arXiv:2510.26745v3 Announce Type: replace
Abstract: Deep sequence models are said to store atomic facts predominantly in the form of associative memory: a brute-force lookup of co-occurring entities. We identify a dramatically different form of storage of atomic facts that we term as geometric memory. Here, the model has synthesized embeddings encoding novel global relationships between all entities, including ones that do not co-occur in training. Such storage is powerful: for instance, we show how it transforms a hard reasoning task involving an $\ell$-fold composition into an easy-to-learn $1$-step navigation task.
From this phenomenon, we extract fundamental aspects of neural embedding geometries that are hard to explain. We argue that the rise of such a geometry, as against a lookup of local associations, cannot be straightforwardly attributed to typical supervisory, architectural, or optimizational pressures. Counterintuitively, a geometry is learned even when it is more complex than the brute-force lookup.
Then, by analyzing a connection to Node2Vec, we demonstrate how the geometry stems from a spectral bias that -- in contrast to prevailing theories -- indeed arises naturally despite the lack of various pressures. This analysis also points out to practitioners a visible headroom to make Transformer memory more strongly geometric. We hope the geometric view of parametric memory encourages revisiting the default intuitions that guide researchers in areas like knowledge acquisition, capacity, discovery, and unlearning.
Science Journals
arXiv:2503.13934v2 Announce Type: replace
Abstract: Mobile robot navigation in dynamic environments with pedestrian traffic is a key challenge in the development of autonomous mobile service robots. Recently, deep reinforcement learning-based methods have been actively studied and have outperformed traditional rule-based approaches owing to their optimization capabilities. Among these methods, those that assume continuous action spaces typically rely on Gaussian distributions, which limit the flexibility of the generated actions. In contrast, the application of diffusion models to reinforcement learning has advanced, enabling more flexible action distributions than Gaussian policy-based approaches. In this study, we apply a diffusion-based reinforcement learning approach to social navigation and validate its effectiveness. Furthermore, by exploiting the characteristics of diffusion models, we propose extensions that enable adaptation to previously unseen scenarios without additional training. As concrete scenario examples, we demonstrate adaptability to scenarios in which static obstacles exist in the environment that were not present during training, as well as scenarios in which the objective differs from training, such as accompanying target pedestrians while avoiding others to reach the destination.
arXiv:2605.17732v1 Announce Type: new
Abstract: In this paper, we consider the non-Hermitian quaternion linear systems arising from color image restoration and three-dimensional signal filtering problems. For exploring to solve such systems, we present two innovative structure-preserving conjugate gradient-type methods, QNHERLQ and QNHERQR, which are based on the unitary equivalence transformations of the non-Hermitian quaternion matrices to tridiagonal forms, called quaternion Saunders-Simon-Yip tridiagonalization procedure. The proposed tridiagonalization procedure for non-Hermitian quaternion matrices is closely related to the quaternion Lanczos process for Hermitian matrices, and is very different from the quaternion Lanczos biorthogonalization process for non-Hermitian matrices. The convergence of QNHERLQ and QNHERQR is discussed, which depends on the singular values of the coefficient matrix. Also we show that both algorithms have the finite termination property and constant costs per iteration step. Numerical results illustrate that the proposed algorithms are with the robustness and effectiveness compared with QGMRES and QQMR.
arXiv:2605.17734v1 Announce Type: new
Abstract: Equipping LLM agents with reusable skills derived from past experience has become a popular and successful approach for tackling complex and long-horizon tasks. However, such lessons are often encoded as textual guidance that remains largely advisory, lacking explicit mechanisms for when and how to intervene in the agent loop. To bridge the gap, we introduce HASP(Harnessing LLM Agents with Skill Programs), a new framework that upgrades skills into executable Program Functions (PFs). Rather than offering passive advice, PFs act as executable guardrails that activate on failure-prone states and modify the next action or inject corrective context. HASP is highly modular: it can be applied at inference time for direct agent-loop intervention, during post-training to provide structured supervision, or for self-improvement by evolving validated, teacher-reviewed PFs. Empirically, HASP drives substantial gains compared to both training-free and training-based methods on web-search, math reasoning, and coding tasks. For example, on web-search reasoning, inference-time PFs alone improve the average performance by 25% compared to (multi-loop) ReAct Agent, while post-training and controlled evolution achieve a 30.4% gain over Search-R1. To provide deeper insights into HASP, our mechanism analysis reveals how PFs trigger and intervene, how skills are internalized, and the requirement for stable skill library evolution.
