Papers

Interference management in multi-stream multi-input multi-output (MIMO) device-to-device (D2D) networks often leads to weighted sum-rate maximization with sum-log-determinant involving matrix-valued signal-to-interference-plus-noise ratios (SINRs). The state-of-the-art paradigms, including weighted minimum mean-square error (WMMSE) and fractional programming (FP), have achieved tremendous success in link scheduling, power control, and beamforming problems. Recently, an upgraded FP approach, named surrogate-enhanced FP (SEFP) in a scalar form, has attained improved performance in joint uplink scheduling and power control in coordinated multicell SISO networks. The proposed SEFP improves the surrogate construction of the classical Lagrangian dual transform plus quadratic transform for logarithmic fractional objectives with a novel reciprocal-inverse transform (RIT), yet its extension to the matrix form for MIMO settings does not seem straightforward because the matrix ratio and the auxiliary matrix generally do not commute. In this paper, by leveraging mathematical tools from Hermitian functional calculus, we extend RIT to matrix ratios by lifting the scalar RIT along the eigendirections of an auxiliary matrix and recasting the resulting directional construction in an operator form. Following, we develop a matrix SEFP framework for weighted sum-log-determinant maximization. Further, we establish a unified view of SEFP and the recently proposed XMMSE method, specify when the two algorithms attribute to identical minorization-maximization (MM) surrogates and variable update trajectories, and prove XMMSE can be considered as a special case of SEFP under the algorithmic family perspective. Inspired by the unified view, we develop SEFPLinQ for the joint scheduling and beamforming optimization in flexible-association MIMO D2D networks. Numerical results demonstrate consistent performance gains.

Pretrained transformers use little of their depth to follow references in context. Thirteen base models reliably follow only 1.4-3.6 lines, and extra pretrained loops add little. A task-trained rank-8 LoRA at one early layer extends this computation with all model weights frozen. Qwen3-8B improves from 15.5% to 99% exact accuracy on 24-line chains; a longer-trained LoRA reaches 50 lines. Ouro-1.4B reaches 60 lines after four loops and at least 160 after eight. The LoRA starts a relay: program lines pass on their chain identity through a short range of middle layers. Frozen heads read progressively further up the chain, and removing parent-line attention stops the relay. A frozen-model measurement locates the last useful intervention layer within tolerance in three of four held-out models. Task-specific LoRAs also improve MuSiQue. Default answers therefore understate the computation accessible through a tiny edit. Code and an interactive demo are available at https://lunamos.github.io/stop-thinking-too-early/

Analog in-memory computing (AIMC) offers an alternative for model training by executing matrix operations directly where weights are stored. However, scaling AIMC to train modern deep models remains an open challenge due to severe hardware non-idealities, including physical weights with finite dynamic range and write granularity, analog-digital converters with finite resolution, and noisy and asymmetric updates. Guided by the insight that gradient accumulation is sensitive to precision and rounding errors, we adopt a mixed-precision training paradigm: executing forward and backward matrix multiplications in the analog domain while computing weight gradients in the digital domain. To enable scalable training, we present a holistic system-algorithm co-design that co-optimizes weight mapping to ensure well-conditioned physical and logical weight profiles, couples a preconditioned optimizer with threshold-triggered open-loop pulsing to stabilize training trajectories, and aligns converter dynamic ranges to suppress quantization errors. Evaluated via hardware-calibrated architectural simulations calibrated with electrochemical RAM measurements, our framework scales Transformer training up to $123\text{M}$ parameters with validation loss scaling as $L\propto N^{-0.231}$, where $N$ is the parameter count, comparable to $L\propto N^{-0.238}$ for digital training.

Wide-area monitoring, protection, and control (WAMPAC) systems damp inter-area oscillations in large interconnected grids, but their reliance on synchronized PMU measurements carried over wide-area networks exposes them to false data injection (FDI) attacks. This paper presents an attack-resiliency analytics framework that formally models the coupled dynamics of the wide-area damping loop, automatic generation control, and the governor and excitation systems, and formulates the optimal stealthy FDI attack as a mixed-integer linear program (MILP). The anomaly detection model (ADM) enters as a replaceable constraint set: either a static bad-data detection (BDD) rule or a boundary learned from benign operation calibrated to a common false-positive rate. On the IEEE 39 and 118 bus systems, with reference dynamics validated on an OPAL-RT hardware-in-the-loop testbed, the optimal attack with wide-area access attains 3.3 and 8.1 times the benign objective and reaches a 0.5Hz frequency excursion more than twice as fast as an attack confined to automatic generation control. Learned boundaries reduce the attack objective by 15.5 55.0% and prevent over-frequency relay trips in all configurations, with feasible stealthy attacks remaining in every case, and residual risk tracks the width of the learned boundary rather than the detector family.

