Computer Vision
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
Sunday 16 August
Conditional image-to-image generators are single-shot: they map input features to an output in one forward pass and treat it as final, with no opportunity to improve on it. Although trained to produce the best possible result in one step, such a model leaves room for improvement if it can adaptively revise its own output over iterations. We propose Recursive Refinement via Feedback Conditioning (RRFC), a novel feedback-conditioning framework for iterative output refinement that teaches a model to adaptively revise its output by conditioning on a new signal, namely its most recent previous prediction, which is fed back as an auxiliary set of channels alongside the original input. This preserves the generator's core architecture while modifying its conditioning interface and, depending on the model family, its training or inference procedure, so RRFC can be attached to existing generators without redesign. We evaluate RRFC across six baselines spanning adversarial, equilibrium, and diffusion-based models and three paired image-to-image translation tasks. Across 18 architecture-task settings, RRFC yields seven Holm-corrected improvements, seven degradations, and four non-significant changes. The gains concentrate on reconstruction-fidelity and identity settings, while five of the seven degradations fall on the single semantic-layout task, where every model declines. These results indicate that feedback-based refinement helps when its objective overlaps with the evaluated property, and that its gains concentrate on the tasks where that overlap holds.
Artificial Intelligence
Large Models for Small Devices: Recent Advances and Empirical Analysis of Edge AI Deployment
Running large AI models on resource-constrained edge devices requires model compression to reduce model size and computation. What compresses well, however, need not deploy well. We survey dozens of recent works that report compression results on real hardware and extract practical deployment guidelines from them. Following these guidelines, we deploy compact language and image models on GPU, CPU, and Raspberry Pi platforms across question answering and image segmentation. No single technique wins across tasks. For question answering, Qwen3.5 0.8B reaches 93.85 SQuAD F1 and 92 EM under Q5_K_M GGUF quantization, while structured pruning at the same precision costs 16 F1 at a 1% ratio. For segmentation, the ranking reverses: default quantization leaves parameters and MACs unchanged, whereas pruning cuts model size by nearly 80% at near-constant mIoU. Pruning can even inflate the deployed artifact by 21-49% by breaking k-quant super-block alignment; combined with longer, less format-compliant outputs, this raises Raspberry Pi latency up to 3.4x. Compression can also manufacture the appearance of competence rather than destroy it visibly: one LoRA-recovered variant stays fully parseable and holds 71% strict BoolQ accuracy while sending 97 of 100 predictions to a single class, at 52.6% balanced accuracy. We explain these effects through neural-flow graph analysis and prefill-decode-level latency decomposition, and condense them into task-specific deployment research directions. The right technique depends on the task, the model, and the hardware. Our experiment code and artifacts are open-sourced at https://github.com/Arnavvvkumar/deployment
Read the full abstract
Computer Vision
Automated Fetal Brain MRI Biometry in Healthy and Pathological Cases
Automated biometric analysis of fetal brain MRI enables reproducible, observer-independent quantitative assessment, yet existing methods are often restricted to few measurements or evaluated only on healthy cases. We assemble and evaluate an automated biometric analysis pipeline that localizes 22 anatomical landmarks on NeSVoR-reconstructed 3D volumes and derives 11 clinically relevant measurements spanning supratentorial, ventricular, cerebellar, and midline structures. We compare two landmark localization models, H3DE-Net and SCN, on a heterogeneous cohort of 122 acquisitions (both healthy controls and range pathologies). Localization accuracy was assessed with a linear mixed-effects model, agreement with normative growth trajectories with calibrated centile charts, and diagnostic utility with a decision tree classifying VM severity. H3DE-Net achieved significantly lower localization error than SCN across all landmarks (mean 1.36 mm vs. 3.58 mm in HC and 1.90 mm vs. 4.13 mm in PC; p < 0.001), and outperformed a GA-based regression baseline in 7 of 11 measurements. H3DE-Net measurements yielded higher classification AUC in every diagnostic group, with the clearest advantage in separating healthy controls from VM. Decision tree thresholds for ventricular width fell near the clinical 10 mm and 15 mm cut-offs used to define and grade VM.
Read the full abstract
Computation and Language
BERTopic-Virality Prioritisation: A Scalable Framework for Thematic and Comparative Analysis of COVID-19 and Monkeypox Misinformation on Twitter
Health misinformation circulating during pandemics can gain traction rapidly, creating harmful narratives that compete with public health guidance. Most topic-modelling pipelines treat engagement as an external outcome, limiting their ability to prioritise semantically coherent topics that are also rapidly diffusing. We introduce BERTopic-VP, a virality-prioritised topic-modelling framework that combines contextual embedding-based clustering (BERTopic) with a post hoc Virality Prioritisation (VP) layer. The pipeline is complemented by a two-stage hybrid misinformation detection module that fuses a supervised content-based classifier with an external verification signal derived from public-health knowledge bases. Applied to three benchmark datasets, COVID-19_FNIR, Monkeypox, and Constraint, the framework achieves strong classification performance, with F1 up to 0.950 and ROC-AUC up to 0.989, while identifying high-impact clusters under top 1%, 5%, and 10% VP thresholds. For datasets without native engagement metadata, prioritisation is based on a logistic propensity-to-spread score, used as an ordinal proxy for diffusion potential rather than a direct measure of engagement. The results show that integrating semantic structure, virality-aware ranking, and affective-linguistic profiling enables scalable and interpretable comparative analysis of misinformation across pandemics. The proposed framework supports monitoring-oriented early warning by surfacing low-volume but high-risk narratives for analyst review.
