Papers

Creating animatable and relightable human avatars from multi-view images remains challenging, as pose-dependent deformation, materials, and light visibility are tightly coupled in images. In this paper, we present ARS-Avatar, a novel method using surfel representation for high-quality, animatable, and relightable human avatars from multi-view images captured under unknown illumination. For faithful surfel estimation, we extract deformation displacement and body model deviation two priors from the mesh, which jointly capture local deformation complexity and serve as auxiliary. We employ deferred shading to estimate BRDF materials and further introduce a differentiable screen-space ambient occlusion formulation that enables gradient-based optimization of body-part specific occlusion radii through finite differences, providing an efficient approximation of light visibility that can be jointly optimized with the avatar. Extensive experiments show that ARS-Avatar achieves competitive or improved radiance reconstruction on multiple metrics and consistently outperforms the evaluated PBR relighting baselines, while enabling realistic animation and relighting under novel poses and illuminations.

Generating controllable and human-like emphasis remains an open challenge in text-to-speech, even when explicit emphasis control signals are provided in the text input, limiting the communicative accuracy of synthetic speech in real-world applications. Reinforcement learning has recently shown promise for post-training TTS systems to align with human preference, yet existing methods have not been applied to word-level prosodic control. We present EmphTTS, a non-autoregressive TTS system that applies Group Relative Policy Optimization (GRPO) to the duration predictor with an emphasis localization reward, enabling direct optimization for word-level emphasis. Evaluations show that EmphTTS achieves the best emphasis controllability and performs the best in emphasis objective evaluation. In subjective preference tests, EmphTTS is significantly preferred over synthetic groundtruth and most baselines. Ablation studies show that GRPO improves emphasis realization beyond supervised-finetuning-based duration modeling and simple speaking-rate adjustment, while alleviating the mismatch between the independently trained duration predictor and TTS model.

The growing demand for automation in the mining industry, particularly for the autonomous operation of articulated dump trucks (ADTs), has drawn increased attention to accurate vehicle modeling. The importance of such models lies in their use in model predictive control (MPC), model-based estimation methods, and vehicle simulation. While dynamic modeling offers a viable solution for these purposes, it is associated with complex setup and parametrization and may require recalibration in changing operating environments. As a result, kinematic models have dominated ADT modeling, especially in MPCs, at the expense of reduced prediction accuracy. In this work, we propose an approach using Gaussian Process Regression (GPR) to learn the sideslip angle of the vehicle, which is identified as the primary contributor to the reduced accuracy of kinematic models. The learned GPR function is augmented into the kinematic model to form a gray-box model that aims to reduce the gap to dynamic models. We show that the gray-box model can predict the sideslip angle and, consequently, the vehicle's lateral velocity, thereby improving the MPC's prediction performance. The resulting gray-box MPC is compared against two white-box MPCs in a simulation environment. The results indicate an improvement in terms of maximum lateral tracking error from over 2 m to 0.56 m.

Wearable sensing enables intuitive hand gesture recognition for human–computer interaction, augmented reality, and prosthetic control, yet subject–independent recognition remains challenging because wearable signals provide only indirect and highly subject-specific observations of hand motion. Although visual information can improve wearable gesture recognition, requiring it during inference increases sensing complexity and limits practical deployment. We propose ViMoWear, a visual-motion-guided framework that leverages synchronized 3D hand motion as training-only supervision while requiring only wearable sensing for gesture classification at inference. Specifically, Motion-Guided Cross-Subject Contrastive Learning (MGCL) promotes subject-robust representations, and Thumb-Aware Masked Motion Reconstruction (TMMR) preserves fine-grained motion information. The leave-one-subject-out experiments on a synchronized sEMG–IMU–pose dataset demonstrate consistent improvements over supervised baselines across multiple sensing configurations, while the learned representations also support classifier-free retrieval. The proposed training-only visual motion supervision improves the generalization of wearable representations to unseen subjects.

