Agents based on Large Language Models (LLMs) have demonstrated strong capabilities across a wide range of tasks. However, deploying LLM-based agents in high-stakes domains comes with significant safety and ethical risks. Unethical behavior by these agents can directly result in serious real-world consequences, including physical harm and financial loss. To efficiently steer the ethical behavior of agents, we frame agent behavior steering as a model editing task, which we term Behavior Editing. Model editing is an emerging area of research that enables precise and efficient modifications to LLMs while preserving their overall capabilities. To systematically study and evaluate this approach, we introduce BehaviorBench, a multi-tier benchmark grounded in psychological moral theories. This benchmark supports both the evaluation and editing of agent behaviors across a variety of scenarios, with each tier introducing more complex and ambiguous scenarios. We first demonstrate that Behavior Editing can dynamically steer agents toward the target behavior within specific scenarios. Moreover, Behavior Editing enables not only scenario-specific local adjustments but also more extensive shifts in an agent’s global moral alignment. We demonstrate that Behavior Editing can be used to promote ethical and benevolent behavior or, conversely, to induce harmful or malicious behavior. Through extensive evaluations of agents built on frontier LLMs, BehaviorBench validates the effectiveness of behavior editing across a wide range of models and scenarios. Our findings offer key insights into a new paradigm for steering agent behavior, highlighting both the promise and perils of Behavior Editing.
论文检索
输入标题、作者或关键词，从 3,314 篇学术成果中精准定位
RESTL: Reinforcement Learning Guided by Multi-Aspect Rewards for Signal Temporal Logic Transformation
PDF ↗Signal Temporal Logic (STL) is a powerful formal language for specifying real-time specifications of Cyber-Physical Systems (CPS). Transforming specifications written in natural language into STL formulas automatically has attracted increasing attention. Existing rule-based methods depend heavily on rigid pattern matching and domain-specific knowledge, limiting their generalizability and scalability. Recently, Supervised Fine-Tuning (SFT) of large language models (LLMs) has been successfully applied to transform natural language into STL. However, the lack of fine-grained supervision on atomic proposition correctness, semantic fidelity, and formula readability often leads SFT-based methods to produce formulas misaligned with the intended meaning. To address these issues, we propose RESTL, a reinforcement learning (RL)-based framework for the transformation from natural language to STL. RESTL introduces multiple independently trained reward models that provide fine-grained, multi-faceted feedback from four perspectives, i.e., atomic proposition consistency, semantic alignment, formula succinctness, and symbol matching. These reward models are trained with a curriculum learning strategy to improve their feedback accuracy, and their outputs are aggregated into a unified signal that guides the optimization of the STL generator via Proximal Policy Optimization (PPO). Experimental results demonstrate that RESTL significantly outperforms state-of-the-art methods in both automatic metrics and human evaluations.
Embodied agents, such as robots, will need to interact in situated environments where successful communication often depends on reasoning over social norms: shared expectations that constrain what actions are appropriate in context. A key capability in such settings is norm-based reference resolution (NBRR), where interpreting referential expressions requires inferring implicit normative expectations grounded in physical and social context. Yet it remains unclear whether Large Language Models (LLMs) can support this kind of reasoning. In this work, we introduce SNIC (Situated Norms in Context), a human-validated diagnostic testbed designed to probe how well state-of-the-art LLMs can extract and utilize normative principles relevant to NBRR. SNIC emphasizes physically grounded norms that arise in everyday tasks such as cleaning, tidying, and serving. Across a range of controlled evaluations, we find that even the strongest LLMs struggle to consistently identify and apply social norms—particularly when norms are implicit, underspecified, or in conflict. These findings reveal a blind spot in current LLMs and highlight a key challenge for deploying language-based systems in socially situated, embodied settings.