Read the full abstract
Sound
Adding Voice Cloning to Text-to-Audio-Video Models with a Single Zero-Initialised Layer
Text-to-audio-video (T2AV) generation models produce a video and its soundtrack from a textual description, but offer no control over whose voice speaks in the output. We show that a base T2AV model can be turned into a voice-cloning model by adding a single zero-initialized linear layer on top of its audio backbone, fine-tuning for a comparatively short training schedule, and conditioning on a short reference recording at inference time. The reference is injected through two complementary signals: its diffusion latents are prepended to the audio stream, and a global speaker embedding modulates token of the target audio. On a benchmark of 674 speaker-text pairs spanning 30 speakers we compare against five strong voice-cloning text-to-speech baselines: our enhanced 5B model attains the highest speaker-encoder cosine similarity (SECS) across three independent verification networks (ECAPA-TDNN, WavLM-SV, Resemblyzer), statistically significantly outperforming every baseline. A side product of the architecture is that the audio path can be evaluated without the video path at inference time, yielding a ~30x speed-up over the full audio-video diffusion loop while preserving the voice-cloning behaviour.
Read the full abstract
Social and Information Networks
Integrating Persuasion Theory into the Epidemiological Modelling of Health Misinformation Spread on Social Media
This study presents a hybrid epidemiological and behavioural framework to simulate the spread of health misinformation on social media. We extend the classical Susceptible–Infected–Recovered (SIR) model to a six-compartment structure (SIRMMM), incorporating Misinformed Susceptible (MS), Misinformed Infected (MI), and Misinformed Recovered (MR) compartments to better reflect the dynamics of the misinformation lifecycle. To account for individual-level behavioural variation, we extend the SIRMMM model by integrating psychological signals from the Elaboration Likelihood Model (ELM), including sentiment polarity, engagement metrics, and cognitive effort, which dynamically modulate the misinformation transmission rate, yielding the ELM-SIRMMM framework. Model parameters were estimated using the FibVID dataset, which captures COVID-19 misinformation on Twitter. Generalisability was tested on two additional datasets: MC-Fake (emotional misinformation) and Monant (general health misinformation). Results show that the ELM-SIRMMM model enhances both predictive accuracy and dynamic realism. On FibVID, it decreases RMSE by 5.5%, delays the misinformation peak from day 150 to day 160, and increases its peak prevalence from 6% to 7%. On MC-Fake, it accurately reproduces a flash-rumour pattern, infecting 38% of users by day 45 and achieving 97% misinformation recovery, all while maintaining model accuracy. In contrast, minimal behavioural signal variability in the Monant dataset leads to marginal benefit, with only a 3% peak and 57% of users remaining susceptible. These findings suggest that structural elaboration alone is insufficient. Functional realism in modelling misinformation spread requires dynamic psychological inputs that vary meaningfully across time and contexts.
Read the full abstract
Computer Vision
Training-Free Long-Term Multi-Object Tracking for Sports Video Analytics
Long-term multi-object tracking in sports remains challenging due to frequent occlusions, rapid camera motion, and repeated player reappearances. We introduce McByte++, a training-free tracking-by-detection framework that integrates lightweight mask propagation, conditional camera motion compensation, and online re-identification within a unified pipeline. Compared to its predecessor, McByte++ substantially improves runtime efficiency while enhancing identity preservation. On SoccerNet-tracking and SportsMOT benchmarks, McByte++ achieves up to +3.0 HOTA and +6.1 IDF1 improvements over the original McByte in the online setting, with further gains when combined with offline global association. Replacing heavy segmentation components and optimizing motion modeling yields up to an order-of-magnitude speed increase. All results are obtained without detector retraining or dataset-specific tuning. Code will be made available at https://github.com/tstanczyk95/McBytePlusPlus.
