arXiv:2606.18151v1 Announce Type: cross Abstract: Channel charting (CC) in real-world coordinates is a recently proposed self-supervised machine learning method that maps high-dimensional channel state information (CSI) to user equipment (UE) position. In this paper, we extend CC to also estimate UE orientation, which can further assist tasks such as beamfinding, precoding, and beam- and cell-assignment. To this end, we propose a novel orientation triplet loss that accounts for angle periodicity and an alignment loss that embeds estimated orientations in real-world coordinates in a self-supervised fashion. Using real-world CSI measurements from a standard-compliant 5G NR system, we demonstrate that the proposed method achieves position and orientation estimation accuracy close to that of supervised approaches trained with ground-truth labels.
Science Journals
arXiv:2503.12209v2 Announce Type: replace-cross Abstract: Beyond the study of individual materials, their interfaces and arising functionality are crucial enablers of fundamental science and technological progress. Recently, the arena of two-dimensional (2D) materials was extended to so-called non-van der Waals (non-vdW) compounds derived from strongly bonded non-layered bulk crystals. These nanosystems have advantageous active surfaces and thus now complement - through their chemically bonded heterostructure (HS) interfaces - the established space of weakly interacting vdW HSs. Here, we study a wide range of 55 candidate non-vdW HSs with extensive density functional calculations and find that the resulting interfacial chemical hybridization and bonding is the key factor determining their electronic and magnetic properties. It gives rise to the formation of unique hybrid interface bands, strong magnetic coupling, and substantial electronic as well as magnetic moire surface property modulations upon twisting. Our work thus provides a valuable platform for charting the new interface class of non-vdW HSs.
arXiv:2509.18448v2 Announce Type: replace Abstract: Liquid crystal (LC) film plasma mirrors (PMs) based on 4-octyl-4'-cyanobiphenyl (8CB) are an enabling technology for reflecting high-fluence laser pulses. These freestanding LC films can achieve high optical quality and are well-suited for rep-rated applications, as motorized devices continuously replenish films over an aperture following each destructive laser shot. However, a systematic characterization of film quality as a function of seminal operating conditions had not yet been performed for the LC "windmill" version of the device, which aims to match the repetition rate of an existing "spinning disk" (SDI) version and the angular stability of the "linear slider" (LSTI) version. We determined the 8CB film quality using low-power wavefront measurements, and studied the film-to-film wavefront stability and formation reliability. The film-formation reliability of 8CB LC films demonstrated >97% formation success at 2.7 mm/s film-forming speeds, but decreased to 45% at 10.8 mm/s. These reliability numbers will inform future designs to reach Hz-level repetition rates and beyond. Depending on area-of-interest within the 10 mm diameter film, the added wavefront root-mean-squared (RMS) variation was as small as 12 nm for a 2 mm diameter region, and <50 nm for a 3 mm diameter region. Within the optimal 21-22 degrees C operating regime, pointing fluctuations remained at or below 0.5 mrad. With a maximum effective film formation frequency of approximately 0.25 Hz, these results establish windmill-formed 8CB films as promising candidates to pursue next-iteration improvements towards rep-rated plasma-mirror operation.
arXiv:2606.16591v2 Announce Type: replace Abstract: Large language model (LLM) agents increasingly rely on agent harnesses that manage context, tools, and multi-turn execution, making tools a central interface for acting in realistic digital environments. As harness-connected tool ecosystems expand to hundreds or thousands of APIs, services, and task-specific skills, exhaustive tool schema injection becomes costly and imposes a closed-world assumption that limits agents to a predefined static inventory. Retrieval-augmented tool selection offers a natural alternative, but existing one-shot retrieval methods often fail to align isolated tool descriptions with the agent's true task intention, especially in long-horizon tasks where required capabilities emerge through decomposition, observations, and newly induced subgoals. We propose SING, an intention-aware active tool discovery framework that builds an intention-tool graph linking user intentions, tool capabilities, and tool collaboration patterns, and dynamically retrieves tools according to evolving task states. Using a unified corpus of 7,471 tools, we evaluate SING on three real-world tool-use benchmarks. SING improves Global Recall@5 by up to 59.8% and downstream success rate by up to 28.9% over baselines, while reducing full-corpus tool-schema exposure by 99.8%, demonstrating that intention-aware graph structure enables more accurate and context-efficient tool discovery in large-scale agentic ecosystems.