We study the problem of computational efficiency for linear bandits in high-dimensional settings with a finite arm set. In linear bandits, the increase in the dimension $d$ of the feature vectors leads to growing computational costs of $O(d^2)$ at each round of update. Traditional sketching-based methods such as SOFUL reduce computation via fixed-size matrix sketching, yet run the risk of incurring vacuous linear regret when the spectral tail of the data is heavy and the sketch size is inadequately selected. To guarantee regret convergence and effectively reduce computational costs, we introduce a clustering mechanism and propose the Cluster Sketch Linear Bandit (CS-LB) algorithm. Our method preserves the full covariance information in each cluster to guarantee robust sublinear regret without spectral-tail vulnerabilities, performs cluster switching by assigning a sentinel for each cluster, and reduces per-round update computation to $O(l^2d)$ via a tunable sketch size $l<d$. Experiments on synthetic datasets demonstrate that our method consistently maintains a favorable trade-off between efficiency and regret.

Concept erasure methods that operate via linear projection assume that features occupy separable subspaces. We show this assumption fails under dense superposition: when two features are forced into an antipodal pair sharing a single subspace, state-of-the-art linear erasure destroys both, not just the target. Networks trained with gradient descent instead solve this problem non-linearly, but not uniformly: they converge to one of two distinct circuit-level solutions depending on initialization, which we call mirror and shadow solutions. We map this bifurcation as a function of feature entanglement, show it reflects a stable attractor structure rather than an artifact of our setup, and use targeted causal interventions to demonstrate that both solutions leave a substantial, measurable trace of the erased feature's representation intact, recoverable through a single scalar patch rather than requiring any further training. This mirrors a failure mode recently observed empirically in LLM unlearning, where suppression rather than deletion allows forgotten knowledge to resurface; our results offer a mechanistic, causally-validated account of why that failure mode occurs.

Long-context evaluations often test whether a model can recover distant evidence, but recoverability does not guarantee behavioral influence. We test the prediction that a distant discourse anchor can remain explicitly recoverable yet fail to change the locally preferred reading of a familiar multiword expression; such failures should concentrate when the model's no-anchor default conflicts with the anchor, while prior-correct decisions remain largely preserved. We introduce Multiword Expression Effective Context Length (MWE-ECL), a bilingual diagnostic whose matched anchor-retrieval, no-anchor prior, and interpretation prompts measure explicit recoverability, model-observed defaults, and anchor-conditioned decisions, respectively. Across eight English deployment panels on a shared 0-128K grid, retrieval-control accuracy on prior-conflict items is 0.989-1.000, prior-conflict override spans 0.806-1.000 (0.809-1.000 after conditioning on correct retrieval), and preservation of prior-correct decisions remains 0.977-1.000. A same-call control querying retrieval and interpretation in one prompt reproduces the gap for DeepSeek V4 Pro (1.000 retrieval versus 0.900-0.920 interpretation), showing that separate invocations are not its sole explanation; smaller or absent gaps in the other two models bound its generality. For DeepSeek V4 Flash, separate prompt-fit tests retain perfect retrieval with lower interpretation at 512K and 1M, while foil-consistent cues shift the no-anchor prior far more than retrieval; cross-model cue effects are heterogeneous. A separately reported 10-family Chinese subset shows similar descriptive gaps, but imperfect retrieval for some models prevents an integration-only attribution. MWE-ECL therefore evaluates whether explicitly recoverable distant context changes a competing local semantic decision.

Existing machine learning (ML) scaling laws relate predictive loss to compute, model parameters, and data. However, as models are increasingly deployed through agentic harnesses, loss alone is insufficient to characterize downstream performance: models with similar loss can exhibit different capabilities in reasoning, retrieval, planning, and adaptation. Yet, no unified framework connects such capabilities to the coupled resources available across the ML lifecycle. We bridge this gap by introducing a capability manifold, a multidimensional framework mapping downstream capabilities to pre-training, post-training, and test-time resources through bounded scaling functions. Analytical Jacobians quantify capability sensitivity to resource changes and interactions. As an initial application, we embed Kaplan- and Chinchilla-type scaling laws and test-time compute within the framework, demonstrating how existing scaling relationships can be unified as trajectories on a common capability manifold.