Foot-ground interaction signals recorded by quadruped robots may enable spatially distributed, in situ characterization of soil strength. As a first step, we test whether the internal friction angle $\phi$ of cohesionless soil can be identified from the force history of a simplified rotating leg. A two-dimensional continuum model implemented with the material point method, benchmarked against measured rotating-leg force histories, generates the training data, and two Gaussian-process surrogates support Bayesian inversion of the full histories. In matched-model experiments, the framework recovers 14 off-grid friction angles with a median absolute error of approximately $0.1^\circ$ (maximum $\sim 0.7^\circ$); the reported credible intervals contain the true value in every case. These results establish that $\phi$ is identifiable when the forward model is correctly specified, and support further development of proprioceptive soil sensing for spatially variable terrain, with applications from physics-grounded world models for robot training to post-wildfire slope assessment.

Query-agnostic streaming video understanding requires vision-language models to continuously compress an indefinitely growing visual stream into a bounded memory before future queries are known. The performance depends critically on the memory mechanism–what observations to preserve, how to represent and consolidate them, and what information to retrieve when a query eventually arrives. Rather than designing a single memory architecture by hand, we formulate memory design as a search problem over executable memory programs. We introduce a lightweight domain-specific language that expresses memory mechanisms through structured primitives for representation, admission, retention, consolidation, budgeting, and retrieval, while enforcing causal and bounded-memory constraints. Although structured, the derived program space remains large and contains heterogeneous, conditionally dependent design choices whose effects can only be assessed via downstream execution. We therefore propose MemEvo, an LLM-driven auto-research framework that uses pretrained LLM as a semantics-aware proposal model to iteratively generate and refine candidate memory programs based on accumulated experimental feedback. At runtime, a deterministic evaluation pipeline validates and evaluates each candidate, while the underlying vision-language model remains frozen throughout discovery. We finally produce a training-free, bounded-memory mechanism. Extensive experiments on StreamingBench and OVO-Bench demonstrate strong streaming video understanding performance together with substantial context and inference efficiency.

LLM agents deployed in real-world environments continually encounter new tasks and safety risks, while execution feedback typically becomes available only after each task is completed. However, existing self-evolving approaches commonly rely on multiple rounds of optimization over fixed and repeatedly accessible task distributions, fundamentally differing from test-time adaptation in real-world deployment, where only experience accumulated from past tasks can be used to improve safety decisions on future unseen tasks. To address this limitation, we propose SafeCoEvo, a test-time Harness-Guard co-evolution framework for LLM agent safety that enables the external safety system to continually adapt from accumulated runtime experience. SafeCoEvo jointly improves two complementary safety capabilities at different timescales: S-Harness rapidly externalizes recent runtime experience into updatable explicit safety knowledge that can promptly influence subsequent tasks, while GuardVPO internalizes accumulated runtime safety experience over a longer timescale into parametric risk-judgment capabilities. By combining short-term rapid adaptation with long-term capability consolidation, SafeCoEvo continually improves the agent's safety capabilities, reducing the unsafe outcome rate by 10.05% while improving the task success rate by 12.15% over the strongest baseline, thereby achieving simultaneous gains in safety and task utility.

