arXiv:2607.09739v1 Announce Type: new
Abstract: We study LLM benchmark coreset selection: selecting a small subset of prompts over multiple benchmarks whose induced model scores and rankings approximate those obtained from the full benchmark suite. In evaluation-unsupervised benchmark coreset selection (our approach), the selection algorithm uses no model evaluation outcomes, and operates on a fine granularity by producing subsets of prompts over multiple benchmarks rather than producing a sub-collection of entire benchmarks. We use submodular subset selection, and we develop and evaluate many different submodular functions for this purpose, including determinantal point process (DPP) based approaches, submodular mutual information functions, and facility location-based functions. On a new large-scale suite of 35 heterogeneous benchmarks spanning five different capability categories, 18 frontier LLMs, and over 61K prompts, we find that the facility location (FL) function operating exclusively on inexpensive semantic prompt embeddings preserves LLM scores better than twelve separate score-based and diversity-based baselines, across a range of coreset budgets. Moreover, we show our proposed objective is not limited to the evaluation-unsupervised regime: in the setting where only a handful of whole benchmarks must be selected and a large amount of model scores are available, the same objective matches or outperforms state-of-the-art baselines on the MMLU and MTEB leaderboards, while being substantially cheaper to compute. Together, our results suggest that submodularity, in general, is a strong and reliable tool for benchmark compression.
Science Journals
arXiv:2607.10004v1 Announce Type: new
Abstract: Unified multimodal models (UMMs) with interleaved reasoning, which generate both textual and visual steps as part of intermediate reasoning traces, have demonstrated great potential for visual mathematical reasoning tasks. However, we identify a key insight in this paradigm: generating intermediate visual reasoning steps is not always beneficial and can even be harmful, as self-generated visual steps may introduce erroneous visual evidence that misleads subsequent reasoning. Moreover, frequently triggering visual steps during reasoning incurs substantial computational and memory overhead, degrading inference efficiency. To address these accuracy and efficiency challenges, we observe that the model's internal signals can indicate whether a visual step will benefit reasoning before the entire visual generation is completed. Specifically, this work identifies two internal signals: 1) Generation Intent, which reflects whether the model has a concrete textual plan for what to draw, and 2) Visual Fidelity, which measures whether the visual generation remains grounded in the original input image. Leveraging these internal signals, we propose AdaViG, a training-free adaptive visual gating method for unified multimodal reasoning. AdaViG dynamically evaluates each triggered visual step at an early visual generation stage and aborts it when both signals are weak, thereby preventing misleading visual evidence from entering the reasoning trace while avoiding unnecessary computation. Comprehensive experiments demonstrate that AdaViG improves accuracy by up to 5.7% while reducing visual generation FLOPs by 25.0%-91.0% and wall-clock latency by 15.4%-45.6%.
arXiv:2410.23187v3 Announce Type: replace
Abstract: We define the class of explorable automata on finite or infinite words. This is a generalization of History-Deterministic (HD) automata, where this time non-deterministic choices can be resolved by building finitely many simultaneous runs instead of just one. We show that recognizing HD parity automata of fixed index among explorable ones is in PTime, thereby giving a strong link between the two notions. We then show that recognizing explorable automata is ExpTime-complete, in the case of finite words or parity automata up to index [0, 2]. Additionally, we define the notion of {\omega}-explorable automata on infinite words, where countably many runs can be used to resolve the non-deterministic choices. We show ExpTime-completeness for {\omega}-explorability of automata on infinite words for the safety and coB\"uchi acceptance conditions. We finally characterize the expressivity of ({\omega}-)explorable automata with respect to the parity index hierarchy.
arXiv:2607.09811v1 Announce Type: new
Abstract: We present a framework for understanding and generating feature rich hair strands. Drawing upon both scientific and cultural expertise, we define strand texture as the various distinctive patterns (curling, switchbacks, twist, etc.) that are formed by forces internal to a hair strand. We begin by proposing a novel five-dimensional parameter space, intended to be a bijection with naturally occurring hair strand textures. This encoding is both qualitatively accessible, allowing users to readily locate their own hair in the parameter space, and quantitatively precise, allowing the generation of individual strands from texture inputs. Importantly, strand texture should be independent from the overall strand direction. In order to disentangle strand texture from the overall strand direction, we identify centerline geometry and use it to map strands into a canonical space (a strand texture space). We construct centerlines using a novel method that cleanly distills complex hair grooms, separating the strand texture from the overall style (parameterized by style guides). We enable the creation of new strands conforming to our parametric description of texture via a generative artificial intelligence approach supervised by a separate neural network trained to label candidate strands according to our five-parameter description. The ability to create new strands conforming to any desired texture enables groom editing using either texture transfer or user-provided inputs. We demonstrate results on a variety of hair types.