LLM coding agents increasingly operate inside game-engine repositories, where application logic is inseparable from serialized assets: a single gameplay change may span C\# scripts, prefabs, scenes, and ScriptableObjects wired together by Unity GUIDs. The retrieval tools agents carry today—shell utilities and code-only indexes—cannot answer basic cross-file questions, because these relationships live in .meta files and YAML assets rather than in code. We present Unity Insight, to our knowledge the first persistent, LLM-facing, agent-integrated cross-file code–asset index for Unity projects, shipping in production with Tuanjie Codely, the agent CLI of Tuanjie Engine, since its public launch on 2026-07-28. In a paired experiment—28 project-specific questions on two Unity games, same model and harness, one run per arm per question—the index-backed agent spent 53% fewer tokens and 52% less wall-clock time than a general-purpose exploration agent (exact paired sign tests, $p{<}0.004$), using only its typed index-query tools.

A de Bruijn sequence packs all $n$-bit binary words into a cycle of length $2^n$. A de Bruijn torus is the two-dimensional analogue in which each word appears exactly once in a rectangular window. Here we consider the natural analogue for permutations using their shorthand representation (i.e., each permutation's final redundant value is omitted from the window). We show that these tori exist when $n = 2m + 1$ is odd and the torus and windows have two rows (i.e., the torus is a "tworus"). These twori can be constructed with a high degree of symmetry. More specifically, there are twori that can be partitioned into $2^{m-1}$ matching blocks where each block contains the same sequence of unordered columns. Furthermore, given one such block we can generate each successive column of a tworus in amortized $\mathcal{O}(1)$-time. We also prove non-existence results for certain sizes of tori and provide algorithms for constructing multiversal cycles (perfect necklaces) of unlabeled binary words.

Step Law gives power-law formulas for the optimal peak learning rate eta* and batch size B* when pre-training language models. It was calibrated on models between 59M and 1B parameters; the small-model regime N < 59M was never tested empirically by its authors. This regime matters for single-GPU training, interpretability research, educational experiments, and settings where larger models are infeasible on memory or cost grounds. We test whether Step Law transfers to small language models. We consider three outcomes: H1, the original coefficients work directly; H2, the power-law form holds but with different coefficients; and H3, a power law does not describe the optima in this regime. All experiments use a single nanoGPT/TinyStories pipeline with a 2048-token BPE vocabulary, AdamW, and a warmup-cosine schedule. The optimum for each (N, D) cell is extracted from the loss surface L(eta, B) via a local quadratic approximation in log-log coordinates over the smoothed training loss. The final dataset contains 29 unique (N, D) cells and 935 analysis-ready runs. The main refit uses 25 cells (815 runs) in the working range 4 <= D/N <= 600. On the pooled data we accept H2: the functional form is preserved, but the coefficients differ from the original. We obtain eta*(N, D) = 0.0985 N^(-0.508) D^(0.238) (R^2 = 0.834) and B*(D) = 3.6 x 10^(-4) D^(0.931) (R^2 = 0.950). Step Law's structural claim that B* is independent of N is reproduced (p = 0.87), but the growth of B* with D is nearly twice as steep as in the original work. Direct transfer of Step Law systematically overestimates the optimal learning rate: the median ratio eta_SL / eta* is approximately 4.0x, with a range of 2.4x to 6.6x.