Passive transformations of an $n$-mode squeezed probe become locally unidentifiable when the quantum Fisher matrix loses rank. For pure zero-mean Gaussian probes, we show that the $C_{n}$ restricted-root system of $\mathrm{Sp}(2n,\mathbb{R})/\mathrm{U}(n)$ governs this exact sloppiness. For squeezing magnitudes $r_{1},\ldots,r_{n}$ in the canonical basis, the roots $2r_{j}$ and $r_{j}\pm r_{k}$ label the local phase rotations and two beam-splitter quadratures, and their vanishing identifies every Fisher-null direction. The associated Fisher weights quantify the approach to each singular wall. At fixed positive mean photon number, we solve the E-optimal probe-design problem of maximizing the smallest passive Fisher eigenvalue in two explicit generator normalizations. In the canonical phase and beam-splitter angle convention, the unique optimum in the fundamental Weyl chamber is an arithmetic progression approaching the consecutive-odd-integer dual-Weyl direction at large resource. With an invariant generator norm, the smallest eigenvalue is independent of the passive frame and the optimum lies exactly on that ray at every resource. Uniform squeezing instead minimizes the local-phase A-optimal cost while leaving one beam-splitter quadrature unidentifiable for every mode pair. Finally, the two quadratures of each identifiable mode pair saturate the quantum-geometric incompatibility bound, and the largest compatible passive submodel has $n(n+1)/2$ parameters at a regular spectrum. The resulting classification separates local identifiability, sensitivity optimization, and simultaneous attainability.

Scheduling problems arise from repeatedly selecting one item from a set of candidates based on their states. These problems often reduce to assigning priority scores and choosing the highest-ranked item. In this work, we propose a factorized scheduling principle (FSP) framework to learn interpretable and transferable scheduling rules. The FSP framework represents system states as condition distributions and decomposes a global scheduling principle into additive univariate and pairwise components with identifiability constraints. The scheduling principle enables the framework to maintain a simple priority-based structure during deployment. This principle is learned by using a policy-based objective combined with a temporal-difference signal defined on the condition distribution. Experiments on synthetic and realistic scheduling tasks demonstrate the FSP framework's strong performance, interpretability, and zero-shot generalization across different system scales.

Audio large language models (ALLMs) can reason about the content of audio recordings to perform complex tasks. However, these capabilities usually collapse in real-world environments when background noise and competing sources mix the target sound. Inspired by long-term memory in human listening, we propose Long-Term Memory-Guided Audio Enhancement (LTM-AE) to improve selective target perception by refining the audio representations of ALLMs without training. LTM-AE extracts representations in hidden states from separate clean reference recordings as long-term memory for each category, guiding enhancement toward a user-specified listening target. We reconstruct incoming audio tokens in the selected category long-term memory and interpolate the reconstructions with the original tokens before language backbone decoding. This interpolation controls the influence of stored auditory experience while keeping all ALLM parameters fixed. Diagnostic readouts across twenty sound categories and three ALLMs show that LTM-AE strengthens responses to a specified target amid three interfering sources. Averaged over constrained and free-form classification, accuracy gains over raw mixtures range from 29.53 to 46.15 percentage points across multiple open source models. For speech content recovery, LTM-AE with an additional learned token-level gate reduces Qwen2-Audio's word error rate from 23.07% to 14.77%. This work takes an initial step toward using principles of human long-term memory to enhance ALLMs for real-world listening. Our code is available at https://github.com/aynlp/ltm-audio-code

Agentic systems are now being widely used to orchestrate tools and reason over long contexts. However, the improving capabilities of the large language models powering these agents also create new attack surfaces for indirect prompt injection. In particular, an attacker may not need to place a complete malicious instruction in retrieved content if the agent can reconstruct the objective from incomplete fragments distributed across a long context. In this work, we introduce adaptive long-context prompt injection (AdaLCPI), which combines long-context fragmentation with adaptive search. AdaLCPI splits an attack objective into incomplete fragments, embeds them in external content retrieved through the agent's tools, and uses a reconstruction cue to prompt the agent to combine them. It then iteratively refines the fragments and cue with OpenEvolve using graded scoring and natural-language execution feedback from the target agent. Empirically, AdaLCPI achieves higher attack success than strong adaptive baselines, reaching 61.4% macro-average ASR compared with 32.8% for Trojan Hippo-style and 30.0% for AgentVigil. Safety evaluations should therefore test whether agents remain robust when harmful objectives must be reconstructed from incomplete fragments.