arXiv:2509.17666v2 Announce Type: replace
Abstract: Object insertion tasks are prone to failure under pose uncertainty and environmental variation, often requiring manual fine-tuning or controller retraining. We present a novel approach for robust and resilient object insertion using a passively compliant soft wrist that enables safe contact absorption through large deformations, without high-frequency control or force sensing. Our method structures insertion as compliance-enabled contact formations, sequential contact states that progressively constrain degrees of freedom, and integrates automated failure recovery strategies. Our key insight is that wrist compliance permits safe, repeated recovery attempts; hence, we refer to it as compliance-enabled failure recovery. We employ a pre-trained vision-language model (VLM) that assesses each skill execution from terminal poses and images, identifies failure modes, and proposes recovery actions by selecting skills and updating goals. In simulation, our method achieved an 83% success rate, recovering from failures induced by randomized conditions, including grasp misalignments up to 5 degrees, hole-pose errors up to 20 mm, fivefold increases in friction, and unseen square/rectangular pegs, and we further validated the approach on a real robot. Project page is available at https://omron-sinicx.github.io/compliance-enabled-failure-recovery/.
arXiv:2510.06809v3 Announce Type: replace
Abstract: Echocardiography is a critical tool for detecting heart diseases, yet its steep operational difficulty causes a shortage of skilled personnel. Probe guidance systems, which assist in acquiring high-quality images, offer a promising solution to lower this operational barrier. However, robust probe guidance remains challenging due to significant individual variability. This variability manifests as differences in low-level features within two-dimensional (2D) images, which complicates image feature understanding, and differences in individual three-dimensional (3D) structures, which poses challenges for precise navigation. To address these challenges, we first propose leveraging the robust image representations learned by ultrasound foundation models from vast datasets. Yet, applying these models to probe navigation is non-trivial due to their lack of understanding of individual 3D structures. To this end, we meticulously design a Vision-Action Adapter (VA-Adapter) to online inject the capability of understanding individual 3D structures. Specifically, by embedding the VA-Adapter into the foundation model's image encoder, the model can infer cardiac anatomy from historical vision-action sequences, mimicking the cognitive process of a sonographer. Extensive experiments on a dataset with over 1.31M samples demonstrate that the VA-Adapter outperforms strong probe guidance models while requiring approximately 33 times fewer trained parameters. Code is available at https://github.com/LeapLabTHU/VA-Adapter.
arXiv:2502.20969v4 Announce Type: replace
Abstract: Retrieval-augmented generation (RAG) extends large language models (LLMs) with external data sources to enhance factual correctness and domain coverage. Modern RAG pipelines rely on large datastores, creating a significant system challenge: achieving high throughput and low latency is difficult, especially when GPU memory is limited. To address these challenges, we propose TeleRAG, an efficient inference system that reduces latency and improves throughput with minimal GPU memory requirements. The core innovation of TeleRAG is lookahead retrieval, a prefetching mechanism that predicts required data and transfers them from CPU to GPU in parallel with LLM generation. In addition, TeleRAG adopts a prefetching scheduler and a cache-aware scheduler to support efficient multi-GPU inference with minimal overhead. Evaluations show TeleRAG achieves up to a 1.53x average end-to-end latency reduction (single-query) and 1.83x higher average throughput (batched), as well as good scalability in throughput. This confirms the practical utility of TeleRAG for faster and more memory-efficient deployments of RAG applications.