arXiv:2607.09877v1 Announce Type: new
Abstract: Vacation rental marketplaces face a structural imbalance on the supply side: a small fraction of properties receive most user interactions, while the long tail of new, niche, and seasonal listings generates too little behavioral signal for collaborative filtering to serve effectively. At Vrbo, item-based k-nearest neighbors (IBKNN) is a core candidate generation channel, but leaves tens of thousands of properties with no candidates and produces weak neighborhoods for sparsely interacted ones. We present a training-free, LLM-based candidate generation pipeline that complements IBKNN using static property metadata alone. An off-the-shelf LLM synthesizes diverse semantic queries per property, a pre-trained text encoder embeds them, and an approximate nearest-neighbor index retrieves candidates from an 11.7M-property catalog. A Union fusion strategy merges these with IBKNN while preserving the behavioral channel's ordering, guaranteeing no degradation on well-served properties, and a downstream learning-to-rank model re-scores the fused pool. Evaluated on 1.6M focal properties, the system extends candidate coverage to tens of thousands of properties IBKNN cannot reach, delivers its largest gains on the long-tail segment where behavioral methods are weakest, and matches or beats IBKNN at every K on shared properties. A downstream learning-to-rank stage further lifts the fused pool, yielding a complete candidate generation and re-ranking stack that serves the long tail without regressing well-served properties. We additionally show that Union fusion collapses the recall gap between a 3B open-weights LLM and frontier API-based models from 27-46% to under 1%, supporting self-hosted small-model deployment at marketplace catalog scale.
arXiv:2411.12721v2 Announce Type: replace
Abstract: Cyber-physical systems rely on integrated circuits (ICs), making them vulnerable to hardware Trojans that can remain dormant until triggered, causing functional disruption or information leakage. Detecting these stealthy attacks is challenging because they introduce only subtle changes in circuit behavior. We present a golden-chip-free hardware Trojan detection framework that combines time-domain and frequency-domain features extracted from side-channel power traces. The framework evaluates six artificial intelligence models, including random forest, gradient boosting, naive Bayes, deep neural network, long short-term memory, and graph neural network, and integrates them using a stacked ensemble classifier. Evaluation on the AES-Trojan benchmark demonstrates that the proposed ensemble consistently outperforms the individual baseline models, achieving a macro-averaged ROC-AUC of 0.987. The results show that combining dual-domain feature extraction with stacked ensemble learning enables accurate and robust detection of hardware Trojans directly from side-channel emissions without requiring a trusted reference IC.
arXiv:2607.09880v1 Announce Type: new
Abstract: Clinical time series are central to patient monitoring, risk assessment, and clinical decision support. However, they are often sparse, irregularly sampled, and asynchronous, making it difficult for models to identify the temporal evidence required for clinical Question Answering (QA). Existing benchmarks primarily focus on regularly sampled time-series QA or medical QA over static data, and therefore rarely assess whether models can faithfully ground their answers in irregular temporal observations. To fill this gap, we introduce CLIR-Bench, a benchmark for irregular clinical time series QA constructed from de-identified ICU records through a principled four-stage pipeline. CLIR-Bench contains 6,600 QA instances spanning 11 clinical variables, organized into four capability dimensions and 11 tasks. Each question is linked to explicit temporal evidence and task-specific answer derivation rules, enabling evaluation of both answer accuracy and evidence use. Experiments show that existing generalist models struggle to retrieve and reason over sparse clinical evidence, highlighting the need for stronger irregular time-series reasoning methods. Our code and data are available at https://huggingface.co/datasets/winall/CLIR-Bench.