Visuomotor policies for humanoid loco-manipulation must generalize across object poses and lighting from only a handful of demonstrations. 3D Diffusion Policy (DP3) conditions a diffusion-based action generator on point-cloud features, but its PointNet-style encoder has no built-in equivariance to the rotations, translations, and scalings (SIM(3)) that manipulation tasks respect. EquiBot closed this gap for wheeled manipulators with a SIM(3)-equivariant Vector Neuron Network (VNN) encoder. We extend this to a substantially more complex embodiment, the 43-joint Unitree G1 humanoid, and propose EquivDP3: a two-stage policy where a high-level diffusion planner with a SIM(3)-equivariant VNN encoder emits 6 Hz whole-body command chunks, executed at 50 Hz by a frozen, pre-trained RL locomotion policy and a differential inverse-kinematics module for the arms, trained end-to-end by behavior cloning. Across two simulated IsaacLab benchmarks and four non-equivariant baselines (5-100 demonstrations, in- and out-of-distribution), EquivDP3's advantage concentrates in the low-data regime: at 5-10 demonstrations it reaches 67.1% success versus 38.2-52.4% for the baselines, while by 50-100 all encoders converge (74.3-85.2%) and the ordering is no longer meaningful. A proprioception-only control confirms this gap is genuinely perceptual: with the point cloud removed, success drops to 31% vs. 60% (EquivDP3) at 5 demonstrations and 78% vs. 99% at 10, but vanishes by 50-100, showing the high-data plateau reflects a benchmark ceiling, not five encoders learning the same invariance. The encoder costs only 0.8 ms of extra latency per action chunk over the PointNet encoder it replaces. Baking geometric symmetry into a hierarchical diffusion policy's perception backbone is a practical, nearly free way to improve data efficiency for humanoid loco-manipulation when demonstrations are scarce.

A layer of diagonal quantum gates can be specified by an integer-valued function on computational basis states. This description suggests implementing the layer by evaluating that function reversibly and translating its value into a phase using a shared quantum register. We develop this viewpoint for parallel phase kickback (PPK): the weighted sum of several rotation parameters is computed coherently, added once to a phase-gradient state, and then uncomputed. The construction separates the cost of representing a phase function from the cost of applying it. We prove correctness, give explicit error bounds, and analyze logical gate counts for general weights, pairs of rotations, and weights with disjoint binary support. With Clifford-only encoding, a batch uses at most $4(b-1)$ T gates at $b$-bit phase resolution, excluding initialization. The first phase reference is prepared with independent rotations synthesized by Gridsynth; additional phase-gradient states then follow from the disjoint-support construction with $O(b)$ T gates each. The resulting bounds identify when shared phase arithmetic reduces amortized rotation cost and when encoding overhead limits that advantage.

While LLM-based attackers exhibit growing proficiency in vulnerability exploitation, most existing cybersecurity benchmarks suffer from single-stage truncation, prematurely terminating evaluation upon initial access. In practice, initial footholds are exceptionally fragile across operational disruptions such as service restarts and host reboots. Whether LLM-based attackers can establish and maintain durable footholds beyond initial compromise remains a central blind spot in cybersecurity evaluation. To bridge this gap, we introduce CyberPersistBench, the first benchmark dedicated to post-compromise installation and persistence. Decoupled from upfront exploitation, CyberPersistBench frames persistence as an adversarial survival task in which agents use native host mechanisms to maintain footholds across staged system disruptions. Deterministic checks support a six-level scoring method (L1–L6) spanning installation and persistence. The benchmark comprises 203 core tasks across seven categories, augmented by multi-host and active defense extensions. Empirical evaluations across five frontier agents show that autonomous persistence remains limited (27.6%–44.8%) and drops further on defense-enabled tasks (5.5%–13.3%); nonetheless, these results reveal an emerging cyberattack risk, underscoring the necessity of benchmarking post-compromise persistence. CyberPersistBench thus establishes a foundational benchmark for post-compromise installation and persistence, delineating the operational boundaries of autonomous cyber agents.