Text-to-SQL is a critical task in natural language processing that aims to transform natural language questions into accurate and executable SQL queries. In real-world scenarios, these reasoning tasks are often accompanied by complex mathematical computations, domain knowledge, and hypothetical reasoning scenarios. However, existing large-scale Text-to-SQL datasets typically focus on business logic and task logic, neglecting critical factors such as vertical domain knowledge, complex mathematical reasoning, and hypothetical reasoning, which are essential for realistically reflecting the reasoning demands in practical applications and completing data querying and analysis. To bridge this gap, we introduce LogicCat, the first Text-to-SQL benchmark dataset specifically designed for complex reasoning and chain-of-thought parsing, encompassing physics, arithmetic, commonsense, and hypothetical reasoning scenarios. LogicCat comprises 4,038 English questions paired 12,114 detailed chain-of-thought reasoning steps, spanning 45 databases across diverse domains, significantly surpassing existing datasets in complexity. Experimental results demonstrate that LogicCat substantially increases the task difficulty for current state-of-the-art models to at most 33.20% execution accuracy, indicating that this task remains exceptionally challenging. The advancement of LogicCat represents a crucial step toward developing systems suitable for real-world enterprise data analysis and autonomous query generation.
AcoustoReinforce: Multi-Particle Acoustophoretic Path Planning with Deep Reinforcement Learning
PDF ↗Acoustophoresis uses sound waves to manipulate small objects in mid-air and has broad potential in various applications. However, stable multi-particle levitation remains challenging due to complex acoustic dynamics and limitations of existing models. We introduce AcoustoReinforce, a reinforcement learning-based path planner that autonomously controls the motion of multiple levitated particles. Leveraging a decentralized architecture, it learns local neural policies that generate particle trajectories independently, enabling scalable, communication-free control even in densely populated acoustic fields. To ensure physical feasibility, acoustic trapping strength is incorporated as a constraint during both training and inference, producing trajectories that are collision-free, acoustically stable, and physically realizable within real-world system constraints. Experiments on a real-world levitation platform show that AcoustoReinforce outperforms state-of-the-art planners, improving task success rates by up to 130% across diverse configurations. These results demonstrate the effectiveness of learning-based decentralized control for complex multi-object acoustophoresis in real environments.
Physics-Informed Koopman Neural Estimation of the Heston Model from High-Frequency Observations
PDF ↗We propose a physics-informed learning framework, called Koopman-PINN, to estimate the parameters of the Heston stochastic volatility model with high-frequency price data in financial markets. The method integrates a nonparametric volatility estimation (known as ART-filter in the literature), moment-based parameter initialization, and a neural Koopman operator constrained by the infinitesimal generator of the underlying stochastic differential equation. By incorporating a generator-based loss, the model bridges Koopman theory and neural modeling to handle partially observed coupled stochastic dynamics in a manner consistent with continuous-time evolution. Across diverse parameter combinations reflecting varying market conditions, Koopman-PINN consistently achieves accurate and robust five-parameter recovery, outperforming existing estimators under a minimal set of initialization assumptions.
Molecular assembly (MA) has long been a fundamental task in chemistry and biology, with the potential to create new materials and enable novel functions beyond the molecular scale. However, its vast conformational search space poses substantial challenges, and current generative models remain limited in capturing molecular flexibility and preventing non-physical poses. In this paper, we propose AssemUDB, a diffusion bridge–based framework that learns transport mappings between two distinct flexible domains for molecular assembly generation. We reformulate the marginal matching constraint of diffusion bridges as a coupling distribution governed by unbalanced transport rather than imposing strict conservation. Subsequently, we employ a progressive process from structural relaxation in Euclidean space to assembly on the SE(3) manifold. This relaxation of marginal conservation grants the generative model greater flexibility and leads to more physically plausible atom placements. Comprehensive experiments demonstrate the superior performance of AssemUDB. Notably, we find that the method demonstrates performance comparable to, or even better than, mature tools such as PackMol for packing tasks.