arXiv:2607.10391v1 Announce Type: new
Abstract: Despite exposing rich intermediate representations, Vision Transformers (ViTs) are almost exclusively utilized as black-box feature extractors, where only the last layer is considered for downstream tasks. We challenge this convention by introducing the notion of recoverability: the capacity of intermediate representations to correct last-layer failures. By evaluating independent classification probes at every model depth across 16 datasets, we observe that intermediate probes correctly classify 18% to 76% of samples that the last-layer probe misclassifies. We show that these gains are not primarily driven by predictive diversity, but by a redundancy-correctness correspondence, where the internal hierarchy acts as a series of stable, redundant probes of a shared discriminative signal. While established horizontal ensemble strategies (i.e., across multiple models) can improve performance, they incur high computational cost and ignore this vertical signal within a single model. To bridge this gap, we propose VFusion, a principled vertical aggregation strategy employing a learnable mapping into a low-dimensional latent space that synthesizes features across the internal ViT hierarchy. VFusion substantially outperforms established aggregation baselines in both in-distribution and out-of-distribution settings, notably closing 45% of the accuracy gap between the best individual layer and a theoretical oracle performance. Our gains consistently generalize across model sizes and pre-training regimes, confirming that VFusion offers a robust and efficient alternative to horizontal ensemble methods. The code is available at https://github.com/francescodisalvo05/vit-vertical-fusion.
arXiv:2607.10912v1 Announce Type: new
Abstract: 3D Gaussian Splatting represents scenes as finite mixtures of anisotropic Gaussians whose number of components $K$ is set by heuristic density control or user caps. Variational Bayes Gaussian Splatting (VBGS) recast splat fitting as conjugate variational inference, but $K$ remains fixed. We replace the finite symmetric Dirichlet over mixture weights with a truncated stick-breaking Dirichlet-process prior -- and, as a theory-backed alternative, a sparse overfitted finite Dirichlet -- so that the number of occupied components adapts to the data while every update remains a closed-form coordinate-ascent step; a natural-gradient stochastic variant makes the per-step cost independent of the number of points. We give an exact monotonicity guarantee, a rigorous truncation-error bound correcting an anti-conservative large-$\alpha$ approximation in common use, and an honest account of what the fitted number of components estimates. Empirically: (i) the effective complexity $\hat{K}$ adapts to scene complexity and recovers the true $K$ within $\pm 1$ on well-separated synthetic data with regime-appropriate concentration; (ii) a deconfounded comparison shows the DP prior's contribution is complexity selection, not per-component efficiency -- converged DP fits exceed single-pass fixed-$K$ VBGS by +2.7 dB at matched budgets yet tie an equally converged fixed-$K$ baseline, and on 3D scenes DP-Splat matches or exceeds VBGS's held-out color prediction with 5.9-7.6x fewer components; (iii) the posterior-predictive color variance is well calibrated on model-matched synthetic data; and (iv) the ordering suggested by exact-posterior asymptotics reverses under mean-field coordinate ascent: the DP prior resists over-splitting while the sparse finite mixture saturates its truncation, a gap between variational practice and posterior asymptotics documented across three orders of magnitude in $N$.
arXiv:2607.09890v1 Announce Type: new
Abstract: CorrNet is a strong baseline for continuous sign language recognition (CSLR) because it models inter-frame correlations inside the visual encoding stage. In this paper, we study two natural extensions of a reproduced CorrNet system: replacing the BiLSTM temporal head with a Transformer encoder, and injecting motion cues after temporal pooling. We find that the Transformer head does not outperform the BiLSTM baseline, even with a training strategy adjusted for the Transformer, and the two heads have almost the same computational and runtime cost. For the second extension, we design a lightweight module called MotionGate. In our experiments, MotionGate consistently collapses to an identity-like mapping: the gate loses motion selectivity, and the injected residual becomes a weak, non-selective perturbation of the pooled features. These results suggest that explicit motion injection after CorrNet's correlation-based encoding is largely redundant, and that natural-looking architectural extensions in CSLR should be tested carefully instead of being assumed to help.