arXiv:2505.03551v5 Announce Type: replace-cross
Abstract: Here we argue that spinor structure arises naturally if relativistic statistical mechanics is formulated directly on phase spacetime. Requiring a first-order phase-spacetime description that retains both mass-shell branches leads to a Clifford factorisation of the relativistic constraint and hence to a $4\times4$ spinor-matrix distribution function. We show that deformation quantisation leads to a phase-space formulation of spin quantum mechanics. We argue that projection onto positive- and negative-energy sectors recovers the standard relativistic classical transport equations in the appropriate scalar limits, while the corresponding left- and right- stargenvalue equations reproduce the constraint structure of the Dirac-Wigner formulation. The result is a phase-space route from relativistic statistical mechanics to spinor quantum mechanics, in which spin algebra emerges as the internal structure required by any relativistic statistical theory containing both mass-shell branches and the dimensions of angular momentum from quantum non-locality.
arXiv:2605.17650v1 Announce Type: new
Abstract: In the framework of Smart Cities and Intelligent Transportation Systems (ITS), efficient parking management is essential to reduce urban congestion and emissions. However, current reservation-based systems often encounter a scenario in which users find their reserved slot occupied by a previous occupant who failed to vacate on time ("No PARK" situation). This paper introduces a dual-mechanism architecture designed to enhance system reliability. A Reservation Module uses a dynamic size buffer of non-reservable slots to grant parking availability. A reputation-based Reward System exploits a "star-based" metric to incentivize punctual departures through financial penalties and access restrictions. The simulations conducted with the SUMO urban simulator are promising, showing that the dynamic buffer strategy provides a better tradeoff between parking availability and reservation success. By progressively adapting to users behavior, the proposed system mitigates "NO PARK" instances and improves resource utilization, significantly enhancing urban viability. Index Terms-Smart City, Intelligent transportation systems, Parking, Reservation systems, V2I, Reputation-based mechanisms, Smart Parking
arXiv:2502.13957v3 Announce Type: replace
Abstract: Large language models (LLMs) are transforming web search by shifting from document ranking to synthesizing answers, and are increasingly deployed as autonomous agentic search systems that iteratively interact with external knowledge sources. Despite this progress, building effective search agents remains challenging because high-quality intermediate search steps are difficult to generate. Previous approaches have primarily relied on outcome supervision, rewarding agents only for producing correct final answers. This often leads to reward hacking and excessive dependence on parametric memory, limiting generalization to out-of-domain tasks. To address these limitations, we introduce RAG-Gym, a framework that shifts supervision from final answers to the search process itself. With RAG-Gym, we systematically investigate architecture design, parameter optimization, and action evaluation, identifying reasoning reflection as a critical capability for search agents. Building on this insight, we propose Re$^2$Search++, a process-supervised agent that achieves substantial improvements on multi-hop information-seeking benchmarks, especially in out-of-domain settings. Performance gains are driven primarily by higher-quality search queries rather than answer optimization alone, and the learned search critics transfer across models, including proprietary LLMs. These findings show that supervising the search process produces more reliable and generalizable information-seeking agents.
arXiv:2510.12008v3 Announce Type: replace-cross
Abstract: APN functions play a central role as building blocks in the design of many block ciphers, serving as optimal functions to resist differential attacks. One of the most important properties of APN functions is their linearity, which is directly related to the Walsh spectrum of the function. In this paper, we establish two novel connections that allow us to derive strong conditions on the Walsh spectra of quadratic APN functions. We prove that the Walsh transform of a quadratic APN function $F$ operating on $n=2k$ bits is uniquely associated with a vector space partition of $\mathbb{F}_2^n$ and a specific blocking set in the corresponding projective space $PG(n-1,2)$. These connections allow us to prove a variety of results on the Walsh spectrum of $F$. We prove for instance that $F$ can have at most one component function of amplitude larger than $2^{3n/4}$. We also find the first nontrivial upper bound on the number of bent component functions of a quadratic APN function, and provide conditions for a function to be CCZ-equivalent to a permutation based on its number of bent components.