Many Minds, One Path: LLM-Augmented Consensus Decision for Distributed Control in Multi-Agent Collaborative Stable Scenarios
PDF ↗Distributed multi-agent systems are increasingly deployed in dynamic and high-stakes environments such as power grids, intelligent traffic systems, and collaborative robotics. In these systems, long-term stability, the ability to maintain coherent and safe system behavior over time, is critical but underexplored in existing research. This paper presents LLMASC, a framework designed to enhance long-term stability in multi-agent collaboration by combining semantic reasoning with decentralized control. LLMASC comprises three key components: a Semantic Perception Encoder that transforms heterogeneous agent observations into structured natural language; an LLM-Guided Consensus Decision module that enables strategic alignment through proposal exchange and voting; and a Policy Execution Controller that maps high-level plans to executable actions via reinforcement learning. We evaluate LLMASC across three representative simulation domains (Multi-Walker, Simulation of Urban Mobility and Power Grid Stabilization), spanning both physical and cyber-physical systems. Experiments show that LLMASC consistently outperforms the best baselines, improving stability rates by up to 44% and long-term success by 31%. Further analysis confirms its decision-making efficiency and robustness under varying agent populations and model choices.
Equivariant Atomic and Lattice Modeling Using Geometric Deep Learning for Crystal Structure Optimization
PDF ↗Structure optimization, which yields the relaxed structure (minimum‑energy state), is essential for reliable materials property calculations, yet traditional ab initio approaches such as density‑functional theory (DFT) are computationally intensive. Machine learning (ML) has emerged to alleviate this bottleneck but suffers from two major limitations: (i) existing models operate mainly on atoms, leaving lattice vectors implicit despite their critical role in structural optimization; and (ii) they often rely on multi-stage, non-end-to-end workflows that are prone to error accumulation. Here, we present E³Relax, an end-to-end equivariant graph neural network that maps an unrelaxed crystal directly to its relaxed structure. E³Relax promotes both atoms and lattice vectors to graph nodes endowed with dual scalar–vector features, enabling unified and symmetry‑preserving modeling of atomic displacements and lattice deformations. A layer‑wise supervision strategy forces every network depth to make a physically meaningful refinement, mimicking the incremental convergence of DFT while preserving a fully end‑to‑end pipeline. We evaluate E³Relax on four benchmark datasets and demonstrate that it achieves remarkable accuracy and efficiency. Through DFT validations, we show that the structures predicted by E³Relax are energetically favorable, making them suitable as high-quality initial configurations to accelerate DFT calculations.
Accurate and efficient simulations of physical phenomena governed by partial differential equations (PDEs) are important for scientific and engineering progress. While traditional numerical solvers are powerful, they are often computationally expensive. Recently, data-driven methods have emerged as alternatives, but they frequently suffer from error accumulation and limited physical consistency, especially in multiphysics and complex geometries. To address these challenges, we propose PEGNet, a Physics-Embedded Graph Network that incorporates PDE-guided message passing to redesign the graph neural network architecture. By embedding key PDE dynamics like convection, viscosity, and diffusion into distinct message functions, the model naturally integrates physical constraints into its forward propagation, producing more stable and physically consistent solutions. Additionally, a hierarchical architecture is employed to capture multi-scale features, and physical regularization is integrated into the loss function to further enforce adherence to governing physics. We evaluated PEGNet on benchmarks, including custom datasets for respiratory airflow and drug delivery, showing significant improvements in long-term prediction accuracy and physical consistency over existing methods.
Towards Real-Time Neutral Atom Array Assembly via Unsupervised Hologram Generation and Path Optimization
PDF ↗The rapid and reliable assembly of defect-free atom arrays poses a fundamental challenge for neutral atom quantum computing. While parallel rearrangement methods using spatial light modulators show promise, they suffer from significant overhead in two sub-tasks: atom-site matching and hologram generation. We propose a framework to address these bottlenecks and enhance the efficiency and fidelity of the assembly process. It features a new optimization objective for atom-site matching that minimizes the longest movement path, and a Fourier U-Net model that integrates Fourier operators with image-to-image translation to enable real-time hologram generation. The model is trained in a fully self-supervised paradigm, leveraging the physical properties of holography to remove the need for costly ground-truth labels. Experimental results show our framework not only significantly outperforms the state-of-the-art supervised CNN-based model but also achieves an inference speed orders of magnitude faster than traditional iterative algorithms, enabling real-time, dynamic atom rearrangement.