arXiv:2606.17873v1 Announce Type: new Abstract: Controller synthesis in power electronics-based systems depends predominantly on the mathematical model of the system, which is a limitation when the actual system is complex and the mathematical model cannot capture all its dynamics. Model-free control addresses this limitation by using an ad-hoc simple model which is compensated by high-rate evaluation of dynamics in terms of their derivatives. However, application of the model-free control strategy to power electronics-based multi-time scale dynamical systems is challenging because of the derivative action needed to implement such control. Grid-connected power converters are examples of such systems, yet experimental validation has not been adequately addressed in the literature. This letter presents the validation of such control including the hardware implementation level. An intelligent proportional-integral (iPI) controller is synthesized and validated on a 16 kW experimental test bench. This proves the benefits of the approach in control of grid-connected power converters, among which their participation in the secondary voltage control.
arXiv:2606.18033v1 Announce Type: new Abstract: Cross-lingual transfer in multilingual NLP has been widely explored in supervised fine-tuning contexts, where factors like data availability and linguistic similarity largely determine transfer quality. As the field shifts toward few-shot In-Context Learning (ICL), it is often presumed that insights from fine-tuning carry over unchanged. Yet this assumption has not been rigorously evaluated, leaving open the question of how to choose source languages for cross-lingual ICL. We conduct a broad empirical study of cross-lingual transfer in ICL spanning seven tasks, six models, and a typologically diverse set of languages. We further analyze language confusion, a key obstacle for generative tasks in cross-lingual ICL. Our results show that conventional fine-tuning-based expectations do not consistently apply in the ICL regime and point to alternative heuristics for selecting source languages effectively.
arXiv:2606.12556v2 Announce Type: replace Abstract: The rapid shift toward agentic and long-context workloads in Large Language Models (LLMs) is pushing the industry beyond the capacity of individual servers toward disaggregated shared storage to handle TB-scale context states. This movement has led to the emergence of specialized shared context layers designed to externalize and share cumulative inference states across distributed clusters. While offloading to a data processing unit (DPU) within just-a-bunch-of-flash (JBOF) architectures accelerates NVMe-over-fabrics (NVMe-oF) target processing, the need for sophisticated software-level optimization and cost-efficiency burdens remain significant. Consequently, the ideal architecture for scaling this shared context infrastructure is still an active area of exploration. In this paper, we propose ITME (Inference Tiered Memory Expansion), which leverages a CXL-hybrid memory to present a massive, TB-scale byte-addressable remote memory expansion. This approach enables cost-efficient scaling and simplifies the software stack through direct byte-addressability, effectively addressing the challenges of shared context infrastructure. Our key insight is that the deterministic access patterns of voluminous model weights and prefix caches enable the system to proactively manage data movement across the memory-storage hierarchy. We validate ITME by evaluating its performance potential with production-grade SK Hynix CMM and PCIe Gen5 NVMe SSDs, while further demonstrating its functional feasibility through an FPGA-based hardware prototype. Overall, ITME enhances conventional CPU-offloading by providing additional remote memory expansion to accommodate large KV cache footprints beyond host memory limits, achieving up to a 35.7\% throughput improvement.