arXiv:2504.07347v3 Announce Type: replace-cross
Abstract: As demand for Large Language Models (LLMs) and AI agents grows rapidly, optimizing systems for efficient LLM inference becomes critical. While significant efforts have targeted system-level engineering, little has been explored from a mathematical modeling and queueing perspective. In this paper, we develop the queueing fundamentals for LLM inference. In particular, we study the throughput aspect of LLM inference systems. We prove that a large class of `work-conserving' scheduling algorithms achieve maximum throughput for both individual requests and AI-agent workloads with directed acyclic graph (DAG) and fork-join routing topologies, establishing `work-conserving' as a key design principle for practitioners. Technically, we develop a fluid-limit framework for multi-class batched processing networks under $K$-FCFS scheduling, which may be of independent interest. Evaluations of real-world systems confirm that Orca and Sarathi-Serve are throughput-optimal, reassuring practitioners, while FasterTransformer and vanilla vLLM are not maximally stable and should be used with caution. Our analysis also reveals how constraints such as batch size limits and cyclic routing topologies complicate the throughput picture, pointing to rich open questions at the intersection of queueing theory and LLM system design.
arXiv:2509.05383v2 Announce Type: replace-cross
Abstract: Models that describe Newtonian liquid films evolving due to the competing effects of surface tension and attractive intermolecular or van der Waals forces are known to rupture in finite time in a self-similar manner. We extend the computation of similarity solutions to non-Newtonian power-law liquid films. The resulting bifurcation diagram, indexed by power law exponent $n$, has a highly nontrivial structure with branches merging via a snaking bifurcation around $n=1$. A countably infinite number of solutions are also found in the extreme shear-thinning ($n\to 0$) limit, in which similarity solutions possess an exponentially small inner region. Numerical simulations of the time-dependent model are shown to be attracted to the single primary branch of similarity solutions. The asymptotic structure of solutions in the small $n$ limit is also determined.
arXiv:2605.15735v2 Announce Type: replace
Abstract: Vision--language--action (VLA) models are typically built by fine-tuning a pretrained vision--language model (VLM) on action data. However, we show that this standard recipe systematically erodes the VLM's multimodal competence, a side effect we call the embodiment tax. But do VLAs have to forget? Inspired by the two-stream organization of biological vision, we trace this degradation to a structural bottleneck: current VLAs ask a single encoder to support both language-grounded semantics and control-relevant visual features, whereas biological vision separates recognition and visuomotor control into distinct pathways. Building on this view, we propose the Unified Action Model (UAM), which adds a parallel Dorsal Expert, an analog of the brain's dorsal pathway. To make the Dorsal Expert an effective second pathway and reduce the control-learning burden on the VLM, we initialize it from a pretrained generative model and train it with a mid-level reasoning objective that predicts visual dynamics. This design allows us to train the whole VLA end-to-end on action data alone: with no parameter freezing, no gradient stopping, and no auxiliary VL co-training, UAM retains over $95\%$ of the underlying VLM's multimodal capability and at the same time achieves the highest average success rate among baselines on a variety of manipulation tasks that probe out-of-distribution generalization, including unseen objects, novel object--target compositions, and instruction variation. Together, these results suggest that semantic preservation in VLAs can emerge from architectural separation itself, rather than being enforced by frozen weights or auxiliary data replay, and that this preserved semantic capability can naturally transfer from VLMs to semantic generalization in actions.