Distributed intelligence concerns systems in which semi-autonomous components with local dynamics and partial observations coordinate through information exchange to maintain a shared function. This paper proposes an operational way to study such systems: measure information at the interface where a message changes a receiving action, then connect that measure to function by intervention and disturbance evaluation. We instantiate this proposal in DI-Walker, a two-dimensional four-limb embodied plant controlled by frozen Cross-Entropy-Method policies. We compare a controller using each limb's own realized-force sensor with one using the realized-force sensors of peer limbs. Under limb loss, limb slip, and weak central-control dropout, Peer-Sensor has lower late tracking error in several conditions. A corrected finite-history action-predictive estimator shows a substantially larger peer-message gain under compound failure. A future scalar functional-prediction estimator does not show the same stable advantage. We interpret this discrepancy as a methodological result: information useful for an intermediate control action can be hidden by later plant dynamics, redundancy, and context. The paper relates this result to Predictive Information, Transfer Entropy, Directed Information, information-to-go/IT-PAC ideas, empowerment, and the robust control data-rate perspective, while explicitly distinguishing operational predictive gains from exact Directed Information, channel capacity, and a formal data-rate theorem.

Multimodal joint training often suffers from modality imbalance, where a dominant modality suppresses the optimization of others. Existing methods mainly balance modality learning by modulating gradient magnitudes or directions, modifying optimization objectives, or adjusting training strategies, with most interventions focusing on the current update. However, when combined with widely used momentum-based optimizers, the update also incorporates accumulated information from previous gradients, which is not explicitly addressed by current-step modulation alone. To address this issue, we propose Variance-Calibrated MomentuM (VCMM), which adapts gradient memory to modality-specific gradient dynamics. Specifically, VCMM estimates minibatch noise and temporal drift online and uses their relative strength to determine modality-specific momentum through a Kalman-inspired controller. We further center the control signal across modalities and apply exact bias correction for the time-varying first moment, enabling adaptive gradient memory without extra network passes or explicit learning-rate scaling. Experiments on four multimodal benchmarks demonstrate consistent improvements with modest training overhead.

This paper deals with the generic modeling of systems of systems (SoSs) using agent-based modeling. SoSs are large-scale systems, including numerous-possibly heterogeneous-interacting component systems evolving in a dynamic environment. The aim of this paper is to provide generic formalism allowing to represent and control the whole complexity of a SoS using agent-based simulations. In particular, organizational aspects of SoSs are managed with the Agent-Group-Role model. Functional aspects, guiding SoSs to accomplish their global goals, are handled via a functional specification. Multilevel aspects are modeled with the Influence Reaction Model for Multilevel Simulation (IRM4MLS) agent-based meta-model. Models generated using this formalism encompass static and dynamic aspects of SoSs. They consider reorganization of SoSs caused by changes of goals or subsystem capacity. All these elements are illustrated in this paper using a SoS case study of Intelligent Autonomous Vehicles initiated by the Intelligent Transportation for Dynamic Environment (InTraDE) European project to automate the port container logistic.

Reinforcement learning (RL) has emerged as a key paradigm for improving the reasoning capabilities of large language models (LLMs). However, existing reward systems, such as rule-based and reward-model-based, often exhibit issues such as unstable optimization and reward hacking. In this work, we revisit the general reasoning of LLMs from a geometric perspective, conceptualizing it as a coupled manifold composed of three interdependent sub-manifolds: logical deduction, evaluation, and representation. Based on this perspective, response generation in RL can be interpreted as a decoupling process from the evaluation manifold, while reward estimation corresponds to a decoupling process from the logical deduction manifold. The limitations of rule-based and reward-model RL systems can be geometrically interpreted as the mismatch of policy-reward manifolds during RL process. To address the aforementioned misalignment, we propose Decoupling and Coupling Reinforcement Learning (DCRL) framework, which incorporates two key components: (1) a syllogistic logic-based prompt evolution mechanism that dynamically refines reward rubrics to enhance the expressiveness of the reward manifold; and (2) a policy-reward re-coupling mechanism that jointly updates the reward and policy models, ensuring consistent evaluation and mitigating manifold mismatch during training. Theoretical analysis and extensive experiments across multiple reasoning domains demonstrate that DCRL consistently outperforms both rule-based and reward-model baselines. Notably, a Qwen3-4B model trained under DCRL surpasses a Qwen3-32B baseline and approaches the performance of a Qwen3-235B model, highlighting superior effectiveness and generalization in RL.