Read the full abstract
Artificial Intelligence
THESIS-MoE: Trainable Hierarchical Extraction and SteerIng of Sycophancy in Mixture-of-Experts
Sycophancy, the tendency of a language model to change its answer to match a user's stated belief, is a common alignment failure. Existing activation steering methods typically apply a single contrastive direction uniformly throughout the model, which is an unconditional intervention that alters activations even when no sycophantic behavior is present, trading knowledge retention for behavioral correction. In Mixture-of-Experts (MoE) models, prior work further suggests that behavior is encoded within expert computations rather than routing decisions alone, making precise behavioral steering particularly challenging. In this work, we introduce a shared contrastive signal, built from matched prompts with and without a stated belief, that identifies where sycophancy lives across the MoE hierarchy and drives interventions that act only where the behavior is present. We formulate localization as a causal search over a granularity ladder of MoE blocks, experts, attention blocks, and heads, and compare unconditional subtraction against two conditional alternatives: an analytic projection-based subtraction and a learned per-token gate that steers the model away from sycophancy while keeping its weights frozen. We evaluate on three MoE models measuring sycophancy alongside general knowledge and reasoning benchmarks. Our conditional interventions removed up to 90% of the belief-induced sycophancy. Our results demonstrate that sycophancy resides in identifiable computational subcircuits and can be selectively steered while maintaining a favorable removal-retention trade-off.
Read the full abstract
Computer Vision
Counterfactual Sensitivity Is Not Repairability: Auditing Replay Probes for Video Evidence
Tool-using video agents retrieve visual evidence before answering, but the final answer is not forced to depend on what was retrieved. The natural black box test is counterfactual: destroy the semantic content of the frames the agent retrieved and check whether the answer changes, against a matched sham that re-executes the identical pipeline on those same frames. We introduce CARVE, a black-box counterfactual probe that compares answer changes under matched SHAM and DESTROY replays. Across three independent k=3 runs on a frozen VideoExplorer-style agent, DESTROY changes the answer 29.3 percentage points more often than SHAM, yielding a large and reproducible aggregate effect. Question-level scores are less stable, and increasing the replay budget from k=3 to k=10 reduces ties but weakens the original zero-threshold routing policy. At k=3, CARVE selects 538 of 1,258 LVBench questions and improves accuracy by 3.26 points, with higher fallback yield than most matched random subsets. The score shows only a weak association with annotated temporal coverage, so CARVE is best understood as a routing signal rather than a direct grounding classifier. Our implementation is available at https://github.com/KurbanIntelligenceLab/CARVE.
Read the full abstract
Numerical Analysis
A Homological Decomposition for the Dimension and Dimensional Stability of Polynomial Spline Spaces over T-Meshes
We study the dimension and dimensional stability of polynomial spline spaces of bi-degree $(m,m')$ with prescribed smoothness orders over planar T-meshes. The homological dimension formula writes the spline dimension as the sum of an Euler characteristic term and a correction term. For a fixed ordered bi-degree, the Euler characteristic term is determined by the mesh structure and the prescribed smoothness orders. The correction term can be written as a quotient of coefficient spaces attached to maximal interior segments (MISs). We prove a weighted deletion theorem: when the available vertex relations generate the coefficient space of an MIS, its summand can be removed from the quotient without changing the correction term. This operation changes neither the T-mesh nor its chain complexes. Repeating the deletion leaves a weighted completely non-diagonalizable component (CNDC). We prove that the weighted CNDC is independent of the order in which eligible MISs are removed and is the same for corresponding pairs in the structural class. The correction term can therefore be represented using only the MISs in the weighted CNDC, while all relations from the original T-mesh are retained. Dimensional stability is then equivalent to constancy of the dimension of the remaining relation space. In particular, an empty weighted CNDC is sufficient for stability. We also derive an upper bound for the remaining correction term and compare it with Mourrain's upper bound based on all MISs.
Read the full abstract
Computer Vision
BASeg: Boundary-Aware Remote Sensing Segmentation with Structural Penalties
Semantic segmentation is a core computer vision task in the remote sensing field, accelerating advancements in ur- ban development, agriculture, ecology, water resources, and environmental monitoring. However, recent methods usually struggle to capture fine-grained object features and bound- ary details. Besides, current widely used datasets often lack city morphology diversity and segmentation on generative im- ages remains largely unexplored. To address these issues, we propose a Mahalanobis-Angle Boundary Loss (MABL) that explicitly enhances boundary and shape consistency. MABL jointly models structural importance and boundary orientation through Mahalanobis distance-based weighting and angle- aware penalty. It can be readily integrated into diverse seg- mentation architectures and consistently improves their accu- racy. Built upon MABL, we introduce BASeg, a boundary- aware remote sensing segmentation framework with Struc- tural Penalties. BASeg integrates a Global Visual State Space module (GSM) with a Cross-Feature Fusion module (CFM) to capture both long-range contextual dependencies and fine- grained local details. Additionally, we establish a global 10- city benchmark dataset (GCD-25k) to facilitate accurate build- ing and road segmentation. Extensive experiments on four remote-sensing benchmarks demonstrate that BASeg consis- tently outperforms existing methods, achieving up to a 2.8% improvement in mIoU while producing more accurate object boundary segmentation across diverse scenes. Moreover, integrating MABL into multiple existing segmentation archi- tectures consistently improves performance across datasets, demonstrating its robustness and broad applicability.