arXiv:2606.18166v1 Announce Type: new Abstract: Classifying Cyber Threat Intelligence (CTI) using MITRE Adversarial Tactics, Techniques, and Common Knowledge (ATT&CK) is essential for proactive defense, but historically required extensive human effort. Pre-Large Language Model (LLM) automation sped up this process, but could not resolve the complex language and multi-step attack patterns found in unstructured CTI reports. LLMs addressed previous limitations by using contextual reasoning to understand unstructured text. However, current evaluations rely on simplified, single-technique sentences that ignore the complexity of real-world CTI reports, which often leads to inflated performance results. Consequently, the baseline performance of open-source LLMs on complex unstructured CTI reports remains unevaluated. To address this gap, we constructed a ground-truth dataset of 2,076 human-annotated sentences (1,281 technique-positive, 795 negative) from 83 complex unstructured CTI reports. These sentences were mapped to 114 unique ATT&CK techniques using a six-phase annotation process, achieving \k{appa} = 0.68 inter-annotator agreement. Using this dataset, we evaluated seven open-source LLMs ranging from 8B to 236B parameters across prompt strategy and temperature configurations. The highest-performing LLM achieved a micro-averaged F1 score of 0.22, establishing the empirical baseline for multi-label ATT&CK classification on complex unstructured CTI. Parameter size showed a statistically significant positive correlation with F1 score. Prompt strategy and temperature produced no statistically significant gains across model configurations. These results indicate that current open-source LLMs are insufficient for production-grade ATT&CK classification. The dataset, benchmark, and findings provide a reproducible foundation for future CTI research.
arXiv:2606.17068v1 Announce Type: new Abstract: This study focuses on the complex coupling between discharge and flow properties in a ns-pulsed He micrometer scale atmospheric pressure plasma jet ($\mu$APPJ). This is investigated by integrating electrical measurements, schlieren photography, ICCD imaging, and space-resolved Optical Emission Spectroscopy (OES) with Computational Fluid Dynamics (CFD) simulations. In the flow rate range QV=0.1-1 slm, a critical threshold emerges at 0.3 slm, where the discharge consumes the highest energy overall, achieving maximum propagation length and remarkable collimation. Below 0.3 slm, insufficient momentum renders the jet susceptible to buoyancy and air entrainment, leading to shorter effluents, while higher flow rates enhance shear layer instabilities. CFD simulations reproduce the schlieren flow profiles to quantify the axial helium mass fraction (YHe) confirming a stable helium-rich core at 0.3 slm (YHe=90%), not seen in other flow rates. Furthermore, lower and higher flow rates promote stochastic branching which is more pronounced at QV>0.3 slm. Numerous lateral branches are clearly distinguished and quantified via single-shot ICCD imaging for the first time in a He $\mu$APPJ. The increase of voltage amplitude (VP) in the range 4-9.5 kV, amplifies their activity in the effluent tip at 0.3 slm. At VP=9.5 kV, their number increases for QV<0.3 slm compared to QV=0.3 slm, while for QV>0.3 slm they occur much closer to the nozzle exit and intensify farther downstream. Timeresolved imaging reveals a distinct peak in ionization wave velocities (up to $\approx$600 km/s at 0.3 slm/9.5 kV) just after the nozzle exit, followed by a progressive decay which becomes more abrupt at the higher flow rates. This correlates spatially with a surge and subsequent axial drop in N2 + (FNS) emission intensity, indicating Penning ionization as a key mechanism behind this acceleration. Average gas temperature estimations (TGas$\approx$350 K) suggest…
arXiv:2606.17451v1 Announce Type: new Abstract: Automated Driving System deployments create a foundational ratemaking challenge: sparse experience, shifting operational design domains, and non-stationary risk across software releases. We propose a hierarchical Bayesian credibility framework pooling across cities, software versions, and territories via a learned ODD-similarity kernel, nesting Buhlmann-Straub as a limiting case. Demonstrated on 648 verified-engaged Waymo crashes across four U.S. metros from the NHTSA Standing General Order database against 116 million matched miles, city-aggregate credibility weights are moderate (0.12-0.46), partial pooling decisively outperforms no pooling, and a power analysis shows the learned kernel's advantage becomes detectable at approximately twelve deployed cities.