Deployment requires an agent to turn application code into a running system whose services connect, become ready, and remain observable. FDE-Bench evaluates this capability with 136 deployment-configuration tasks spanning Docker images, multi-service Compose stacks, and Kubernetes, in greenfield and diagnose-and-repair modes. Agents submit declarative artifacts that are collected, rebuilt, and redeployed in a pristine environment. Four gated binary check layers measure build, readiness, behavior, and conformance to the deployment specification, using programmatic checks without an LLM judge. A four-arm release gate requires a resolving reference solution and rejects tasks solved by do-nothing, specification-transcription, or generic-stub submissions. The released check annotations expose the link between 2,145 checks and their specifications, including seven documented gaps. Three additional adversarial strategies test shortcuts in the grading signals; none resolves any of the 135 tasks they cover, while a vacuous health probe passes readiness and exposes the need for downstream checks. On the 136-task evaluation grid, seven language models from four providers use the same four-tool scaffold and resolve 52.9-75.0 percent of tasks. The three zero-intelligence floors resolve none and reach a mean Deployment Score of at most 0.44. Readiness is the largest failure stage, accounting for 110 of 313 unresolved episodes. Mean resolution rate is 30.7 percentage points higher on the repair task group than on the disjoint greenfield group, with a positive gap for every model; ten tasks resist all seven. In a 25-task case study, one practicing engineer directing Claude-Sonnet-5 resolves 92 percent against 72 percent for the autonomous baseline. FDE-Bench links deployment success and failure to artifacts that can be inspected and replayed.

Standardised, machine-readable cybersecurity playbooks provide a basis for portable, shareable, and reusable incident-response logic. OASIS CACAO provides a vendor-neutral representation for such playbooks, but not the product-specific integration artefacts needed to invoke external products and services. We introduce the Common CACAO Registry (CCR), an open, provenance-aware registry of CACAO HTTP-API connector envelopes. Each envelope captures an API operation's command, inputs, target, authentication-related information, provenance, validation evidence, and maturity metadata. CCR is quality-gated, with acceptance requiring both CACAO v2 schema validity and a mean back-validation score of at least 0.8 against the source OpenAPI operation, while a six-level maturity model records progressively stronger evidence and distinguishes gate acceptance from operational readiness. To seed CCR, we develop a hybrid OpenAPI-to-CACAO pipeline. Deterministic code extracts source-derived interface facts, generates identifiers, wires cross-references, and validates structure, while a constrained LLM provides bounded semantic enrichment, including action naming, authentication interpretation, and CACAO activity annotation. Evaluation across eight security APIs yields 713 CACAO-schema-valid envelopes with a mean back-validation score of 91.5%, of which 675 produce well-formed, dispatchable HTTP requests in a local harness. Comparison with a deterministic rule-based baseline shows that mechanical API structure is preserved more reliably through rule-based translation, while the LLM contributes bounded semantic enrichment, most notably CACAO activity annotation. Together, these results support CCR as reusable integration infrastructure for CACAO action steps and as an initial foundation for a broader common European registry.

The brain does not rely on a single type of neuron to perform all kinds of tasks; instead, it designs different neurons for different tasks. The concept of task-based neurons represents a paradigm shift compared to task-based architectures. It argues that solving a specific problem requires customized neurons, as task-based neurons capture useful prior knowledge from task-related data. To facilitate the use of task-based neurons in scientific research and industrial applications, we introduce TNLearn, an open-source Python package that provides automated construction of task-based neurons and networks, enabling smooth training of task-based networks. Comprehensive documentation, including technical exposition, API reference, and representative examples, is available online. TNLearn is open-sourced at https://github.com/NewT123-WM/tnlearn and has become a PyTorch ecosystem project.