arXiv:2605.17737v1 Announce Type: new
Abstract: The rapid advancement of generative AI has made audio deepfakes increasingly indistinguishable from authentic human vocals, posing significant threats to persons-of-interest (POI) such as public figures. Current detection systems primarily rely on generic, black-box models that fail to capture speaker-specific idiosyncratic traits and lack interpretability. In this paper, we propose Phoneme-based Voice Profiling (PVP), a novel personalized defense framework. By shifting the detection paradigm from macro-utterance analysis to micro-phonetic modeling, PVP captures the unique acoustic distributions underlying a POI's habitual articulatory patterns. Specifically, our framework models speaker-specific phonetic realizations using lightweight Gaussian Mixture Models (GMMs) estimated solely from bona fide reference speech. This design enables data-efficient profiling and robust generalization to previously unseen spoofing attacks without requiring heavy spoof-specific training. Furthermore, we introduce the first large-scale Chinese POI deepfake dataset to benchmark speaker-specific detection. Experimental results demonstrate that PVP significantly outperforms state-of-the-art generic detectors in POI spoofing scenarios, achieving substantial EER reductions while providing fine-grained, phoneme-level interpretability for forensic analysis. Code and data are available at: https://github.com/JunXue-tech/PVP
arXiv:2605.17697v1 Announce Type: new
Abstract: Cities are increasingly turning to large-scale data analysis and machine learning to make consequential decisions. While the algorithmic fairness community has focused on analyzing the risks and benefits associated with these complex methods, there has been much less scrutiny of the many simpler, but still widely used, data-driven tools that support government decision-making in a variety of settings. In this work, we study hand-crafted indices for geographic targeting and decision-making in emergency management -- a field responsible for coordinating preparedness and response efforts to hazards ranging from natural disasters to human threats. Indices, which capture abstract principles and overarching priorities (e.g., reducing social vulnerability), are low-complexity models that statistically aggregate chosen variables. They are generally flexible and interpretable, but can also be sensitive to key design choices and require strong assumptions. Through a case study of decision-making for extreme heat emergencies in NYC, we examine the challenges that practitioners may face in selecting an index for preparedness and response actions. We map empirical findings from index-based simulations to concerns related to validity and reliability from the measurement literature and show via sensitivity analyses that different reasonable choices of input variables or spatial scale can result in substantive differences to index risk scores, thereby affecting downstream government decision-making. We contrast these challenges with considerations for developing predictive algorithms that more narrowly relate to concrete, measurable outcomes. Ultimately, we provide generalizable recommendations that practitioners and public-sector technologists can use for navigating the trade-offs between indices and predictive algorithms in other government settings.
arXiv:2605.17739v1 Announce Type: new
Abstract: We study hyperinterpolation and its spectral multiplier variants on the sphere under weak cubature assumptions formulated through Sobolev discrepancy estimates. In contrast with classical hyperinterpolation theory, our framework does not require exact polynomial cubature formulas or Marcinkiewicz--Zygmund inequalities. The main idea is to interpret the discretization error as the action of a spectral multiplier operator on the cubature discrepancy measure. This viewpoint separates approximation properties of the underlying spectral operator from geometric properties of the sampling measure, leading to stable Sobolev approximation estimates under weak cubature assumptions. The resulting theory applies to a broad class of spectral approximation operators, including sharp spectral projections, compactly supported smooth filters, Bessel potential operators, and heat kernel operators. For sufficiently localized spectral multipliers, we additionally obtain uniform $L^\infty$-stability of the corresponding discrete approximation operators. The results establish a direct connection between hyperinterpolation, Sobolev discrepancy, and quasi-Monte Carlo (QMC) designs, showing that stable approximation from scattered data can be achieved without exact polynomial reproduction.
arXiv:2605.17818v1 Announce Type: new
Abstract: Open-set recognition systems face a neglected failure mode: high-confidence near-known unknowns, which lie outside the known label set but are close enough to known classes that a closed-set classifier accepts them with high confidence. We show that this failure is widespread across scalar-threshold methods, including recent post-hoc detectors, and that stronger encoders can amplify rather than remove the risk. We propose EGUR-A, which changes the decision from ``is this sample's score high enough?'' to ``does this predicted known class have sufficient evidence to accept this sample?'' EGUR-A combines class-conditional local acceptance evidence with global residual evidence, and selects their relative weight from known-sample statistics without unknown validation data. Across CUB, FGVC-Aircraft, and ImageNet-hard, EGUR-A substantially reduces high-confidence false known acceptance at matched known-rejection operating points. The result is not a stronger threshold; it is a different question: whether a known class is entitled to accept a sample.