arXiv:2607.09974v1 Announce Type: new
Abstract: We formulate rate-independent elastic-ideally-plastic response directly in hereditary form and study its approximation by finite history surrogates. At the material-point level, the constitutive law is the vector play operator generated by metric projection onto a closed convex elastic domain in stress space. Starting from the closest-point return mapping for step inputs, we pass to absolutely continuous driving histories, for which the constitutive law admits a differential form: the stress remains in the elastic domain and the difference of rates belongs almost everywhere to the normal cone. In this $W^{1,1}$ setting, the hereditary law is causal, contracts variation, and satisfies a $BV$-to-$L^\infty$ stability estimate. We then approximate histories by right-continuous step surrogates with at most $N$ constant pieces. For absolutely continuous inputs, we prove a sharp minimax theorem for input approximation in $L^\infty$, normalized by the $BV$ norm: the optimal encoder is given by equal-variation sampling. For constitutive approximation, the correct vector-valued minimax statement is obtained by allowing the encoder to be material-law aware: it may compress the exact stress history $\mathcal P(\pi)$ rather than only the driving history $\pi$. The resulting stress-aware encoder, followed by the same discrete hereditary decoder, gives the sharp value $(2N)^{-1}$ under the natural nondegeneracy assumption $0\in\operatorname{int}C$. The scalar complementary-variable case is also recorded: there the input equal-variation encoder is sharp because the scalar stop/play operator is $L^\infty$-nonexpansive in the complementary variable. The results identify cumulative variation as the natural variable for sampling and compressing both driving and constitutive histories.
arXiv:2412.16632v3 Announce Type: replace
Abstract: Ride-hailing platforms frequently face spatiotemporal supply-demand imbalances caused by uneven passenger demand and decentralized driver decision-making. Existing vehicle rebalancing methods typically assume drivers always follow repositioning recommendations or model adherence using static probabilities. In practice, adherence evolves through repeated interactions with the platform. We propose the Adherence-Aware Vehicle Rebalancing (AAVR) model, which generates simultaneous fleet-wide repositioning recommendations while explicitly accounting for driver preferences and dynamically evolving adherence. The resulting optimization problem is computationally intractable, so we derive a tractable upper-bound reformulation that enables real-time recommendation generation for large-scale systems. Simulations on the NYC taxi dataset under dynamic adherence updates show that AAVR consistently outperforms state-of-the-art methods, improving served demand by 26.72%, reducing passenger waiting time by 26.45%, increasing platform and driver profits by 25.90% and 28.75%, respectively, and improving fleet adherence by 30.06%. These results demonstrate that modeling evolving driver adherence improves both operational performance and long-term adherence to platform recommendations.
arXiv:2607.10428v1 Announce Type: new
Abstract: Evaluating large language models (LLMs) as multi-turn conversational partners requires probing capabilities that single-turn benchmarks miss: persona consistency, evolving intent tracking, emotional dynamics, and goal completion. We introduce EYT-Bench, a human-centered benchmark built around a three-party decoupled design: a persona-grounded user simulator, a target model that separates intent perception from response generation, and an independent third-party LLM judge with optional multi-judge ensembling. Personas are sampled from public human-curated corpora, Nemotron-Personas-USA and PersonaMem-v2, rather than synthesized, reducing LLM-induced persona bias. EYT-Bench also introduces two trajectory-level metrics: embedding-based intent drift and final-intent completion rate (FICR), inspired by tau-bench. In a 17-target x 200-dialogue evaluation, EYT-Bench reveals four findings: (i) state-of-the-art closed- and open-source models are statistically close on subjective dimensions (empathy / persona / anthropomorphism vary within <= 0.3), but differ by up to 9x on objective intent tracking; (ii) reasoning ("thinking on") sharply improves objective tracking on long-context personas (+0.47-0.50 latent-intent accuracy on Gemma-4) while leaving subjective scores nearly unchanged; (iii) persona format dominates trajectory spread, with FICR saturating above 0.95 on Nemotron-USA but spreading from 0.53 to 0.88 on PersonaMem-v2; and (iv) the warm-up effect is robust on 16/17 models (one outlier, GPT-5.5, reverses the effect), with stable rankings across alpha in [0.05, 0.15]. A cross-judge ablation using deepseek-v4-pro confirms that target rankings and final-intent satisfaction are preserved across judges.