arXiv:2606.17456v1 Announce Type: new Abstract: Rolling over is one of the earliest milestones in infant motor development, reflecting the emergence of coordinated, whole-body sensorimotor control. Here, we conduct a computational study of infant rolling using MIMo, a virtual infant embodiment equipped with proprioception and vestibular sensation. MIMo learns supine-to-prone rolls with reinforcement learning. Interestingly, the learned behaviors capture developmental trends and coordination patterns consistent with those reported in real infants, including improved performance and faster execution with age. Our results explain how infant capabilities and constraints can give rise to realistic behaviors in artificial agents, with a particular emphasis on how motor development is shaped by the changing body morphology. This work highlights the role of embodied computational models as a powerful tool for studying sensorimotor development.
arXiv:2606.17458v1 Announce Type: new Abstract: With the rapid advancement of Deep Research Agents in knowledge-intensive domains such as finance, establishing reliable and domain-aligned evaluation standards remains a critical challenge. Existing benchmarks focus on either closed-ended question answering or open-ended report evaluation, failing to jointly capture retrieval-reasoning accuracy and end-to-end research quality required in real-world workflows. We introduce ICBCBench, a consortium-driven benchmark for financial deep research, developed in collaboration with domain experts from a broad range of financial institutions and academia, involving over 50 experts across more than 40 organizations. It adopts a dual-track paradigm integrating objective tasks with verifiable answers and subjective long-form report evaluation, enabling complementary assessment of retrieval-reasoning accuracy and end-to-end report quality in terms of expert alignment, citation consistency, and source quality. Experiments on state-of-the-art DRAs and large language models reveal substantial gaps in complex reasoning, factual grounding, and report quality, highlighting the challenges of achieving industry-level performance. Our dataset and evaluation framework are available at https://github.com/DeepFin-Intelligence/ICBCBench.
arXiv:2606.17507v1 Announce Type: new Abstract: Generative AI and large language models (LLMs) are increasingly applied to question generation and automated assessment. However, deploying LLMs in preparation for high-stakes exams requires more than prompt engineering; it demands software pipelines that systematically ground model outputs in authorised curriculum artefacts and marking guidelines issued by education authorities. This paper presents a curriculum-grounded, configurable LLM-as-Judge pipeline for question-level marking, co-developed with an industrial partner, to support exam preparation for university admission. The pipeline identifies the relevant topics, subtopics, and cognitive demand of a question, and assembles verifiable and authorised context to support LLM judgement. Curriculum intent is operationalised through concrete syllabus artefacts, including prescribed verbs and outcomes, performance band descriptors, glossary definitions, and marking-guideline principles. A staged LLM workflow is employed to first generate question-specific rubrics, capturing structured expectations of performance, and then derive and evaluate marking criteria used to allocate marks to student responses. This design improves consistency, transparency, and alignment with official marking practices. Preliminary evaluation shows that the proposed LLM-as-Judge pipeline delivers marking outcomes comparable to human tutors, while yielding justifications that are more traceable to authorised curriculum artefacts and marking standards. The pipeline has also been integrated into an online study platform, where early deployment data provide initial insights into operational usage and manual overrides.