Reinforcement Learning with Verifiable Rewards (RLVR) has been extended to Large Vision-Language Models (LVLMs), and perception-aware methods further encourage policies to rely on visual evidence. Yet relying on the image does not guarantee that visual claims are supported by it. Before RL training, 27.81% of the correctly answered responses of Qwen2.5-VL-7B on four multimodal reasoning benchmarks contain at least one direct visual claim that the image does not support. Since outcome-level RL rewards each response as a whole, these claims inherit the positive credit of the correct answer. We introduce a fixed-rollout counterfactual diagnostic that re-scores the same response under an intervened image to separate Evidence-Function Sensitivity (EFS), how strongly the model's predictions change, from claim persistence, whether the model keeps supporting the same claim rather than retracting it. The diagnostic reveals Sensitivity-Persistence Decoupling (SPD): under DAPO and VPPO, EFS increases and claims become more retractable overall, yet unsupported claims become significantly more persistent, whereas GRPO raises EFS without this deterioration. We therefore propose Persistence-Aware Credit Gating (PACG), which attenuates positive credit for unusually persistent visual claims and leaves all other credit unchanged. It requires no supported/unsupported labels and adds no inference cost. On Qwen2.5-VL-7B, PACG raises the nine-benchmark average over three seeds from 58.1% to 59.9% with DAPO and from 59.8% to 60.9% with VPPO, while making unsupported claims more retractable. The gains extend to a larger model, a newer backbone, and the accuracy of HallusionBench also improves consistently. These results suggest that visual sensitivity and claim retractability are complementary dimensions of multimodal credit assignment.

Quantum Fisher information (QFI) quantifies the sensitivity of a quantum state to a parameter change and plays a key role in quantum metrology. Meanwhile, quantum coherence is a crucial resource for quantum information processing. However, despite extensive studies on various aspects of quantum metrology, the interplay between the dynamics of the QFI and quantum coherence remains unexplored. Here, we explore the factorization relationship between the QFI and quantum coherence, both theoretically and experimentally. We prepare pure states in qubit and qutrit systems and apply unitary evolution to investigate this relationship. Our results confirm the factorization law, offering strong evidence for this relationship. Furthermore, we establish a lower bound on the variance of the estimated parameter that does not require the final state information. This approach reveals a deep connection between the QFI and quantum coherence, providing insights into quantum parameter estimation. These findings have potential applications in quantum sensing and quantum metrology.

Indirect prompt injection makes an LLM agent treat untrusted retrieved text as instructions. We present CounterSteer, an inference-time defense that suppresses this behavior inside the model. Per model, a five-step recipe fits a residual-stream direction from paired episodes differing only in whether an embedded instruction is followed, and retains it only if it passes pre-specified causal and capability gates. At deployment, the direction is subtracted from every tool-result token during prefill. The edit is always on–there is no detection decision to evade–and requires no fine-tuning, auxiliary model, or added tokens, only white-box serving and tool-result span boundaries. Across five open-weights models (8B-106B, five vendor lineages), held-out attack success falls from 0.21-1.00 undefended to 0.00-0.17 defended, and AgentDojo compromise rate from 0.10-0.49 to 0.006-0.079, at 93-100% typography-normalized benign utility, with larger task-dependent costs when reasoning over steered content. A benchmark-level adaptive attacker reaching 0.67-0.73 undefended is held to roughly a quarter of that on the two most deeply evaluated models. Among the defenses we measured on capable models, those achieving lower compromise rates either lost 22-89% of benign utility or fine-tuned the served weights. White-box gradient attacks through the deployed vector compromise at most 2 of 52 episodes, and none of 2,052 replayed human red-team attacks succeeds. CounterSteer largely neutralizes instructional takeover: a black-box framing search cracks 3 of 18 development samples. Parameter manipulation–attacker-chosen arguments in otherwise legitimate calls–is only partially resisted (13 of 18); the decision becomes linearly readable at argument emission but not at the examined pre-generation sites, and is not removed by the tested prefill- or decode-time steering, motivating argument-provenance controls.

Checkpoint selection is a routine decision in supervised fine-tuning (SFT): training produces multiple checkpoints, but only one is retained. Yet fixed-budget comparisons do not by themselves distinguish three empirical claims: whether more validation data improve checkpoint selection, whether a selection rule outperforms validation-loss selection, and whether it improves over simply retaining the final checkpoint. We therefore treat checkpoint selection as a finite-information decision problem. Holding completed training trajectories, candidate checkpoints, and independent test items fixed, we vary the validation budget and separately measure improvement from additional validation data, gain over negative log-likelihood (NLL) selection, and gain over the final checkpoint. Across 60 mathematical SFT trajectories and 19 configurations, increasing the validation budget from 32 to 305-313 examples raises independent-test accuracy by 0.32 percentage points (pp) for generated-accuracy selection and 0.29 pp for checkpoint agreement, with 95% configuration-bootstrap CIs of [0.10, 0.56] and [0.11, 0.50], respectively. At the full validation budget, the two generation-based rules outperform matched NLL selection by 0.71 and 0.85 pp, respectively, while their gains over the final checkpoint remain unresolved. A cross-domain replication on 12 newly trained Commonsense trajectories shows the same qualitative separation: increasing the validation budget from 32 to 1,024 questions improves generated-accuracy and checkpoint-agreement selection by 0.87 and 0.27 pp, while gains over the final checkpoint again remain unresolved. Together, these results show that benefiting from more validation data, outperforming NLL selection, and outperforming the final checkpoint are distinct empirical claims that require separate evidence.