arXiv:2607.10009v1 Announce Type: new
Abstract: Non-fullerene acceptors such as Y6 have enabled high-efficiency organic photovoltaic devices and motivated the development of single-component architectures; however, the microscopic mechanisms governing exciton transport and charge dissociation remain under active investigation. In particular, the interplay between Frenkel-charge-transfer excitations and their coupling to environmental fluctuations complicates the description of light absorption and subsequent exciton dynamics. Here, ultrafast transient absorption spectroscopy is used to probe exciton quenching dynamics in Y6 films interfaced with hole-transport layers. To interpret these measurements, we develop an analytical model based on hybrid Frenkel-charge-transfer states that enables direct extraction of intermolecular electronic couplings, charge-transfer character, and system-bath interaction strengths from experimental data. The analysis reveals a substantial charge-transfer admixture of 20-40% in the exciton states and identifies a transport regime characterized by delocalization-mediated exciton motion rather than purely diffusive hopping. Consistent with this interpretation, the corresponding quenching dynamics occur on a ~1 ps timescale within ~4 nm of the interface, suggesting a short-range injection mechanism facilitated by exciton delocalization. In addition to providing physical parameters for Y6, these results establish a quantitative framework that connects spectroscopic observables to microscopic transport mechanisms and can be generalized to other non-fullerene acceptors.
Fourth-order Optoelectronic Response from Cascaded Circular Photogalvanic and Nonlinear Hall Effects
arXiv:2607.10012v1 Announce Type: new
Abstract: The interplay between nonlinear optical transitions and topological band structure offers a route to control photocurrents. We reveal a fourth-order optoelectronic response that emerges due to an interlink between the circular photogalvanic effect (CPGE) and the Berry curvature dipole (BCD) in noncentrosymmetric 2D materials. Using monolayer $\Td$-WTe$_2$ as a prototype, we predict that circularly polarized mid-infrared light produces a steady dc injection current that induces an internal electric field, which in turn drives a transverse nonlinear Hall response through BCD. The resulting cascaded photovoltage scales as the fourth power of the optical field $E_0^4$. By mapping the full injection current tensor, we show that this cascaded voltage is strongly tunable by the optical geometry: normal incidence drives an in-plane resonance $\mathrm{Im}(\eta_{yxy})$, whereas oblique illumination ($\theta = 45^{\circ}$) recruits a dominant out-of-plane component $\mathrm{Im}(\eta_{yyz})$ and amplifies the signal by more than two orders of magnitude (${\sim}10^2~\mu$V). While the massive linear Drude background typically screens nonlinear responses in semimetals, we argue that the amplitude modulation of the optical pump allows lock-in detection to cleanly isolate the frequency-doubled cascaded response. The proposed mechanism converts mid-infrared light into a gate-tunable transverse signal, providing a route for probing quantum geometry and realizing topological photodetectors and frequency doublers.
arXiv:2607.10041v1 Announce Type: new
Abstract: We study a symbolic search space for the Collatz conjecture based on finite exponent codes of the accelerated map. Each code records the number of divisions by two after every 3n + 1 step and determines three quantities: real drift, a 2-adic start representative, and a 3-adic endpoint representative. Their combination defines the 2-3-infinity diagnostic. Counterexample-like codes should exhibit near-critical drift, small 2-adic start representatives, and endpoints compatible with growth on the scale of (3/2)^k. We prove that every infinite code generated by a fixed positive integer has asymptotically vanishing 2-adic and 3-adic residue rates. Experiments with random critical codes, mechanical critical codes, and adaptive evolutionary search at lengths 100, 200, and 400 show that adaptive search improves finite-length trade-offs, while all methods retain clearly positive residue rates. The proposed framework is not a verification method for the Collatz conjecture, but a symbolic diagnostic approach for investigating obstruction structures in exponent-code space.
arXiv:2607.09911v1 Announce Type: new
Abstract: Multi-robot path planning in human-shared environments requires a delicate balance between robust inter-robot coordination and socially aware behavior. While diffusion models excel at generating predictable, human-like paths, existing generative planners are often restricted to paths of fixed duration and high computational latency, limiting their adaptability to varying goal distances and hindering real-time deployment. We present Multi-Robot Rolling Diffusion (MRRD), a novel framework that enables real-time, long-horizon navigation for large robot teams through dense crowds. MRRD combines a rolling-horizon scheme to accommodate the limited prediction horizon of human motion, parallelized diffusion inference for scalable generation of human-like paths, and a conflict-based-search mechanism for resolving inter-robot collisions. It further incorporates urgency-based temporal conditioning to generate paths with varying speeds and employs differentiated guidance terms to maximize both social awareness around humans and efficient coordination between robots. Experimental results in crowded environments demonstrate that MRRD successfully scales to 15 robots in real-time, significantly outperforming existing baselines in both safety and mission success rates.