arXiv:2606.17334v1 Announce Type: new Abstract: Event cameras are bio-inspired sensors that asynchronously capture logarithmic intensity changes, offering inherent advantages in high-speed and high-dynamic-range scenarios. However, the sparse and asynchronous nature of event streams poses a fundamental challenge for modern deep learning architectures. To enable compatibility with standard models, most existing approaches partition the accumulation window into fixed temporal sub-bins. While effective for spatial processing, this internal discretization discards fine-grained temporal structure and constrains inference to the low temporal frequencies imposed by training supervision. To address this limitation, we propose FATE, a unified framework built upon a novel Pillar Encoding (PE). While operating over discrete macro-accumulation windows dictated by the target frequency, PE avoids internal temporal sub-binning. It organizes events into spatial pillars and approximates their intra-window evolution via projection onto a continuous-time orthogonal polynomial basis. This formulation yields an L2-optimal representation that retains rich temporal dynamics in a dense pseudo-image, mitigating information loss under sparse event conditions. To fully leverage this representation, we introduce Frequency-Aware Training (FAT), a soft mean-teacher curriculum that generates temporally dense pseudo-labels, effectively bridging the mismatch between low-frequency supervision and high-frequency inference. Extensive experiments demonstrate that FATE generalizes across architectural paradigms and consistently outperforms strong baselines. It enables robust object detection at high temporal resolutions up to 200 Hz, while incurring minimal overhead in parameter count and inference latency
arXiv:2509.05879v2 Announce Type: replace Abstract: Millions of users visit Stack Overflow regularly to ask community for answers to their programming questions. However, like many other platforms, Stack Overflow consistently struggles with low user retention and declining levels of user contributions to the platform. With the introduction of ChatGPT in November 2022, these ongoing difficulties on Stack Overflow were further magnified, as many users moved toward ChatGPT for programming help. In this paper, we build upon recent research on this phenomenon by analyzing the transformation of user-generated content on Stack Overflow during the post-ChatGPT period. Specifically, we analyze two years of Stack Overflow data and fit multiple causal regression models to estimate the effect of ChatGPT on the length and difficulty of user questions and code examples. We confirm an acceleration of decline in user contributions but find that ChatGPT had a significant positive effect on question and answer length, code length, and question difficulty on Stack Overflow across programming languages. Our results suggest that ChatGPT has effectively raised the bar for questions on Stack Overflow, as users increasingly turn to crowdsourced platforms for help with more complex and challenging problems. With our work we contribute to the ongoing discussion on the impact of tools such as ChatGPT on help-seeking in programming and, more broadly, on collaborative knowledge creation. Our results provide actionable insights for platform operators to support information management and user retention in the aftermath of ChatGPT's launch.
arXiv:2606.17459v1 Announce Type: new Abstract: Evaluating the decision-making capabilities of large language models (LLMs) is a growing research priority, yet existing benchmarks focus on isolated cognitive tasks such as reasoning, knowledge retrieval, and economic rationality in stylized settings. These evaluations overlook the defining challenge of real executive decision-making: integrating conflicting recommendations from specialized stakeholders under information asymmetry, organizational constraints, and temporal dependencies. We introduce \textsc{CEO-Bench}, a multi-agent benchmark that evaluates LLMs on CEO-level strategic resource reallocation -- the process of redirecting capital across business units in a multi-round, constraint-rich organizational environment. In \textsc{CEO-Bench}, LLM agents receive conflicting advice from four role-conditioned C-suite advisors (CFO, CTO, COO, CMO), each with private signals and distinct priorities, and must synthesize these into a concrete allocation plan evaluated along four dimensions: role integration, conditional boldness, history-sensitive judgment, and plan validity. Experiments across five frontier models on 13 scenarios reveal that all models achieve high structural validity but diverge sharply on strategic calibration -- the hardest capability layer. We identify systematic failure modes including single-advisor capture, conservative default under ambiguity, and historical amnesia, and uncover a structural integration-boldness tradeoff: models that engage more deeply with conflicting perspectives tend to produce less decisive action. These findings delineate the current capability boundary of LLMs as organizational decision-makers and inform the design of future AI-assisted executive systems.