Read the full abstract
Robotics
Robo-Dopamine 2.0: History-Conditioned and OOD-Aware Process Reward Modeling for Robotic Manipulation
Vision-language-action (VLA) models improve robotic manipulation but remain vulnerable to compounding errors, scene changes, and off-trajectory states. Reinforcement learning can refine pretrained VLA policies, yet sparse success signals hinder exploration, while engineered dense rewards are costly and task-specific. Existing learned visual reward models often rely on static before-after observations, causing temporal ambiguity and weak discrimination between robustness-preserving variations and task-invalid failures under out-of-distribution (OOD) execution. We introduce Robo-Dopamine 2.0, a history- and OOD-aware process reward model with a pairwise prediction interface. It combines (1) history-conditioned pairwise rewards that use source-aligned reference panels for synthetic OOD queries and observed rollout history for online queries, while preserving the queried endpoints, and (2) an OOD-aware signed progress space that represents valid progress, robustness, failure, and recovery. A Signed-Hop Curriculum with transition-aware replay learns coarse execution ordering before fine-grained progress calibration. We also construct an OOD trajectory dataset and a five-family benchmark. Reference panels improve mean visual order consistency (VOC) from 0.967 to 0.986 and OOD-robust VOC from 0.906 to 0.958. With the same 400K pairwise-reward budget, Signed-Hop training with 25% replay reaches 0.9872 mean VOC, compared with 0.9858 for a matched-pool shuffled control. In downstream reinforcement learning, the full model achieves 86.8% mean RoboTwin success and 71/80 successful real-world insertions.
Read the full abstract
Software Engineering
Where Accountability Lives: Mapping Human Responsibility to Workflow Artifacts in Agentic Software Development
Coding agents author commits, open pull requests and push code in production repositories. Responsibility is settled in two layers that do not refer to each other: platform controls gating what an agent may do, and provider terms allocating responsibility for its output. Objective: Where the two layers disagree at a workflow event, and what each states there about authority, execution, verification, consequence and record. Method: A qualitative document study of 121 items archived byte-exact: documentation for four agentic coding tools, the platform controls on their output, and eighteen policy documents from seven providers, read by deductive content analysis against an a-priori system of five dimensions and nine workflow events. An independent second coder blind-recoded the verification-mechanism classification, a complete enumeration and not a sample (Cohen's kappa 0.81, n = 13, on whether a mechanism compels; 0.75, n = 12, on who performs it). Results: Cursor's terms make the user responsible for evaluating the use of any suggestion; its documentation describes a product that performs that evaluation and records it as an approval. No collected artifact records that the verification the terms make the user's duty took place. At merge two tools compel a person, one documents an agent approving below a configured risk threshold, and one only advises. Six of the fifteen mechanisms that compel a check have a default the vendor states; nine have one this study inferred. Readers who had not made them refuted five of the eight absence claims; three survived, one materially qualified. Conclusions: The terms attach duty and consequence to output as a class, the platform records events, nothing records the duty discharged. One vendor documents a product that forms the approval judgement, and the terms allocate consequence against the artifact regardless.
Read the full abstract
Machine Learning
PL-Guard: Probabilistic Logic Reasoning for LLM Guardrails
Large language model guardrails can be viewed as policy-consistency problems: a system must determine which policy-relevant facts hold in a prompt-response pair and what those facts imply under a given policy. Common approaches, including policy prompting and LLM-as-a-judge pipelines, often overlap the tasks of semantic grounding and policy reasoning: the model both interprets the prompt-response pair and reasons about whether a policy has been violated. This can lead to unsafe compliance with harmful prompts, or refusals to assist benign ones. To separate grounding and reasoning roles, we propose PL-Guard, a neurosymbolic guardrail architecture. Using a symbolic policy interface consisting of predicates and ProbLog rules, a local LLM grounds prompt-response pairs into predicate probabilities using renormalized True/False token scores, while ProbLog performs explicit probabilistic rule inference over the symbolic policy. On the XSTest benchmark, an offline Qwen-based evaluator finds that PL-Guard with a hand-curated policy reduces unsafe compliance from 22.0% for the base model to 0.5%, and below the 6.0% rate of an LLM-as-a-judge baseline. This comes at the cost of higher over-refusal than the LLM-as-a-judge baseline, 14.4% versus 5.2%. These results suggest that separating neural grounding from probabilistic symbolic reasoning can expose the safety-helpfulness tradeoff while making the guardrail's intermediate reasoning steps explicit and auditable.