arXiv:2605.17820v1 Announce Type: new
Abstract: We propose an adaptive Hermite spectral method for the Vlasov-Poisson system based on a recently developed frequency indicator that measures the contribution of the high-order expansion coefficients. Precisely, the symmetrically weighted Hermite basis with a scaling factor is utilized to approximate the distribution function to satisfy the increasing resolution requirement, which, for example, is induced by filamentation. To implement the scaling adjustment, a fast conservative projection operator is constructed in two steps. The first step is to formulate the projection as a constrained optimization problem to preserve key invariants, including mass, momentum, energy, and the $L^2$ norm of the distribution function. The second step is an ODE-based approximation developed to compute the updated expansion coefficients with linear complexity. Numerical experiments with 1D1V and 2D2V settings validate the feasibility and efficiency of this proposed adaptive Hermite method.
arXiv:2605.13322v2 Announce Type: replace
Abstract: Kamon (family crests) are an important part of Japanese culture and a natural test case for compositional visual recognition: each crest combines a small number of symbolic choices, but the space of possible descriptions is sparse. We introduce KamonBench, a grammar-based image-to-structure benchmark with 20,000 synthetic composite crests and auxiliary component examples. Each composite crest is paired with a formal kamon description language - "kamon y\=ogo" - description, a segmented Japanese analysis, an English translation, and a non-linguistic program code. Because each synthetic crest is generated from known factors, namely container, modifier, and motif, KamonBench supports evaluation beyond caption-level accuracy: direct program-code factor metrics, controlled factor-pair recombination splits, counterfactual motif-sensitivity groups under fixed container-modifier contexts, and linear probes of factor accessibility. We include baseline results for a ViT encoder/Transformer decoder and two VGG n-gram decoders, with and without learned positional masks. KamonBench therefore provides a controlled testbed for sparse compositional visual recognition and factor recovery in vision-language models.
arXiv:2509.17680v2 Announce Type: replace
Abstract: Table question answering (TableQA) is a fundamental task in natural language processing (NLP). The strong reasoning capabilities of large language models (LLMs) have brought significant advances in this field. However, as real-world applications involve increasingly complex questions and larger tables, substantial noisy data is introduced, which severely degrades reasoning performance. To address this challenge, we focus on improving two core capabilities: Relevance Filtering, which identifies and retains information truly relevant to reasoning, and Table Pruning, which reduces table size while preserving essential content. Based on these principles, we propose EnoTab, a dual denoising framework for complex questions and large-scale tables. Specifically, we first perform Evidence-based Question Denoising by decomposing the question into minimal semantic units and filtering out those irrelevant to answer reasoning based on consistency and usability criteria. Then, we propose Evidence Tree-guided Table Denoising, which constructs an explicit and transparent table pruning path to remove irrelevant data step by step. At each pruning step, we observe the intermediate state of the table and apply a post-order node rollback mechanism to handle abnormal table states, ultimately producing a highly reliable sub-table for final answer reasoning. Finally, extensive experiments show that EnoTab achieves outstanding performance on TableQA tasks with complex questions and large-scale tables, confirming its effectiveness.
arXiv:2605.16836v1 Announce Type: cross
Abstract: Hypergraphs provide a principled framework for modeling polyadic interactions, with applications in recommendation systems, social networks, and molecular modeling. Hypergraph generation remains challenging because incidence structures are discrete, sparse, and governed by heterogeneous higher-order interactions. Existing generators often rely on implicit latent spaces or continuous incidence decoders, which provide limited mechanistic interpretation of how node-hyperedge incidences arise. To address these limitations, we propose HYVINT, an intensity-driven hypergraph generative framework. Our key innovations are twofold: (i) we develop an intensity-driven incidence formation mechanism for hypergraphs that links latent interaction strength to binary incidence, and (ii) we derive a tractable lower-bound variational estimator for learning latent representations. We provide generation error bounds with asymptotic convergence rates and empirically show that HYVINT achieves strong fidelity while maintaining substantial novelty and diversity on synthetic and real-world hypergraphs.