arXiv:2606.17539v1 Announce Type: new Abstract: Spatial VLMs have made substantial progress in geometric perception, yet complex spatial reasoning requiring multi-step inference over depth, distance, and scene relations remains challenging. Moreover, different spatial queries call for fundamentally different strategies: some are best addressed through purely linguistic, step-by-step deduction, while others require explicit 3D grounding before quantitative inference. We present Dual-Path Spatial Reasoning via Reinforcement Learning for Spatial VLMs (SR-REAL), a unified framework that equips a spatial VLM with two complementary reasoning paths: Language-Only Reasoning (LOR), which performs step-by-step linguistic deduction, and Detect-Then-Reason (DTR), which detects 3D geometric cues (e.g., centers or bounding boxes) via region tokens before explicit geometric inference. SR-REAL begins with a cold-start supervised fine-tuning stage that constructs LOR and DTR chain-of-thought supervision and exposes a region-to-3D interface, followed by RL that optimizes the policy model with accuracy and format rewards; for DTR, a discrete center-based detection reward further refines geometric alignment. Across diverse spatial benchmarks, SR-REAL significantly outperforms spatial VLM baselines: (i) a single RL-trained model supports both reasoning paths, with DTR excelling in region-aware tasks through precise 3D localization and LOR enhancing general spatial reasoning; (ii) jointly training both paths fosters mutual reinforcement; (iii) high-quality, blended cold-start data is crucial for stable RL optimization; and (iv) the model generalizes across datasets and domains without per-task tuning, demonstrating positive transfer between LOR and DTR.
arXiv:2606.17744v1 Announce Type: new Abstract: We explore the numerical solution of large-scale non-autonomous Differential Riccati Equations (DREs). While we assume to discretize the differential operator using a Backward Differentiation Formula (BDF) of order s, we solve the generalized Algebraic Riccati Equation (gARE) resulting at each time step by different state-of-the-art methods. In particular, we compare the performance of the inexact Newton- Kleinman method with line search and the low-rank RADI iteration, considering for both methods two different initialization strategies: zero initialization and warm-start. A comprehensive panel of numerical results illustrate the potential and limitations of these methods when employed within a numerical pipeline for the solution of DREs, rather than for the isolated solution of a single gARE, as commonly considered in the existing literature.
arXiv:2602.06257v2 Announce Type: replace Abstract: Online strategic classification studies settings in which agents strategically modify their features to obtain favorable predictions. For example, given a classifier that determines loan approval based on credit scores, applicants may open or close credit cards and bank accounts to obtain a positive prediction. The learning goal is to achieve low mistake or regret bounds despite such behavior. While randomized algorithms have the potential to offer advantages to the learner in strategic settings, they have been largely underexplored. In the realizable setting, no lower bound is known for randomized algorithms, and existing lower bound constructions for deterministic learners can be circumvented by randomization. In the agnostic setting, the best known regret upper bound is $O(T^{3/4}\log^{1/4}T|\mathcal H|)$, which is far from the standard online learning rate of $O(\sqrt{T\log|\mathcal H|})$. In this work, we provide refined bounds for online strategic classification in both settings; our bounds depend on the Littlestone dimension $\mathrm{Ldim}(\mathcal H)$ of the hypothesis class $\mathcal H$ and the maximum degree $\Delta$ of the manipulation graph. In the realizable setting, we extend, for $T > \mathrm{Ldim}(\mathcal H) \Delta^2$, the existing lower bound $\Omega(\mathrm{Ldim}(\mathcal H) \Delta)$ for deterministic learners to all learners. This yields the first lower bound that applies to randomized learners. We then provide the first randomized learner that improves the known (deterministic) upper bound of $O(\mathrm{Ldim}(\mathcal H) \cdot \Delta \log \Delta)$. In the agnostic setting, we give an improper randomized learner that improves the regret upper bound to $O(\sqrt{T\log|\mathcal H|})$, matching the standard online learning rate. We also show a larger lower bound for all proper learning rules, demonstrating that improperness is necessary to achieve the optimal rate.
