arXiv:2607.06083v1 Announce Type: new Abstract: Existing deep learning methods perform well in medical image classification but struggle with multi-scale morphology and limited annotations due to fixed sampling and data-hungry training. Existing approaches address these challenges in isolation: DCN-based models provide adaptive sampling but lack explicit multi-scale attention fusion and label-efficient regularisation; multi-scale architectures typically rely on static fusion; and semi-supervised methods target label scarcity without jointly modelling adaptive cross-scale representations. We propose MSA-DCNN, a scale-consistent deformable attention learning framework that introduces adaptive multi-scale sampling, within-scale saliency refinement, learned cross-scale fusion, and auxiliary self-distillation within a unified optimisation scheme, with potential to generalise to structurally heterogeneous anatomy. We evaluate on three public benchmarks and an external hold-out set for leukaemia. MSA-DCNN demonstrates competitive and often better performance against ViT baselines, CNN baselines, and a MICCAI semi-supervised baseline under distribution shift and label scarcity in accuracy, F1, and AUC (binary), while using fewer parameters. Ablations confirm complementary component contributions, supporting MSA-DCNN as a practical foundation for data-efficient medical image classification.
Science Journals
arXiv:2607.06080v1 Announce Type: new Abstract: Putnam's Social Capital Theory is a foundational framework for collective action and community prosperity. However, traditional empirical methods face practical limits on control and replication. Meanwhile, LLM-based social simulations are typically behavior-driven and lack theory-aligned environments for modeling Putnam's core propositions. To address these gaps, we introduce SocaSim, an LLM-based multi-agent simulation framework to study Putnam's Social Capital Theory from theoretical blueprint to simulated reality. Specifically, we build an environment integrating social network evolution, trust dynamics, and norm propagation, where agents engage in repeated collective-action experiments, and then apply the three dimensions to analyze adaptation challenges in smart elderly care. Our simulations reproduce Putnam's macro-level patterns and exhibit strong human-agent alignment at the group level. Unlike traditional methods, SocaSim traces micro-level causal pathways of social network, trust, and norms via round-by-round simulations and counterfactual interventions, enabling process-level interpretability. Taken together, these capabilities establish a research paradigm that leverages LLM agents to bridge social science and computer science.
arXiv:2607.05958v1 Announce Type: new Abstract: We present a distributed, vendor-agnostic multi-MCP architecture for SDN-based automation and autonomous control of multi-vendor, multi-layer IPoDWDM networks. The framework enables E2E service lifecycle automation, closed-loop cross-layer control using GNPy model and optical telemetry, and is experimentally validated on a IPoDWDM testbed.
arXiv:2607.05983v1 Announce Type: new Abstract: In this paper, we introduce the \emph{Surplus Parking Gathering Problem} ($\mathcal{SPG}$), a new coordination problem for robots deployed on an infinite grid. The input consists of a set of designated parking nodes, each associated with a prescribed capacity, while the total number of robots exceeds the total parking capacity. The objective is to saturate every parking node exactly according to its capacity while gathering all remaining surplus robots at a common grid node that is not specified a priori. The robots are assumed to be autonomous, anonymous, oblivious, identical, disoriented, and homogeneous. We consider the asynchronous (\textsc{async}) model with global visibility and global strong multiplicity detection. We first establish necessary conditions for the solvability of $\mathcal{SPG}$ by characterizing the initial configurations that admit no deterministic distributed algorithm. For all the remaining solvable configurations, we present a deterministic distributed algorithm that correctly solves the problem. The proposed algorithm proceeds in several phases and avoids collisions throughout its execution. We prove that the algorithm terminates in finite time and, upon termination, every parking node is saturated according to its prescribed capacity while all surplus robots are gathered at a uniquely determined gathering node. We further analyze the move complexity of the proposed algorithm, obtaining an $O(n(a+b)+n^2)$ upper bound together with an $\Omega(n(a+b))$ worst-case lower bound for the $\mathcal{SPG}$ problem.