Avatar-streaming systems are commonly evaluated with image and video quality assessment (IQA/VQA) metrics, implicitly treating visual fidelity as a proxy for communicative success. We test this assumption through a controlled behavioral study of rendered 3D avatars across a pristine condition and fourteen geometric, photometric, temporal, and combined distortions. Fifty-nine participants contributed 2,688 judgments of perceived action, response confidence, and visual quality. We identify Misleading Quality in this dataset as distorted renderings that retain above-average perceived quality but yield below-average action-recognition accuracy. We also derive an Intent Quality Score (IQS) combining recognition correctness and confidence as the behavioral target for objective metrics. Among 126 distorted content–condition cells, 31 (24.6%) exhibited Misleading Quality; temporal and geometric distortions showed the highest rates, at 50.0% and 31.1%, respectively. The results reveal a quality–accuracy dissociation where distortion families affect appearance and communication differently. Across 24 direct-scoring IQA/VQA metrics and three supervised feature-regression baselines, alignment with IQS remained limited; at $\lambda=0.5$, the best leave-one-content-out baseline reached PLCC $=0.4435$. Under this controlled protocol, visual fidelity alone is insufficient for avatar communication, motivating intent-aware quality assessment and streaming objectives.

Studying syntactic patterns in naturally occurring language requires a large parsed corpus, but manual annotation is costly and difficult to scale. Thai has a manually annotated dependency treebank for training and evaluating parsers, but lacks a large automatically parsed corpus for quantitative syntactic research. We present ThaiTrees, a 342M-token corpus drawn from news, Wikipedia, spoken transcripts, and social media. We develop a reproducible pipeline for cleaning, processing, and parsing Thai text under the Universal Dependencies framework. The resulting corpus makes grammatical relations searchable and supports the study of syntactic distributions. We release a frequency lexicon and CoNLL-U parses in machine-readable formats suitable for both AI-assisted and conventional programmatic analysis.

We investigate the semi-discrete Active Flux numerical method on Cartesian grids applied to two-dimensional linear and nonlinear shallow water equations with Coriolis source terms. These hyperbolic systems admit non-trivial stationary solutions governed by a geostrophic equilibrium. In this setting, we analyze the stationarity-preserving properties of our semi-discrete Active Flux formulation. To do so for the linear system, we apply a discrete spatial Fourier transform to the semi-discrete method and analyze the resulting evolution matrix. We show that its discrete kernel is a non-trivial discretization of the geostrophic equilibrium, which implies that the method preserves a discrete geostrophic equilibrium state. This is an inherent feature of Active Flux and is achieved without any additional modifications. Numerical experiments confirm the method's ability to maintain geostrophic stationary states for the linear system. Additionally, we numerically investigate the performance of the Active Flux method for the nonlinear shallow water system.

Multimodal learning is emerging as a powerful paradigm for AI for Physics (AI4Physics), where predicting physical systems requires the joint interpretation of heterogeneous observations, measurements, and domain knowledge. However, existing approaches typically represent physical quantities and governing equations as generic numerical or textual tokens, overlooking the physical constraints that determine their spatiotemporal interactions. To address this limitation, we introduce the physical-field modality and propose PhyMo, a physics-grounded multimodal framework that organizes heterogeneous measurements through PDE-associated operators. PhyMo follows a three-stage learning procedure: the physical-field encoder is first pretrained through field reconstruction under PDE residual supervision, its representations are subsequently aligned with visual embeddings in a shared latent space, and the fused multimodal representations are finally processed by corresponding downstream prediction heads. Experiments on five datasets spanning diverse physical environments show that PhyMo achieves state-of-the-art performance, compared to the strongest baseline on each dataset, demonstrating the superiority of PhyMo on multimodal representation learning in AI4Physics.