arXiv:2607.10372v1 Announce Type: new
Abstract: The transition of autonomous mobile robots from controlled industrial settings to dynamic, human-centric environments, such as manufacturing, logistics, and healthcare, has made their safe and autonomous operation a critical area of research. These sophisticated machines must be capable of perceiving, understanding, and interacting with their surroundings to navigate freely and perform complex tasks. A significant obstacle to achieving this is the lack of comprehensive contextual awareness, which requires a robot to recognize its spatial environment and identify the objects and actors within it. Without this perceptual knowledge, robots struggle to plan adaptive behaviors or engage in meaningful interaction with humans.
This thesis presents novel solutions to this challenge by exploring two distinct but complementary research directions. The first direction involves human re-identification and tracking to improve Human-Robot Collaboration. Our developed approach enables a mobile robot to recognize a specific person, facilitating targeted collaboration while ignoring other individuals. The second direction focuses on enhancing the robot's overall perceptual capabilities to understand its environment geometrically and semantically. Geometric information is vital for motion planning and collision avoidance, while semantic knowledge provides the robot with a richer understanding for more advanced interaction. Both solutions are driven by the improvement of the semantical understanding of robots that enhance their knowledge of their surroundings, allowing a smoother and more natural interaction between robots, humans, and the environment. The contributions of this work in human re-identification and environmental understanding represent a significant step toward a future where robots are more contextually aware, enabling safer coexistence and more effective collaboration.
arXiv:2501.13692v2 Announce Type: replace
Abstract: Diffusion models have recently unlocked new possibilities in editing images of real-world objects. Yet, transforming objects in non-rigid ways, such as modifying poses or applying image-based conditioning, continues to present significant challenges. Retaining the unique identity of objects during these edits is a complex task, and current techniques often fall short of delivering the precision needed for industrial settings, where consistency is non-negotiable. Additionally, adapting diffusion models demands custom training data, which is often unavailable in real-world scenarios. To address these gaps, we present FashionRepose, a novel, training-free pipeline designed to handle non-rigid pose adjustments specifically for the fashion industry. This approach combines pretrained off-the-shelf models to modify the poses of long-sleeve garments while safeguarding their identity and branding characteristics. By adopting a zero-shot methodology, FashionRepose enables near real-time edits, entirely eliminating the requirement for specialized training data. FashionRepose has been deployed for a global fashion firm, OVS, handling more than 30,000 long-sleeve garments.
arXiv:2607.10232v1 Announce Type: new
Abstract: We study the fundamental problem of fairly dividing indivisible items among agents with additive utilities. In our model, an item can be a good yielding non-negative utilities to some agents and simultaneously a chore yielding negative utilities to others. We take the best-of-both-worlds perspective and our goal is to construct a randomized allocation that is exactly fair ex ante while also being supported on ex post approximately fair allocations. The fairness notions examined in this paper are envy-freeness (EF) and its well-known relaxation envy-freeness up to one item (EF1). Our main result is that ex-ante EF and ex-post EF1 can be achieved simultaneously. To achieve this, we introduce a novel probabilistic Hall-type matrix decomposition that intricately correlates the fractional assignments of goods and chores. We resolve this decomposition problem by combining continuous minimax duality -- via Sion's minimax theorem -- with carefully designed biased flow networks.
arXiv:2607.11025v1 Announce Type: new
Abstract: Face super-resolution is the task of increasing the resolution of an image containing a face thereby adding finer detail. It is a ubiquitous task in many computer vision applications and quite often the user isn't even aware that it is being performed. However, doing it with high fidelity is challenging as it is an ill-posed problem. In this paper we present a reference-based solution for face super-resolution that uses higher resolution reference images to aid in the task. We show an alignment module based on the spatial transformer that is considerably more stable than the popular deformable convolutions. We also show an aggregation function that can take good quality information from the reference images when available or suppress the function when such information is unavailable. Finally, we show that our relatively smaller model can achieve state of the art results on multiple datasets. The source code is available at https://github.com/varun-jois/FSRST.