The discovery rate of optical transients will explode to 10 million public alerts per night once the Vera C. Rubin Observatory’s Legacy Survey of Space and Time comes online, overwhelming the traditional physics-based inference pipelines. A continuous-time forecasting AI model is of interest because it can deliver millisecond-scale inference for thousands of objects per day, whereas legacy MCMC codes need hours per object. In this paper, we propose SELDON, a new continuous-time variational autoencoder for panels of sparse and irregularly time-sampled (gappy) astrophysical light curves that are nonstationary, heteroscedastic, and inherently dependent. SELDON combines a masked GRU-ODE encoder with a latent neural ODE propagator and an interpretable Gaussian-basis decoder. The encoder learns to summarize panels of imbalanced and correlated data even when only a handful of points are observed. The neural ODE then integrates this hidden state forward in continuous time, extrapolating to future unseen epochs. This extrapolated time series is further encoded by deep sets to a latent distribution that is decoded to a weighted sum of Gaussian basis functions, the parameters of which are physically meaningful. Such parameters (e.g., rise time, decay rate, peak flux) directly drive downstream prioritization of spectroscopic follow-up for astrophysical surveys. Beyond astronomy, the architecture of SELDON offers a generic recipe for interpretable and continuous-time sequence modeling in any time domain where data are multivariate, sparse, heteroscedastic, and irregularly spaced.
Decoupled Spatiotemporal Forecasting from Extreme Sparse Observations via Quantized Latent Space
PDF ↗Predicting spatiotemporal fields governed by partial differential equations (PDEs) from sparse sensor data is a critical and long-standing challenge in science and engineering. Recent deep learning approaches, particularly neural operators, have shown considerable promise in solving PDEs. However, their performance degrades significantly in the demanding regime of extreme sparsity, characterized by spatial sensor coverage of less than 1% and limited temporal observations. To overcome this limitation, we propose a novel framework that decouples the task into two stages: spatial reconstruction and temporal extrapolation. In the first stage, rather than reconstructing the high-dimensional physical field directly, our model learns to reconstruct the complete latent features from sparse observations—features that would otherwise be extracted from a dense field. This process is stabilized by a Vector Quantization (VQ) bottleneck, which discretizes the latent space. In the second stage, a decoder-only Transformer performs temporal extrapolation by autoregressively predicting the future sequence of these discrete latent indices. This design inherently allows the model to generalize to new initial conditions and varying forecast horizons, akin to standard autoregressive models. We validate our framework on three challenging benchmarks, achieving state-of-the-art (SOTA) performance under severe sparsity constraints. Furthermore, we introduce a challenging benchmark dataset based on fire dynamics simulations. On this benchmark, our model successfully forecasts the field's evolution 30 frames into the future from a single timeframe with less than 0.1% spatial observations—a result that pushes well beyond the capabilities of existing methods.
Tabular data is the most abundant data type in the world, powering systems in finance, healthcare, e‑commerce, and beyond. As tabular datasets grow and span multiple related targets, there is an increasing need to exploit shared task information for improved multitask generalization. Multitask learning (MTL) has emerged as a powerful way to improve generalization and efficiency, yet most existing work focuses narrowly on large‑scale recommendation systems, leaving its potential in broader tabular domains largely underexplored. Also, existing MTL approaches for tabular data predominantly rely on multi-layer perceptron-based backbones, which struggle to capture complex feature interactions and often fail to scale when data is abundant, a limitation that transformer architectures have overcome in other domains. Motivated by this, we introduce MultiTab-Net, the first multitask transformer architecture specifically designed for large tabular data. MultiTab-Net employs a novel multitask masked‑attention mechanism that dynamically models feature–feature dependencies while mitigating task competition. Through extensive experiments, we show that MultiTab-Net consistently achieves higher multitask gain than existing MTL architectures and single‑task transformers across diverse domains including large‑scale recommendation data, census‑like socioeconomic data, and physics datasets, spanning a wide range of task counts, task types, and feature modalities. In addition, we contribute MultiTab-Bench, a generalized multitask synthetic dataset generator that enables systematic evaluation of multitask dynamics by tuning task count, task correlations, and relative task complexity.