Magnetic navigation has emerged as a promising alternative for navigation in Global Positioning System (GPS)-denied environments, leveraging geomagnetic field maps in conjunction with onboard magnetometer measurements. However, its performance is highly sensitive to trajectory-dependent observability, which limits its practical effectiveness under conventional flight paths. This paper proposes an optimal trajectory design framework for magnetic navigation that maximizes information content along the flight path. The trajectory generation problem is formulated as an optimal control problem that minimizes the posterior Cramér–Rao lower bound on the position estimation error, subject to a penalty on path length. The resulting trajectories are non-intuitive and significantly enhance the observability of the navigation system. Simulation results demonstrate that the proposed optimal trajectories yield substantial reductions in estimation error compared to conventional straight-line trajectories, highlighting the critical role of trajectory design in enabling high-accuracy magnetic navigation. These findings suggest that trajectory optimization can substantially improve the viability of magnetic navigation as a robust alternative for aerospace applications in GPS-denied environments.

The increasing adoption of software defined networking and network function virtualization, combined with rapid advances in Machine Learning (ML), is driving the evolution toward intelligent network softwarization across cloud, edge, and distributed environments. However, deploying complex learning models in such heterogeneous environments introduces challenges in latency, computation, and energy consumption. Knowledge Distillation (KD) has emerged as a promising approach to enable lightweight and efficient intelligence by transferring knowledge from high-capacity teacher models to compact student models. Despite its extensive study in general ML domains, the integration of KD into intelligent softwarized networks remains fragmented and underexplored. In this paper, we review and classify recent efforts that incorporate KD within softwarized networks, analyze current trends and limitations, and outline open challenges and future directions toward scalable, adaptive, and energy-aware distillation mechanisms.

We propose a time-stepping minimum-residual (MinRes) framework for transient coercive variational problems, applicable to finite-element and neural-network trial approximations. For the Backward Euler scheme, we measure the residual at each time level in the dual norm induced by the steady test-space norm, scaled by the time step. This choice yields stability estimates controlling a discrete parabolic energy quantity, up to the time-discretization defect. We develop conforming and broken-test formulations and specialize the analysis to diffusion-advection-reaction problems. For finite-dimensional test spaces, we derive fully discrete reliability estimates by supplementing the computable discrete residual with a complementary contribution that accounts for residual components the test space does not resolve. We then extend the framework to neural-network trial classes and obtain computable residual decompositions for conforming and broken polynomial test spaces. Numerical experiments confirm the expected finite-element convergence rates, examine the effect of test-space refinement for a global space-time neural approximation, and demonstrate residual-driven spatial refinement for a transient problem with a moving localized feature.

Embodied reinforcement learning (RL) improves model capabilities with a pipeline of environment simulation, action generation, and model updates. These stages show heterogeneous CPU and GPU demands, making efficient resource utilization difficult. Recent systems overlap rollout (simulation and generation) with training for efficiency, but exclusive GPU allocation and synchronized barrier in rollout still leave substantial hardware resource waste. In this paper, we present EBRL, an asynchronous embodied RL training system with two core techniques. The asynchronous pipelined scheduler overlaps rollout and training, pipelines simulation and generation across environment groups, and carries out each environment independently, eliminating synchronization stalls. The fine-grained resource manager pools CPU cores and GPU streaming multiprocessors, and uses stage profiles and runtime feedback to adjust resource quotas and batch sizes to meet the shifting demands among stages. We implement EBRL on RLinf and evaluate it with four embodied policies and four simulation benchmarks across heterogeneous GPU testbeds. Experiments show that EBRL achieves 1.30-3.47 times the end-to-end rollout throughput and 2.5 times of training convergency compared to the SOTA embodied RL systems.