arXiv:2606.17107v1 Announce Type: new Abstract: Prefix caching reuses prefill only across an exactly shared prefix, so one changed field invalidates the entire downstream cache. Yet overwriting the field's own key/value vectors and reusing the rest leaves the model acting on the old value. The reason, established causally across four model families: at prefill the model has already written the field-conditioned conclusion onto downstream notes; the field's own key/value drives under 1% of the decision. Read as a notebook of memoized conclusions, two capabilities follow. (1) It is editable. A salient erratum amends the notes; and with chain-of-thought, editing the field alone recovers the decision (1.00 at 8B, ~1% compute), while without CoT it is ignored. (2) It is composable. The notes are position-portable, so a precompiled skill can be RoPE-repositioned and spliced into any context, indistinguishable from full recompute (logit cosine 0.90-0.999, twelve models) at O(L) rather than O(L^2) time-to-first-token. A unified edit+compose agent stays decision-identical to recompute at up to 14.9x lower latency. The approach applies to any per-token attention KV cache, validated across scale, quantization, Mixture-of-Experts, and multimodal caches, and extends to several attention variants through small adapters. Because the erratum is append-only, it composes with production prefix caching: in an online vLLM benchmark it keeps the prefix cache-aligned (98.5% hit-rate), cutting p90 time-to-first-token by 53-398x.
arXiv:2606.17630v1 Announce Type: new Abstract: Safe and efficient trajectory planning in unknown, cluttered 3D environments constitutes a critical bottleneck for deploying Unmanned Aerial Vehicles (UAVs) in real-world applications. This challenge is further exacerbated by the limited field-of-view (FOV) and sensing range of onboard sensors. Many existing methods either make simplistic assumptions about unexplored space or rely on conservative heuristics such as speed limits or fixed perception patterns, reducing efficiency and generalizing poorly across different sensor types. In this work, we propose a novel planning framework that directly integrates active perception into trajectory optimization, thereby improving safety while preserving efficiency. The perception constraints are derived from the UAV's dynamic model and formulated in the sensor coordinate frame, which enables precise handling of FOV geometry. The velocity-triggered activation mechanism enables the planner to balance perception and motion efficiency. We introduce an active perception sub-trajectory segment with parametric start-time optimization, mitigating collision risks from late obstacle detection. Our formulation enables active perception during arbitrary 3D maneuvers, extending beyond prior methods designed mainly for horizontal motion. All constraints and penalties are incorporated into a differentiable optimization problem, so the planner requires only a simple front-end global path for guidance, rather than a computationally expensive perception-aware path generator. Extensive simulations and real-world experiments demonstrate robust performance across diverse unknown environments with varying sensor configurations.
arXiv:2606.17160v1 Announce Type: new Abstract: Contrastive language-audio pretraining (CLAP) enables zero-shot audio classification, but standard inference classifies each clip in isolation and ignores the structure of the unlabeled test set. We present the first systematic study of TransCLIP-style transductive inference for CLAP: a text-anchored spherical Gaussian-mixture EM that refines zero-shot posteriors using the audio-embedding statistics of the test batch, with no labels, no gradients, and negligible compute (about 15 ms on one CPU core for 2,000 clips). Across ESC-50, UrbanSound8K, and VocalSound, this consistently improves top-1 accuracy by +4.6 to +9.2 points over the zero-shot baseline (e.g., 89.1 -> 94.8% on ESC-50, 73.8 -> 81.8% on UrbanSound8K). We further show that the gain (i) is governed by a simple operating boundary -- roughly 2.5 test samples per class per batch are required, with diminishing returns beyond ~5; (ii) is complementary to entropy-guided prompt weighting, with the combination reaching 96.2% on ESC-50; and (iii) attenuates but remains positive under long-tailed batches (+4.9 -> +3.1 points at a 20:1 imbalance), which we report as an explicit limitation. We also document a negative result: on TUT Urban Acoustic Scenes 2018, where zero-shot CLAP is near chance, transduction has no signal to amplify.