arXiv:2607.05775v1 Announce Type: new Abstract: Large language model (LLM) agents are increasingly evaluated on their ability to use tools, plan multi-step tasks, coordinate with other agents, and operate over extended horizons. Reported benchmark gains often obscure recurring failure modes documented across otherwise unrelated evaluation efforts. This paper synthesizes 27 benchmark, taxonomy, and audit papers (2023-2026), spanning 19 distinct benchmarks, into a cross-cutting taxonomy of agent limitations. To our knowledge, this is the first synthesis that integrates evidence across tool use, planning, long-horizon reasoning, multi-agent coordination, safety, and measurement validity into a single, unified taxonomy of LLM agent limitations. We identify six failure clusters: (1) tool invocation and parameter-level errors, (2) planning and constraint-satisfaction failures, (3) long-horizon degradation from context accumulation, (4) multi-agent coordination failures, (5) safety and security failures under adversarial or underspecified conditions, and (6) measurement validity problems. The taxonomy was derived iteratively by grouping independently reported error categories into themes corresponding to distinct stages of the agent reasoning-to-action pipeline. Across the literature, we find that failures compound nonlinearly with task length, that strong performance on individual sub-tasks does not reliably translate into end-to-end success, and that additional scaffolding does not consistently improve reliability. At the same time, substantial progress has been demonstrated in single-turn tool use, short-horizon web navigation, and narrowly scoped coding tasks.
arXiv:2607.05777v1 Announce Type: new Abstract: Reliable calibration of multi-fisheye camera systems remains challenging as rig size, camera arrangement diversity, and field of view increase. Existing pipelines can jointly optimize intrinsics, extrinsics, and target poses, but their success still depends heavily on empirical capture rules and the quality of the observations supplied to the solver. This paper studies this dependency through a failure-oriented analysis. We reveal that calibration failures are not sufficiently explained by detector recall loss or global image-plane distribution imbalance. Instead, the dominant failure factor lies in intrinsic initialization: observations with limited radial span couple focal scale with fisheye projection-shape parameters, producing ill-conditioned updates. Guided by this insight, we propose CO-Calib, a plug-in calibration-data construction framework that combines a robust learning-based target detector with an error-analysis-guided frame selector. CO-Calib constructs initialization-friendly anchors, co-visible multi-camera constraints, and coverage-completion frames without changing the existing calibration workflow or optimization backend. Extensive experiments on synthetic and real multi-fisheye systems demonstrate that CO-Calib improves the overall success rate from 68.1% to 99.3%, increases extrinsic accuracy, and augments real-world calibration stability. The source code will be made publicly available at https://github.com/HKUST-Aerial-Robotics/CO-Calib.
arXiv:2607.05531v1 Announce Type: new Abstract: Variational Autoencoders (VAEs) frequently suffer from posterior collapse, a failure mode in which the approximate posterior converges to the prior, rendering the latent code uninformative. Despite extensive research, a unified account of why collapse occurs has remained an open question. We identify and formalize two logically independent but coupled causes. \emph{Gradient imbalance} occurs when the decoder's reconstruction signal vanishes faster than the $\mathbb{KL}$ regularization pressure as the posterior widens. \emph{Information gap} occurs when the stochastic sampling step discards a substantial fraction of the encoder's computed representation, attenuating decoder sensitivity and making collapse inexpensive. Both causes share the same collapse trajectory, and we show that the information gap is algebraically equivalent to mismatch between the aggregate posterior and the prior, unifying two pathologies. Subsequently, we introduce $\lambda$-VAE, which resolves both causes through a single modification to the reparameterization step: the sampling noise is scaled by per-dimension exponent, while the $\mathbb{KL}$ penalty retains the original posterior variance. This asymmetry shifts the stable training attractor away from the degenerate collapsed state, driving all latent dimensions toward the same equilibrium -- a mechanism we term \emph{variance equalization}. A closed-form optimal exponent per dimension follows from a net information gain objective, with a single hyperparameter controlling the reconstruction-generation tradeoff. We validate on standard benchmarks (Binary MNIST, Binary Omniglot, CIFAR-10, CelebA-64), showing consistent reductions in collapsed dimensions, information capacity gains of up to $2.8\times$ nats, and reconstruction quality improvements of up to $+0.33$ BPD.
arXiv:2511.21245v3 Announce Type: replace Abstract: Monocular 3D face reconstruction estimates a 3D morphable model (3DMM) representation from a single image, providing geometry-aware expression codes that are useful for facial expression analysis and affect understanding. Despite strong progress, most pipelines are trained with image-level self-supervision and evaluated primarily by geometric fidelity, which does not necessarily maximize the affective utility of the learned expression representation and may encourage intensity-amplifying shortcuts when affect supervision is naively coupled. We propose FIELDS (Face reconstruction with accurate Inference of Expression using Learning with Direct Supervision), a task-driven framework that learns FLAME expression codes for facial expression recognition (FER) under a geometric plausibility constraint. Using hybrid 2D/3D supervision, FIELDS improves affect prediction in both in-domain and external evaluations while maintaining competitive geometric fidelity on held-out and out-of-domain 3D benchmarks.