Score-based diffusion models are increasingly considered in settings where the underlying data distribution may differ from the training distribution, yet existing theoretical guarantees largely focus on the no-shift setting. In this work, we study robust score estimation under Wasserstein perturbations of a reference distribution. For the Ornstein–Uhlenbeck diffusion, we show that robust estimation decomposes into two fundamental components: the statistical cost of learning the reference distribution and the intrinsic cost of distribution shift. The latter scales quadratically with the Wasserstein radius, and this dependence is minimax optimal. We construct an explicit finite-sample estimator achieving the resulting robust minimax rate without knowing the shift radius. When the reference distribution lies on an unknown low-dimensional subspace, the statistical term adapts to the intrinsic dimension while the shift cost remains unchanged. Finally, we show that the same decomposition governs positive-time reverse sampling and obtain matching minimax guarantees in KL divergence. Together, these results characterize how finite data, intrinsic dimension, and distribution shift affect the robustness of score-based diffusion models.

This paper presents a representation-aware framework for geometry-based screening of hull-form producibility at early design stages. The proposed signature combines dimensionless total and signed developability deviation with curvature-class area fractions, distributed fields, metric-specific validity, and representation provenance. These descriptors characterize surface features relevant to plate forming and developability, but are not calibrated predictors of fabrication cost, forming effort, or process feasibility. Native IGES/STEP boundary representations (BReps) are evaluated through direct differential geometry and trimmed-domain quadrature, whereas triangulated surfaces use discrete curvature recovery and area-weighted aggregation. Analytical and semi-analytical controls verify the formulation, while matched-face BRep-to-mesh tests assess discrete curvature recovery. Application to DTMB 5415, KCS, JBC, and KVLCC2M shows that curvature intensity and areal extent provide complementary information and that derivative-based outcomes can be representation sensitive. KCS, for example, exhibits approximately 24% greater developability deviation than DTMB 5415, while double-curved regions occupy 72.9% of its valid surface versus nearly the entire DTMB valid surface. The resulting quantities provide an early geometric screening layer for subsequent use as objectives, constraints, surrogate responses, or design-space features. HullProd, the companion open-source software, implements the signature, distributed fields, validity, and provenance.

The safety of a tool-using language model agent is usually treated as a property of the model alone. We give controlled, full-precision evidence that it is instead a joint property of the model and the software that renders its chat template and parses its tool calls, the decoding harness, and that both halves are attackable from untrusted input. On the released gpt-oss-20b reasoning model under its published tool sandbox, appending a single string of the model's own channel-control tokens to a user message makes the tokenizer render a reasoning turn that is already complete, so the model writes no chain-of-thought and proceeds directly to the tool call. Across forty tasks the model already completes, the reasoning channel falls from a mean of 52.5 tokens to zero on every trial while the http.post still fires on every trial. A rule monitor and a cross-family language-model monitor detect the unsafe request on all plain trials and no forged trials, and on overtly malicious requests the attack converts 39.6% of the model's refusals into completed exfiltrations. Separately, whether an identical tool-call generation fires is decided by the harness parser, not the model: a truncation-tolerant regular expression fires a call whose closing token is missing while a strict one drops it, and two parsers shipped for the Gemma agent give opposite outcomes on identical greedy generations, firing on all twenty-four trials and on none. We show the suppression can be delivered indirectly and characterize its dependence on the chat template across two more reasoning models, and we evaluate input sanitization, parser hardening, and empty-reasoning detection as defenses; flagging an absent trace catches the basic attack but not an adaptive benign decoy. All measurements use greedy decoding on publicly released models. Code and per-trial logs: https://github.com/Usama1002/deleting-the-trace

A major video platform runs on thousands of microservices, each request propagating a context so downstream work can be traced and governed. At handoffs outside instrumented paths, e.g., custom queues and callbacks, the payload continues but the context does not, and the request still succeeds under existing tests. Such breaks are silent and widespread: 673 of 1,133 services carried at least one. Backstitch, a specialized agentic system, repairs them using the surviving execution as its reference: replay determines whether a suspicious call is request-correlated, source analysis reaches the responsible handoff, a bounded change restores its contract, and the same replay validates the fix. Repairs restore the causal chain without disturbing the work it describes: breaks at 240 of the repaired calls fell from 90.46% to 4.69%, and over 112 days the fleet's break rate more than halved.