Read the full abstract
Discrete Mathematics · Cross-listed from math.CO
Repetition Avoidance in Curling-Number Transforms
We study repetition avoidance in a word ${\bf w}$ and its curling-number transform $C({\bf w})$. For alphabets of sizes $2$, $3$, and $4$, we use Thue-Morse-based morphic constructions and exhaustive finite searches. A ternary word for which both ${\bf w}$ and $C({\bf w})$ are overlap-free has length at most $84$, whereas over four letters an infinite example exists. Hence $4$ is the smallest alphabet size admitting simultaneous infinite overlap-freeness. The infinite constructions are verified in Walnut; the finite maxima are obtained by exhaustive breadth-first search and checked independently.
Read the full abstract
Machine Learning
Large Discovery Models: Empirically-grounded Model-Based Open-Ended Search
Scientific discovery often involves optimising expensive-to-evaluate objectives over vast, structured, and open-ended hypothesis spaces, such as molecules, protein sequences, and computer programs. Generative models such as large language models (LLMs) provide expressive priors over such spaces, but their likelihoods and self-assessments are unreliable proxies for the objectives and calibrated epistemic uncertainty, especially for novel candidates outside the observed data distribution. We introduce the Large Discovery Model (LDM), an empirically grounded recurrent architecture that couples a generative model with a Bayesian non-parametric reward surrogate model. The generative model proposes and refines candidate designs, while the surrogate predicts their performance and quantifies uncertainty, yielding an uncertainty-aware value that guides candidate generation, refinement, and selection. The discovery memory and the surrogate model are continually updated as each new experimental observation arrives. We evaluate LDM on three scenarios spanning different design modalities and objectives, including neural-network training, antibody design, and molecular optimisation. Compared to LLM-only reflection or traditional statistical search across these domains, LDM achieves a $2.4\times$ greater reduction in validation BPB, an $18.2\%$ relative decrease in binding energy, and more than $60\%$ relative gains in molecular multi-objective performance. These results suggests that LDM could serve as a general-purpose discovery engine for effective search over open-ended hypothesis spaces.
Read the full abstract
Software Engineering
An Empirical Comparison of Monolithic and Microservices Architectures for an E-Commerce Application
Microservices architectures are widely adopted for their promised scalability and modularity, yet empirical evidence comparing their runtime performance to monolithic designs remains context-dependent. This paper presents an experimental comparison of a monolithic and a microservices implementation of the same e-commerce application, both backed by a shared PostgreSQL database. Using k6, we subject both systems to identical HTTP workloads at 50 and 100 virtual users (VUs) over 60-second runs, measuring throughput, latency, and error rates. At 50 VUs, both architectures perform similarly with no errors. At 100 VUs, the microservices design achieves 5.4% higher throughput, 25% lower average latency, and 39% lower p95 latency than the monolith, while exhibiting a lower median error rate (0.00% vs 0.69%). The monolith shows consistent order-creation failures under load, whereas microservices failures are transient and confined to the cart service in one run. These results suggest that, in this deployment context, decomposing the system into microservices improves scalability and tail latency under stress, while introducing distinct, service-specific failure modes that must be managed.
Read the full abstract
Machine Learning
SubZero+: Memory-Efficient Adaptive Zeroth-Order LLM Fine-Tuning in Random Subspaces
Zeroth-order (ZO) optimization with SGD in random subspaces enables memory-efficient fine-tuning of large language models without backpropagation. However, high gradient estimation noise fundamentally undermines adaptive optimizers like Adam. We propose SubZero+, which achieves practical adaptive ZO optimization through a carefully designed dual low-dimensionality strategy: (i) multi-query forward-difference gradient estimation in periodically refreshed random subspaces to mitigate noise amplification in moment buffers, and (ii) Adam updates with periodic restarts performed directly in low-dimensional space rather than full-parameter space. In experiments, this dual design retains memory overhead comparable to momentum-free ZO methods while achieving stronger optimization performance than the evaluated ZO baselines. Theoretically, in the exact-directional limit, $K$-query averaging preserves conditional unbiasedness, while the coefficient estimator's covariance and mean-squared error, as well as query-induced second-moment inflation, scale exactly as $1/K$. Extensive experiments across SuperGLUE with models from 1.3B to 32B parameters under both full fine-tuning and LoRA schemes demonstrate consistent improvements over competing ZO methods. SubZero+ significantly narrows the performance gap with first-order optimization while preserving ZO's inference-time memory efficiency.
Read the full abstract
Machine Learning
Sequential Multimodal Evidence Optimization for Product Media Ranking in E-Commerce
On modern e-commerce stores, customers consume ordered slates of heterogeneous product media, such as images, videos, and 3D renders, before making purchase decisions. Existing media-ranking systems often optimize myopic engagement proxies such as clicks or dwell time, even though product media assets are cooperative informational components of the same item that together help customers find the information they need through sequential interaction. We present Sequential Multimodal Evidence Optimization (SMEO), a two-stage utility-guided framework for customer-oriented media sequencing. SMEO first learns a trajectory utility model from consumed media prefixes to estimate how ordered evidence helps customers reach a purchase decision, while mitigating position-bias and variable-depth imbalance in logged data. Recognizing that customer attention is a limited resource, it then trains an autoregressive ranking policy with survival-weighted reward-to-go that prioritizes the most decision-relevant information early, so customers can find what they need with less effort. By decoupling utility learning from policy optimization, SMEO enables stable offline learning from biased logs and post-hoc media attribution without explicit media-level labels. Evaluated offline on large-scale e-commerce sessions using doubly robust off-policy estimation, SMEO improves estimated conversion by 5.5% and helps customers reach a purchase decision with 15% fewer swipes than existing baselines.