Accurate and efficient modeling of soft-tissue interactions is fundamental for advancing surgical simulation, surgical robotics, and model-based surgical automation. To achieve real-time latency, classical Finite Element Method (FEM) solvers are often replaced with neural approximations; however, naively training such models in a fully data-driven manner without incorporating physical priors frequently leads to poor generalization and physically implausible predictions. We present a novel physics-informed neural simulation framework that enables real-time prediction of soft-tissue deformations under complex single- and multi-grasper interactions. Our approach integrates Kelvinlet-based analytical priors with large-scale FEM data, capturing both linear and nonlinear tissue responses. This hybrid design improves predictive accuracy and physical plausibility across diverse neural architectures while maintaining the low-latency performance required for interactive applications. We validate our method on challenging surgical manipulation tasks involving standard laparoscopic grasping tools, demonstrating substantial improvements in deformation fidelity and temporal stability over existing baselines. These results establish Kelvinlet-augmented learning as a principled and computationally efficient paradigm for real-time, physics-aware soft-tissue simulation in surgical AI.
Emerging from recent advances in foundation models, Large Wireless Models (LWMs) represent a new paradigm of general-purpose intelligence for wireless communications that transcends task-specific engineering. The success of foundation models is critically underpinned by scaling laws, which provide a predictable roadmap for how performance scales with resources. However, established scaling laws from language and vision, charting performance as a power-law of model and dataset sizes, are ill-suited for the wireless domain, as their core formulations cannot model the structured nature of the physical channel. To address this, we propose a novel wireless scaling law that extends the classical formulation by modeling two wireless-native factors: channel heterogeneity and discretization granularity. These two factors reshape scaling behavior via nested linear and power-law relationships, recasting the scaling law's parameters (notably the scaling exponent and irreducible loss) from universal constants into dynamic variables dictated by the physical environment. Our physics-aware formulation reveals two key insights: first, that compute-optimal scaling is not dictated by a fixed model-data ratio but is instead a dynamic function of heterogeneity and granularity, and second, that this dependency is particularly sensitive to granularity, allowing significant performance to be unlocked from existing data simply by refining its resolution. Crucially, this establishes a reliable roadmap for designing powerful yet resource-efficient LWMs, translating theoretical insights into actionable engineering principles. Extensive experiments validate our wireless scaling law, showing a 32.31% prediction accuracy improvement over classical laws in diverse wireless scenarios where they fail.
We propose BRIC, a novel test-time adaptation (TTA) framework that enables long-term human motion generation by resolving execution discrepancies between diffusion-based kinematic motion planners and reinforcement learning-based physics controllers. While diffusion models can generate diverse and expressive motions conditioned on text and scene context, they often produce physically implausible outputs, leading to execution drift during simulation. To address this, BRIC dynamically adapts the physics controller to noisy motion plans at test time, while preserving pre-trained skills via a loss function that mitigates catastrophic forgetting. In addition, BRIC introduces a lightweight test-time guidance mechanism that steers the diffusion model in the signal space without updating its parameters. By combining both adaptation strategies, BRIC ensures consistent and physically plausible long-term executions across diverse environments in an effective and efficient manner. We validate the effectiveness of BRIC on a variety of long-term tasks, including motion composition, obstacle avoidance, and human-scene interaction, achieving state-of-the-art performance across all tasks.