Diffusion models have emerged as state-of-the-art generative models, with recent extensions from Euclidean spaces to Riemannian manifolds. However, existing convergence guarantees for Riemannian diffusion models typically require $\tilde{O}(\mathrm{poly}(d,T)/\epsilon^2)$ score evaluations, with potentially unfavorable dependence on the dimension. In this work, we develop a general framework that separates score discretization from Brownian-motion simulation and allows multiple geodesic random-walk steps per score evaluation. Under nonnegative Ricci curvature assumption and an exact Brownian-motion simulation oracle, we show that $\tilde{O}(d/\epsilon^2)$ score evaluations suffice to achieve an $\epsilon^2$ KL divergence from the target distribution, matching the existing convergence rate of Euclidean diffusion models. We further show that $\tilde{O}(d^4T/\epsilon^2)$ geodesic random-walk steps suffice to approximate the required drifted Brownian motion to $\epsilon$ total variation error. Combining these results yields a sampling scheme with $\tilde{O}(d/\epsilon^2)$ score evaluations and $\tilde{O}(d^4T/\epsilon^2)$ geodesic random-walk steps, motivating multiple random-walk steps between consecutive score evaluations. Our results provide a sharper characterization of the convergence and sampling complexity of Riemannian diffusion models.

The one-dimensional compressible Navier–Stokes–Fourier–Poisson system under the Boltzmann relation is studied for the existence of small-amplitude viscous shock profiles connecting the Rankine–Hugoniot states of the quasineutral Euler system with effective pressure $P_{\mathrm{eff}}=\rho(\theta+1)$. This profile is unique up to translation, of Lax type, and orbitally stable in time without a zero-mass assumption on the perturbation. The proof is based on a center-manifold reduction of the traveling-wave ODE and on the method of $a$-contraction with shifts, together with a relative-entropy functional augmented by electric energy and a sharp estimate of the viscous flux associated with $P_{\mathrm{eff}}$.

In a referential that rotates with Earth, the dynamics of the configurations of satellites in a Delta Walker constellation can be analyzed as a dynamical system as a function of a translation on the torus. This paper extends this type of analysis to routing over a Walker constellation, leveraging fixed ground relays. It considers any deterministic, time-invariant routing rule on such a collection of satellites and relays. The main object of interest is the routing table associated with this rule, which specifies, for all source-destination pairs, a route made of satellites and relays between them, for instance the shortest, together with angular information on next hop at each step of the route, which is essential for beamforming in this context. The main result is that, for each such routing rule, the routing table inherits the dichotomy of the torus flow: when the ratio of the Earth-spin and satellite angular speeds is rational, the routing table process is periodic with the same period as the constellation. When it is irrational, the table is non periodic but admits ergodic long-run averages which can be computed as ensemble averages w.r.t. an explicit invariant measure. Simulations with mixed satellite–gateway greedy routing illustrate these findings in terms of both temporal and spectral properties of the Walker routing table time series. The paper also discusses how to define metrics that cope with such non periodic fluctuations when present. This is illustrated by a comparison of the round-trip times between two far away cities as offered by a Walker constellation using greedy routing on one side, and by the currently available terrestrial fiber on the other side. It is shown how ergodicity can be used to make this comparison between the time varying instantaneous round trip times of the constellation and the constant round trip time of the fiber network meaningful.

Generative video communication has shown promise for bandwidth-constrained wireless transmission and has the potential to support personalized content delivery. In this article, we propose a dual-stream digital generative video communication (DGVC) framework that integrates a traditional digital link with a generative link. The traditional link provides source-grounded visual references, while the generative link conveys compact semantic and perceptual information for receiver-side generation. We further discuss three bandwidth-dependent operating regimes and key technologies for dual-stream coordination, synchronization, reliability, and latency control. A practical case study demonstrates the perceptual and temporal-quality benefits of DGVC under wireless fading channels. Finally, we discuss open challenges and future research directions for generative video communication.