Read the full abstract
Distributed Computing
Adaptive Heterogeneous Compression for Resource-Efficient Federated Knowledge Distillation
Federated learning (FL) enables privacy-preserving distributed model training but faces challenges from heterogeneous model architectures and limited communication resources at the network edge. Federated knowledge distillation (FedKD) alleviates model heterogeneity by combining prototype-wise parameter aggregation and knowledge transfer across heterogeneous models. However, transmitting gradients still introduces considerable communication overhead, while existing compression approaches typically apply a uniform strategy across clients and ignore their diverse model characteristics and resource capacities. To address this issue, we propose a heterogeneous compression framework for FedKD that enables each client to select a compression strategy from a candidate strategy set. We formulate the compression strategy selection problem as a non-stationary stochastic multi-armed bandit (MAB), where each arm corresponds to a compression strategy. An efficiency-aware reward is designed by jointly considering local optimization improvement, global knowledge alignment, and execution time. Based on this formulation, we develop an Adaptive heterogeneouS Compression algorithm for fEderated kNowledge Distillation (ASCEND), which employs an exponential moving average (EMA)-enhanced $\epsilon$-greedy policy to balance exploration and exploitation. Experimental results on multiple datasets demonstrate that ASCEND effectively adapts to heterogeneous model and resource settings, reducing communication overhead and training time while maintaining competitive model accuracy.
Read the full abstract
Computer Vision
WorldRover: A Scalable Synthetic Video Data Engine for World Exploration with Rich Annotations
Learning to generate or reconstruct explorable worlds requires video paired with more than RGB: camera motion, scene geometry, temporal correspondence and, for interactive models, control signals. Real capture can provide some of these signals, but dense geometry and long-range correspondence usually rely on estimation or specialised instrumentation. Rendering provides these quantities directly, yet existing synthetic resources rarely combine them on the same frames while also supporting controlled changes of viewpoint and appearance. We introduce WorldRover, a data engine for generating richly annotated, long-range explorations of artist-built environments. At its core, WorldRover-Engine is an Unreal Engine pipeline that executes and offline-renders minute-scale routes while preserving their full trajectories and scene geometry. The same exploration can be replayed from first-person, third-person, and 360 panoramic cameras under different environmental states. Using WorldRover-Engine, we construct WorldRover-10M, whose sequences pair RGB with metric depth, camera trajectories, and trajectory-derived action signals throughout each exploration. Third-person subsets additionally provide dense optical flow, long-range 2D/3D point tracks with visibility, and a character trajectory distinct from the camera trajectory. The engine can render a traversal from first-person, third-person and 360 panoramic viewpoints, under different environmental states or with a neutral white material, while preserving the route and scene geometry. WorldRover therefore turns long-horizon world exploration into a scalable data-generation problem, providing supervision for models that must build, maintain, and revisit coherent representations of an explorable world.
Read the full abstract
Artificial Intelligence
A Responsible Artificial Intelligence Framework for Groundwater Modeling
The rapid development and widespread application of artificial intelligence (AI) have sparked intense discussions on how to deploy responsible AI systems in a manner aligned with human values and ethical standards. Compared to fields like healthcare, energy, or finance, the application of AI in groundwater is relatively limited, and research on responsible AI is even more scarce. Taking the middle reaches of the Heihe River Basin as the study area, this paper proposes six Responsible AI principles: transparency, technical robustness, privacy governance, fairness, accountability, and sustainability. LSTM and Transformer time-series models are developed using multi-source hydrometeorological data, and validated via post-hoc interpretability, Monte Carlo simulation, and scenario analysis. The results show that Transformer outperforms LSTM in accuracy, robustness, and interpretability, demonstrating the operability and practical value of Responsible AI principles in groundwater prediction to support sustainable water management under climate change and human activities.
