential strength dominates the electron-electron interaction, band mixing is suppressed and the dispersion bands consist of ``localized states" with vanishing total Chern number. Nevertheless these dispersive bands can have highly nontrivial Berry curvature distribution, and even non-zero Chern numbers when $q>1$. Interestingly even in the limit of large short range interaction, band mixing between dLL and dispersion bands can be strongly suppressed at low filling factor, leading to robust topological phases within the dLL stabilized by the one-body potential. The dLL and the associated dispersive bands can serve as minimal theoretical models for correlated physics in lattice or moiré systems; they are also highly tunable experimental platforms for realizing rich phase diagrams of exotic 2D quantum fluids.

cond-mat.str-el↗