We consider the isoperimetric problem with planar density equal to 1 on a horizontal strip and to a constant $\lambda>1$ outside the strip. We resolve a conjecture made by Cañete-Miranda Jr-Vittone by showing that every four-arc region admits a three-arc competitor with the same weighted area and strictly smaller weighted perimeter. The argument applies to every $\lambda>1$ and gives an explicit positive lower bound for the perimeter difference.

Visual emotion recognition commonly assumes that all evidence required for prediction is contained in the observed image or video. Yet the same visible reaction can convey different emotions depending on events beyond the input: tears, for example, may indicate grief or joy. We formulate Event-Grounded Emotion Recognition (EGER), where emotion recognition requires recovering the affect-determining event. We construct EGER-Bench, comprising 10,052 videos and 10,734 images across 11 emotions, two source domains, and four visual settings. A study with six annotators shows that event context raises human recognition accuracy from 33.96% to 72.08%, confirming that visual evidence alone is often insufficient. Semantic relevance alone does not solve EGER: a plausible event may imply the wrong emotion if its identity, focal-person role, relationship, or outcome is misinterpreted. We therefore propose AffectReveal, a tuning-free framework that first constructs and independently verifies evidence-grounded alternatives over these affect-critical factors. It then cross-checks the recovered event against face-masked in-media facts through bidirectional atomic evidence support, while retaining the original unmasked input for final prediction. Across three downstream models and four input settings, AffectReveal yields average UAR gains of 5.26–10.53 points. For three fine-tunable models, it also enables untuned models to outperform their fine-tuned visual-only counterparts in all 12 accuracy comparisons, without updating downstream parameters.

While Multimodal Large Language Models (MLLMs) are increasingly deployed in safety-critical domains, their reliability is threatened by multimodal implicit risks. Unlike explicit threats, these hazards emerge when individually benign text and neutral visual entities logically converge to induce unsafe outputs. Current detection methods fail to address this because they overlook the underlying risk activation mechanisms that govern cross-modal risk activation, leading to single-modality shortcut learning and hallucinated rationalizations. To bridge this gap, we first construct TriggerBench, the first dataset explicitly modeling risk compositionality (5,600 instances). By formally isolating Key Elements and Trigger Elements to build counterfactual contrastive pairs, TriggerBench eliminates risk residues and forces models to perform genuine logical deduction rather than superficial pattern matching, which provides a rigorous foundation for both large-scale training and fine-grained evaluation. Building on this, we propose a Step-Supervised Structured Reasoning training framework and employ it to train ThinkingGuard, a specialized guard model. Inspired by Situation Awareness theory, we decouple implicit risk identification into progressive cognitive stages, and utilize a step-reward Monte Carlo Tree Search algorithm to explore optimal reasoning trajectories, which are then distilled into the model through Dual-Constraint Preference Alignment. Extensive experiments across both standard and implicit safety benchmarks demonstrate that ThinkingGuard achieves strong performance. Project resources are available at https://github.com/FroggyChen/ThinkingGuard.

Motion planning on arbitrary Riemannian manifolds is an important and difficult problem that frustrates typical planning methods for Euclidean spaces. In particular, motion planning methods that approximate optimal time-to-go functions with neural networks, e.g., Neural Time Fields (NTFields), cannot be directly applied without using ad-hoc coordinate projections into higher dimensions. Using these methods directly without such projections is desirable, as it promises to provide the lowest-possible-runtime method for obtaining optimal plans on high-dimensional manifolds while using minimal model capacity. In this work, we develop a model that requires no coordinate projection and can learn arbitrary functions on Riemannian manifolds by combining splat regression models with splats defined by wrapped Gaussian distributions. We successfully apply this model for learning arrival time fields on several Riemannian manifolds, and we compare the accuracy and model size of this approach with multi-layer perceptrons adapted to work on each manifold individually.