Read the full abstract
Emerging Technologies · Cross-listed from quant-ph
Rand-SEMI-QAOA: Finite-Budget Depth-One MaxCut Ensembles on Compressed Quantum Registers
We introduce Rand-SEMI-QAOA, a finite-budget QRAO–QAOA ensemble for MaxCut based on a $(3,1)$-QRAC relaxation. The method samples labels uniformly without replacement from the product-$X$ family of an implemented Galois stabilizer mutually unbiased basis (MUB) system. Each selected state–mixer pair is optimized independently for relaxed QRAO energy and then evaluated by deterministic Pauli-sign decoding. Exhaustive scans of the implemented noncomputational MUB catalog place the product-$X$ family first in 18 of 20 validated family-mean cells and in every tested cell for $r=5,6,7$. On a complete cohort of $2{,}400$ random connected 3-regular MaxCut instances with $n\in\{18,20,22\}$, the capped matched-cardinality schedule attains a graph-mean decoded best-of-set approximation ratio of $0.9421$ with one QAOA layer, while the $K=r^2$ schedule attains $0.9319$. These statistics are conditional on the disclosed frozen selector pools and do not estimate variability over selector seeds. An exact gauge identity shows that product-family labels generate the orbit of the QRAO Hamiltonian under an $r$-dimensional sign-gauge group. For common angles and relaxed energy, deterministic syndrome analysis proves branchwise dephasing, while an anisotropic Gaussian surrogate controls the coherent label-averaged response on the scale $\beta=b/r$. The theory does not order independently optimized decoded maxima. The results are finite-size and resource-explicit and do not establish quantum advantage.
Read the full abstract
Computation and Language
When Stories Evolve: Benchmarking LLM Storytelling Across Agent Architectures in Open-Ended World Simulations
Large language models can write fluent stories, but open-ended storytelling requires more than local fluency. In evolving world simulations and AI-native games, models must preserve facts, relationships, causal dependencies, and character states as the world changes. We introduce WSE-bench, a process benchmark that separately evaluates sustained generation, canonical coherence, and meaningful development in dynamic LLM storytelling. Generation Coverage records the proportion of planned narrative steps produced; Consistency tracks when canon breaks; and Richness measures how meaningfully branching, player-shaped trajectories develop. Across frontier models, Consistency and Richness do not form a smooth trade-off: their empirical Pareto frontier is non-concave, with several non-dominated intermediate configurations that no positive linear weighting can select. Added structure can enrich trajectories, but it does not uniformly improve coherence and may shorten them. Model scale chiefly improves sustained generation, without producing reliable gains in canonical coherence or meaningful development. These results show that sustained generation, canonical coherence, and meaningful development are distinct and sometimes competing capacities. WSE-bench makes those dynamics visible by extending narrative evaluation from finished stories to the processes that create them.
Read the full abstract
Computer Vision
Scalable Black-Box Model Attribution for Images
The rapid proliferation of generative models raises the model attribution problem: given only an image, can we determine which model produced it? We propose a lightweight CNN to solve this problem in a strict black box setting. The CNN operates on multiple image patches to handle varying image size and improve accuracy. It attributes more models at higher accuracy than prior work, reaching 98.9% on 25-class DRAGON and 95.0% on 27-class OpenFake; runs in a few milliseconds at a cost nearly independent of candidate-set size; and remains robust to transformations encountered in the wild. Beyond closed-set attribution, its learned representation serves as a reusable fingerprint backbone: it supports open-set rejection and few-shot enrollment, recovers model-family structure without lineage supervision, and groups images from unseen generators by source. Controlled ablations and causal perturbations show that its attribution decisions are driven by a spatially local, low-level signal that behaves as a generator-specific fingerprint.
Read the full abstract
Computer Vision
Gaussian-JEPA: Joint-Embedding Predictive Learning for 3D Gaussian Splats
3D Gaussian Splatting (3DGS) represents 3D content with anisotropic primitives that jointly encode geometry and appearance. Fixed-budget encoders consume sampled observations of Gaussian assets, so the same object may be observed through different primitive realizations. Existing self-supervised methods mainly reconstruct masked Gaussian attributes, tying supervision to one sampled realization and requiring an input-space decoder. Latent prediction offers an alternative, but its application to Gaussian tokens requires targets that accommodate coupled attributes and heterogeneous spatial support. We introduce Gaussian-JEPA, which predicts representations of held-out Gaussian token blocks from visible context. An online encoder processes the context, while a shared exponential-moving-average encoder supplies stop-gradient features for multi-scale targets. Complementary target projections and feature-space grounding provide latent supervision without reconstructing Gaussian attributes. We evaluate the features under Gaussian resampling, partial observations, and renderable shape completion, together with transfer to part segmentation and object classification. Compared with matched reconstruction pretraining, Gaussian-JEPA is more consistent across resampled inputs, retains more instance information under partial observations, and provides stronger frozen features for Gaussian completion. These results support latent prediction as an effective objective for reusable 3D Gaussian representations. Code is on the project page (https://amazingren.github.io/Gaussian-JEPA/).