Dynamic Semantic Tokenization for Time Series via Elastic Sampling on Physics-aware Perception
PDF ↗Despite the remarkable success of semantic token learning in NLP and vision domains, token-level representation mechanisms face fundamental challenges when extended to continuous time series analysis. We identify a core limitation lies in the intrinsic absence of semantically meaningful tokenization boundaries within time-series, which differs substantially from discrete text tokens and presents unique complexities compared to spatially coherent image patches. While existing works mechanically apply fixed-length partitioning, recent evidence from time series foundation models reveals performance ceilings in prediction tasks under such paradigms. This paper introduces a novel tokenization framework known as physics-aware tokenization (PATK), designed to implement adaptive time-frequency tokenization via distribution-sensitive sampling strategies. Key innovations include: 1) A Rate-of-Variation (RoV) distribution is meticulously structured to encompass multi-scale temporal dynamics in the time domain, alongside a Spectral Energy Intensity (SEI) distribution devised to reveal global seasonal patterns within the frequency domain; 2) A physics-aware hidden Markov modeling (PA-HMM) is then established to adaptively breaks down continuous time-series into distinct tokens with elastic lengths, responding to physics-aware probabilities sampled from RoV and SEI distributions. The proposed PATK allows steady integration with both conventional Transformers and advanced large-scale time series models (including LLM-transferred methods and pretrained time series foundation models). Simulations across various datasets demonstrate that PATK excels in classification and forecasting tasks, showing notable adaptability to model long-term dependencies, strengthening resilience against disturbances, and robustness to missing data events.
The rapid expansion of the Internet of Things (IoT) has created a growing demand for large-scale sensor deployment. However, the high cost of physical sensors limits the scalability and coverage of sensor networks, making fine-grained sensing difficult. Inductive Spatio-Temporal Kriging (ISK) addresses this challenge by introducing virtual sensors that infer measurements from physical sensors, typically using graph neural networks (GNNs) to model their relationships. Despite its promise, current ISK methods often rely on standard message-passing and generic architectures that fail to effectively capture spatio-temporal features or represent virtual nodes accurately. Additionally, existing graph construction techniques suffer from sparse and noisy connections, further hindering performance. To address these limitations, we propose DarkFarseer, a novel ISK framework with three key innovations. First, the Style-enhanced Temporal-Spatial architecture adopts a temporal-then-spatial processing scheme with a temporal style transfer mechanism to enhance virtual node representations. Second, Regional-semantic Contrastive Learning improves representation learning by aligning virtual nodes with regional component patterns. Third, the Similarity-Based Graph Denoising Strategy mitigates the influence of noisy edges by leveraging temporal similarity and regional structure. Extensive experiments on real-world datasets demonstrate that DarkFarseer significantly outperforms state-of-the-art ISK methods.
Uncovering Hidden Degeneration: A Physics-Guided Bidirectional Inference Framework for Industrial Time Series Prediction
PDF ↗Hidden degenerations in industrial time series often precede observable failures, they remain undetected by standard monitoring systems until anomalies become apparent. This gap between microscopic degradation and macroscopic observation renders conventional predictors inherently reactive, as they rely on correlations in sensor data rather than uncovering the underlying, physics‑consistent degradation states. Crucially, the microscopic mechanisms governing system evolution depend on macroscopic state variables—whose measurements are expectations over microscopic probability distributions—so purely data‑driven “top‑down” or purely physics‑guided “bottom‑up” approaches cannot forecast degeneration‑entangled industrial faults. To address these challenges, we propose a Physics-Guided Bidirectional Inference Framework that represents hidden microscopic states from macroscopic measurements. Our approach uniquely combines: (1) bottom-up physics-based simulation using Continuum Damage Mechanics to model micro-scale damage evolution under environmental stressors, and (2) top-down probabilistic inference via maximum entropy formalism to estimate latent microstate distributions from sparse sensor data. This bidirectional mechanism enables early failure prediction by bridging observable measurements with unobservable degeneration. Validation on real-world railway infrastruc datasets demonstrates significant improvements in early fault prediction compared to state-of-the-art baselines. Our method establishes a new paradigm for safety-critical industrial applications requiring reliable prediction of hidden degeneration processes.