arXiv:2605.17718v1 Announce Type: cross
Abstract: Feature learning is widely regarded as the key mechanism distinguishing neural networks from fixed-kernel methods, yet its impact on the induced function space remains poorly understood. In this work, we precisely characterize how the function space spanned by the features of a two-layer neural network evolves during gradient descent training. We prove that, in the high-dimensional proportional regime, after a large gradient step the post-update feature distribution is well approximated by a target-dependent spiked Gaussian covariance. This induces a data-adaptive kernel that reshapes the function space and modifies its spectral structure. Our analysis reveals that feature learning can be interpreted as a distributional transformation in either parameter space or input space, equivalently as the introduction of a target-dependent kernel. In particular, it selectively amplifies eigenvalues aligned with the target direction and mixes leading eigenfunctions, coupling the top radial mode with a target-aligned quadratic harmonic. Overall, our results provide a precise function-space perspective on early-stage feature learning: rather than just rescaling a fixed kernel, gradient descent induces a data-adaptive deformation that preferentially enhances directions aligned with the signal in the data.
arXiv:2605.17743v1 Announce Type: new
Abstract: Continual test-time adaptation adapts a source-pretrained model to non-stationary, unlabeled target streams while retaining past competence, yet texture-biased backbones risk error accumulation and catastrophic forgetting. Drawing inspiration from the process of decoupling shape and texture in the human visual system, we introduce MoASE, a plug-in mixture-of-experts that disentangles domain-agnostic structure from domain-specific texture using Activation Sparsity Experts with Spatial Differentiable Dropout, forming complementary high- and low-activation pathways, while high- and low-rank bottlenecks diversify representations. The Activation Sparsity Gate produces input-adaptive SDD thresholds for precise token selection, and the Domain-Aware Router assigns per-sample expert weights using texture-sensitive cues. To curb confirmation bias on unlabeled streams and stabilize supervision, we then introduce Domain-Adaptive On-Policy Distillation to constitute MoASE++, with an EMA-anchored on-policy reverse KL distillation and an augmentation policy conditioned on entropy and confidence that aligns predictions across the same views and improves the robustness-plasticity balance. Extensive experiments on classification (CIFAR-10/100-C, ImageNet-C) and semantic segmentation (Cityscapes->ACDC) demonstrate consistent state-of-the-art performance, offering a principled, controllable approach to continual adaptation in dynamic visual environments.
arXiv:2605.17709v1 Announce Type: cross
Abstract: We develop and analyze mathematical models for residential burglary that incorporates police deployment through a delayed feedback mechanism. Motivated by empirical observations from publicly available crime and policing data, we extend a well-known agent-based model by introducing a dynamic police response driven by crime information that becomes available only after a finite delay. Taking the mean-field limit, we derive a coupled continuum system consisting of three partial differential equations and one ordinary differential equation describing the interactions among criminal density, environmental attractiveness, delayed crime signal, and police deployment. Linear stability analysis of homogeneous steady states reveals that response delays can destabilize otherwise stable equilibria through Hopf bifurcations. As a result, the model predicts sustained temporal oscillations and dynamically evolving crime hotspots. Numerical simulations of both the agent-based and continuum models confirm the theoretical analysis and uncover rich spatio-temporal behaviors, including moving, splitting, and merging hotspots. Through a parametric study, we investigate the roles of police density, crime information delay, and neighborhood effects in controlling stability, hotspot size, and oscillatory behavior. Our results indicate that timely access to crime data plays a more important role than police density in stabilizing crime levels.