Read the full abstract
Machine Learning · Cross-listed from stat.ML
On Stopping Rules and Spatial Adaptation for CART
The popular CART algorithm for regression trees combines a greedy splitting rule with a stopping rule, but while the splitting rule has been well studied, the statistical role of stopping rules is less well understood. Meanwhile, although regression trees fit using Bayesian methods or via empirical risk minimization (ERM) have been shown to be spatially adaptive to local smoothness and anisotropy, it is unknown whether CART can achieve the same adaptation. We address these gaps by proving that, under spatially heterogeneous and anisotropic smoothness and appropriate structural assumptions on the regression function and covariate distribution, CART with the minimum impurity decrease (MID) stopping rule and a suitable threshold achieves pointwise rates that are minimax up to logarithmic factors. These rates hold simultaneously over all points in the domain. Moreover, we prove that spatial adaptation cannot be achieved under the widely used minimum leaf size stopping rule. Together, these results establish a precise statistical role for the MID stopping rule and provide a theoretical basis for the empirical success of CART.
Read the full abstract
Computer Vision
Hierarchical Adaptive Feature Refinement Network for VHR Remote Sensing Image Segmentation
Semantic segmentation of very-high-resolution (VHR) remote sensing imagery increasingly benefits from strong pretrained hierarchical encoders, yet exploiting their multi-stage representations remains difficult. Nearby regions demand different balances between fine detail and semantic context, aggressive task-specific transformations perturb useful pretrained features, and conventional semantic supervision provides limited structural guidance. We present HAFR-Net, a progressive refinement framework that adaptively organizes and conservatively refines hierarchical representations instead of replacing them with a monolithic decoder transformation. Heterogeneity-Guided Stage-Adaptive Fusion (HG-SAF) predicts dense stage weights conditioned on local feature variation. A Frequency-Residual Adapter (FRA) then injects frequency information through a bounded, zero-initialized residual branch that keeps the fused representation as its reference. A Confusion-Aware Tri-Prior Decoder (CATP) finally regularizes the prediction with boundary, objectness, and training-derived class-relation cues. Under a matched Swin-B training and single-scale inference protocol, HAFR-Net attains 84.12%, 87.86%, 55.17%, and 67.70% mIoU on ISPRS Vaihingen, ISPRS Potsdam, LoveDA, and OpenEarthMap, improving the matched UPerNet baseline by 0.55, 0.95, 1.55, and 1.84 percentage points, respectively. Controlled analyses further show consistent spatial reweighting beyond content-only routing, improved boundary and thin-structure accuracy over matched spatial and spectral alternatives, and reduced confusion on pre-declared class pairs.
Read the full abstract
Human-Computer Interaction
Situated Practice Systems: A Computational System for Supporting the Coaching and Practice of Regulation Skills for Innovation Work
Students are increasingly expected to prepare for open-ended innovation work, which requires well-developed cognitive, metacognitive, and emotional regulation skills. College learning environments offer opportunities to work on real-world problems–such as in design and engineering–but students often remain unaware of their ineffective work practices and recurring regulation challenges, and may struggle to improve. Coaching from experts can help, but students' practices and regulation behaviors are largely invisible from work artifacts alone and are difficult to diagnose and track without computational support. We introduce Situated Practice Systems (SPS), which provide: (1) an Interactive Context-Assessment-Plan (CAP) Notes tool to support coaches' understanding and modeling students' regulation-informed practices, and (2) Practice Agents that help students develop more effective practices. SPS uses Practice Objects to represent practices and regulation behaviors computationally, and Practice Scripts to automatically present suggested practices to students in relevant situations. In a formative 3-week field study, SPS helped coaches identify recurring regulation gaps and provide tailored practices. SPS also guided students in adopting more effective ways of working on their own and with others. We demonstrate how CSCW systems and learning environments can be designed to support the development of students' work practices and regulation skills, enabling them to lead innovation work.
Read the full abstract
Machine Learning
Generalised Transportability via Causal Abstractions
Transporting a causal conclusion from a source study population to a target one is a fundamental problem in causal inference. The theory of transportability provides a criterion for when this is possible: given experimental data from the source and observational data from the target, it determines whether a target query is identifiable and does so completely; i.e. if the query can be transported, the criterion finds the exact formula. However, it works one query at a time and returns an expression rather than the value itself. It is also silent in two practically important regimes: when the query is not transportable and when no target data exist at all. To tackle both, we take a model-level perspective grounded in Causal Abstraction theory. Source and target share variables, graph, and interventions, differing only at a known set of mechanisms, which makes transportability a special case of same-level abstraction. Thus, instead of asking whether one query transports, we ask whether a single map aligns the source and target across their interventional behaviour. We characterise when such a map exists in both the Markovian and semi-Markovian settings; when it does, every target query transports at once. Our main contribution lies in the approximate case. When no exact map exists, the best approximate one still yields certified query intervals, recasting abstraction error as a quantitative notion of approximate transportability. We formulate model-level transport as distributionally robust optimisation over mechanism and environment perturbations of the unseen target and derive certificates for both challenging regimes: bounds for non-transportable queries, and guarantees under target-agnostic settings. We evaluate our framework on synthetic Markovian and semi-Markovian benchmarks and a real ecological dataset, and we show that the certified intervals bracket the true interventional query.