arXiv:2603.03965v3 Announce Type: replace Abstract: This paper develops an adaptive modular geometric control framework for robotic manipulators with uncertain inertial parameters. The manipulator is decomposed into rigid-body and joint modules, where each rigid-body module is represented by an Euler-Poincar\'e-type spatial dynamics on the Lie algebra se(3), and configuration errors are defined intrinsically through logarithmic maps on SE(3). The joint modules impose local screw constraints that relate adjacent body twists, accelerations, and transmitted wrenches, yielding a recursive propagation structure for the interconnected multibody system. Within this formulation, local geometric control laws are constructed at the module level, while the interconnection among modules is characterized by power-conjugate twist--wrench pairs induced by the natural duality pairing between the Lie algebra se(3) and its dual space se(3)^*. For the nominal case, exponential tracking stability of the interconnected system is established using local configuration energy functions on SE(3) and the power-preserving structure of the modular interconnection. To address inertial parametric uncertainty, a geometric adaptation law is introduced on the manifold of symmetric positive-definite matrices, ensuring physically consistent parameter estimates while retaining compatibility with the Lie-algebraic control formulation. Under the adaptive controller, semi-global uniform ultimate boundedness of the closed-loop tracking and parameter estimation errors is proven. Numerical simulations on a redundant high-inertia robotic manipulator demonstrate accurate pose tracking, smooth transient behavior, orientation regulation, and robustness under inertial perturbations. Comparative studies with state-of-the-art methods further illustrate the effectiveness of the proposed framework for complex robotic manipulation tasks.
arXiv:2606.17174v1 Announce Type: new Abstract: While parasocial interactions (PSIs) and parasocial relationships (PSRs) have been studied in conventional media settings, we investigate whether PSI- (colloquial) relational cues also exist in online communities where both sides are autonomous AI agents. We analyze 4,434 posts and 50,338 comments from Moltbook through three theory-based textual indicators: attachment/intimacy language, reciprocity bids, and self-identification to original poster (OP). The combined results across methods based on keyword matching, few-shot large language model (LLM) annotation, and grouped-context LLM annotation reveal that PSI colloquial cues prevail and are strongly associated with OP re-engagement and a reciprocal reply structure. These results are robust across negative controls, nullification, clustered-standard-error re-estimation, and multiple-testing correction. A dyadic persistence test further affirms reciprocity bids aligned with sustained OP-involving mutual recurrence, providing empirical evidence for bridging interaction-level PSI scripts with PSR-consistent repeated dyadic patterns. We interpret the evidence as a behavioral structure in discourse by LLM-enabled agents.
arXiv:2606.17851v1 Announce Type: new Abstract: A wide range of neurosymbolic (NeSy) systems compute one functional: a belief-weighted sum of a logical quantity over a space of $\sigma$-structures, of which weighted model counting, fuzzy logic, and probabilistic logic are special cases. This account is built on sets, and a set deliberately forgets two things that are important for NeSy: when two $\sigma$-structures are the same up to a symmetry of the theory, and how many distinct proofs witness a query. Replacing the underlying sets by types, in the sense of homotopy type theory, preserves this information, and turns this functional into a belief-weighted homotopy cardinality, a notion of size that counts each object in inverse proportion to its symmetries. We develop the framework from scratch for NeSy systems, prove a conservativity theorem that recovers the classical functional when symmetries are trivial, and show that the symmetry our framework exposes is exactly the one behind reasoning shortcuts. The payoff is concrete: the shortcut-aware concept posterior that recent methods reach by ensembling or expressive density estimation is the only symmetry-invariant point of the confusion-set simplex, computable in closed form by averaging a single model over the symmetry group. On MNIST reasoning-shortcut benchmarks this single-model wrapper is better calibrated than a diversity-trained ensemble, while leaving label accuracy and identifiable concepts untouched. Code is freely available at https://github.com/bio-ontology-research-group/hott-nesy.