arXiv:2502.00632v2 Announce Type: replace
Abstract: This study investigates the use of AI tools in academic writing through an analysis of AI usage declarations in journals. Using a mixed-methods approach combining content analysis, statistical analysis, and text mining, this study analyzed 135 AI declarations from 8633 articles across 27 categories. Results show that ChatGPT dominates academic writing assistance (73.3 percent usage). The primary purposes of AI integration are concentrated on lower-level cognitive tasks, specifically improving readability (57.8 percent) and grammar checking (19.3 percent). Statistical analysis indicates a highly significant association between team composition and AI-use purposes (p = 0.0008), highlighting international teams' reliance on grammar assistance, while no significant association was found regarding authors' native-speaker status (p = 0.2359). These findings provide insights for journal policy development and for understanding the evolving role of AI in academic writing.
arXiv:2607.09936v1 Announce Type: new
Abstract: Cybersecurity systems must adapt rapidly to emerging threats. However, labeled data for new threat categories is unavailable when those threats first appear. Generalized zero-shot learning offers a natural solution by enabling recognition of unseen classes through auxiliary semantic knowledge rather than labeled examples. Large language models are particularly promising in this setting because they can convert unstructured CTI reports into semantic prototypes for emerging threats. However, applying language-driven zero-shot learning to cybersecurity is difficult due to strong semantic overlap between threat descriptions, heterogeneity between behavioral attributes and text, severe class imbalance, and open-set conditions where unseen threats are unknown during training. We propose SMETA-ZSL, that learns semantic prototypes from overlapping language descriptions through contrastive finetuning, aligns behavioral features through episodic meta-learning and knowledge distillation, and performs adaptive routing for generalization across seen-unseen classes. Across 7 benchmarks, SMETA-ZSL delivers the strongest overall generalized zero-shot performance under the strictest inductive setting, surpassing prior methods by 10.8 points on average, with gains up to 18.1 points. Github:https://github.com/Security-And-Intelligence-Lab-UTEP/SMETA-ZSL
arXiv:2502.11007v5 Announce Type: replace
Abstract: Compared to traditional machine learning models, recent large language models (LLMs) can exhibit multi-task-solving capabilities through multi-modal data sources and multi-turn conversations. These unique characteristics of LLMs, together with their large model size, make their deployment more challenging. Specifically, (i) deploying LLMs on devices faces computational, memory, and energy resource issues, while (ii) deploying them in the cloud cannot guarantee real-time service and incurs communication/usage costs. In this paper, we design TMO, a device-cloud LLM inference system with Three-M Offloading: Multi-modal, Multi-task, and Multi-turn. TMO incorporates (i) a lightweight on-device LLM that can process simple tasks at high speed and (ii) a large-scale cloud LLM that can handle multi-modal data sources. We develop a resource-constrained reinforcement learning (RCRL) strategy for TMO that optimizes the inference location (i.e., device vs. cloud) and multi-modal data sources to use for each task in multi-turn conversations, aiming to maximize the long-term reward (response quality, latency, and usage cost) while adhering to resource constraints. We also contribute M4A1, a new dataset we curated across multiple modalities, tasks, conversation turns, and LLM configurations, enabling evaluation of offloading decisions. We demonstrate the effectiveness of TMO compared to several exploration-decision and LLM-as-Router baselines, showing significant improvements in latency, cost, and response quality. Our code and dataset are available at https://github.com/liangqiyuan/TMO.
arXiv:2607.09996v1 Announce Type: new
Abstract: Automated failure attribution uses LLMs to identify where and why agentic systems fail. As agents become more capable, their failures become subtler, making automated attribution increasingly important. We introduce Who&When Pro, a large-scale benchmark for automated failure attribution in agentic systems. Using a strictly controlled pipeline that injects a failure only after exactly replaying a successful prefix, we construct 12,326 failed trajectories with golden labels across 3 modalities and 26 benchmarks covering various scenarios. Beyond benchmarking, we conduct extensive experiments and analyses, revealing systematic patterns in how models attribute failures across modalities, protocols, and model families, and providing empirical guidance for future automated failure attribution systems.