Object re-identification (ReID) faces a recurring challenge: different identities can share highly similar global appearances, while the cues that distinguish them are localized, heterogeneous, and visible only under particular viewpoints. This challenge arises in animal ReID through markings, contours, and scars, in person ReID through subtle clothing and accessory cues, and in vehicle ReID through localized appearance details. Although visual foundation models encode such information in dense tokens, a single holistic descriptor can obscure discriminative local signals. We propose FM-ReID, an end-to-end framework that formulates local representation learning as selective competitive token routing. Its Competitive Fine-grained Mining module uses multiple mining queries and a residual query to compete for dense DINOv3 tokens. Above-prior selection retains tokens preferentially allocated to each mining query, while the residual slot receives tokens excluded from the retrieval descriptors. The resulting multi-query descriptors are jointly trained with a holistic representation for retrieval, without fixed spatial partitions or equal-area constraints. FM-ReID achieves strong results on animal, person, and vehicle ReID benchmarks, supporting competitive token routing as an effective way to augment holistic foundation-model representations.

Extrapolative temporal knowledge graph reasoning (TKGR) predicts future facts from historical snapshots. Most existing methods train once on an early prefix of the timeline and then use a frozen model for all future timestamps. We argue that this fixed-prefix protocol is misaligned with extrapolation. It learns from a static prefix, whereas the target stream is non-stationary: new entities and facts emerge, temporal dependencies shift across regimes, and recurring historical signals must be refreshed online. As a result, models trained only on early snapshots become outdated and degrade over long horizons. We address this mismatch by formulating extrapolative TKGR as continual learning over streaming snapshots. Under this view, effective extrapolation must jointly handle current dynamics, stable knowledge, and recurring historical evidence. Based on these requirements, we propose History-enhanced Two-Step Continual Learning (HiTS-CL), a backbone-agnostic continual learning framework for extrapolative TKGR. HiTS-CL tracks current dynamics via continual fine-tuning, preserves stable knowledge via multi-teacher adaptive distillation, and retains recurring historical evidence via a selective memory of recent and frequent facts. We integrate HiTS-CL into five representative TKGR backbones and evaluate it on four benchmark datasets. HiTS-CL consistently improves extrapolation accuracy, reduces long-horizon degradation, and outperforms strong continual-learning baselines, including a recent method for temporal knowledge graphs. Source code and data are available at https://github.com/liuyansong98/HiTS-CL.

Integrating tackiness sensation into the artificial skin of humanoid robots significantly enhances their cognitive and operational capabilities. However existing tactile sensors face challenges in decoupling of the multimodal signal and stability. Here we present a surface-soft tactile sensor that incorporates a Hall effect sensor and a soft magnetic composite within a robust elastic framework. The sensor surface indents under pressure and bulges prominently when retracted from sticky surfaces dynamically altering the Hall sensor-magnet distance. This generates whole-process-traceable and baseline-separated signals enabling real-time differentiation between pressure and pull-off force. This single-sensing-element design facilitates bimodal sensing at the same contact spot while eliminate stress cross-talk enhancing both accuracy and sensitivity. The fusion of a robust framework and magneto-mechanical sensing mechanism equips the sensor with exceptional reliability and excellent signal baseline stability. This tactile sensor holds substantial potential for advancing robotic capabilities in evaluating adhesive properties monitoring rubber aging precisely handling lightweight objects and cognizing natural objects surface characteristics.

Medical Vision-Language Models (VLMs) show significant promise for clinical image understanding, offering accurate diagnosis with interpretable reasoning. However, a critical performance gap exists between their strong vision encoders and the full multimodal model: in dermatology, the MedSigLIP encoder outperforms MedGemma by an average of 10.26 percentage points even when both use zero target-task labels; few-shot linear probing provides further evidence of strong visual representations. This gap motivates an investigation of how visual information is used in end-to-end diagnosis and why plausible-sounding predictions can lack grounding in image evidence. Using dermatology as our primary testbed, we systematically investigate three hypotheses for this phenomenon. We further provide a mechanistic analysis of the model's internal attention patterns, showing that a simple describe-then-decide prompting strategy increases vision attention by 30-40% during generation. Task-specific fine-tuning improves dermatology classification but reduces cross-domain medical question-answering performance in our evaluation. To address these challenges, we combine label-free prompting with low-label encoder-assisted reranking while keeping the VLM frozen. We validate the interventions across five VLM backbones in dermatology and provide supporting representation and attention analyses across additional medical modalities.