arXiv:2602.14668v2 Announce Type: replace Abstract: We consider the problem of fair allocation of indivisible goods among agents with additive valuations, aiming for Best-of-Both-Worlds (BoBW) fairness: a distribution over allocations that is ex-ante fair, and additionally, it is supported only on deterministic allocations that are ex-post fair. Existing BoBW algorithms are far from achieving the best possible ex-post fairness guarantees, even in the well studied special case when there are only few agents. We focus on BoBW for few agents, and our main result is the design of the first poly-time BoBW algorithms achieving near-optimal fairness for three agents. We also present optimal poly-time BoBW results for two agents. For three agents, we prove that there exists an ex-ante proportional distribution over at most six allocations, each of which is Epistemic EFX (EEFX) and gives every agent at least $\tfrac{9}{10}$ of her maximin share (MMS). Since MMS allocations need not exist, some agent may fall below her MMS; we guarantee that any such agent is EFX-satisfied -- a new criterion we call "Individually MMS-satisfying or EFX-satisfying (IMMX)". We complement this with an FPTAS preserving all envy-based guarantees, and also preserving all value-based guarantees up to $(1-\varepsilon)$. Furthermore, we present an FPTAS which guarantees exact ex-ante proportionality, while guaranteeing each agent receives $(1-\varepsilon)$ of her MMS or is EFX-satisfied ex-post -- notable, as computing EFX allocations in polynomial time is open even without BoBW constraints. For two agents, we give an FPTAS that is ex-ante envy-free, ex-post EFX, and guarantees each agent $(1-\varepsilon)$ of her MMS, matching the strongest guarantees achievable in polynomial time.
arXiv:2502.00181v2 Announce Type: replace Abstract: Muon colliders offer a compelling opportunity to explore the TeV scale and conduct precision tests of the Standard Model, all within a relatively compact geographical footprint. This paper introduces a new detector concept, MAIA (Muon Accelerator Instrumented Apparatus), optimized for $\sqrt{s}=10$ TeV $\mu^+ \mu^-$ collisions. The detector features an all-silicon tracker immersed in a 5T solenoid field. High-granularity silicon-tungsten and iron-scintillator calorimeters surrounding the solenoid capture high-energy electronic and hadronic showers, respectively, and support particle-flow reconstruction. The outermost subsystem comprises an air-gap muon spectrometer, which contributes to muon identification. The performance of the MAIA detector is evaluated in terms of differential particle reconstruction efficiencies and resolutions. Beam-induced background and incoherent pair production simulations are overlaid to single particle gun samples to assess detector reconstruction capabilities under realistic experimental conditions. Even in the presence of backgrounds, reconstruction efficiencies exceed approximately 95\% for energetic tracks, photons, and charged pions in the central region of the detector. This paper outlines promising avenues for future work, including forward region optimization, opportunities for enhanced flavor tagging and boosted object reconstruction, and technological developments needed to achieve the desired detector performance.
arXiv:2601.06441v2 Announce Type: replace Abstract: Learning activation functions has emerged as a promising direction in deep learning, allowing networks to adapt activation mechanisms to task-specific demands. In this work, we introduce a novel framework that employs the Gumbel-Softmax trick to enable discrete yet differentiable selection among a predefined set of activation functions during training. Our method dynamically learns the optimal activation function independently of the input, thereby enhancing both predictive accuracy and architectural flexibility. Experiments on synthetic datasets show that our model consistently selects the most suitable activation function, underscoring its effectiveness. These results connect theoretical advances with practical utility, paving the way for more adaptive and modular neural architectures in complex learning scenarios.
arXiv:2605.13380v2 Announce Type: replace Abstract: This work compares three interaction modalities for human-robot collaboration: passive, reactive, and proactive. We studied 18 participants assembling a seven-layer colored tower from memory while using nearby and distant blocks. In the passive modality participants worked alone; in the reactive modality a mobile robot helped only upon request; in the proactive modality it initiated brick delivery and error signaling without explicit requests. Although robot assistance increased completion time, most participants preferred collaboration: 67% preferred proactive behavior and 78% judged it most useful. These results suggest that timely proactive support can improve user experience in controlled collaborative tasks.
arXiv:2603.18846v3 Announce Type: replace Abstract: Foundation models are used to extract transferable representations from large amounts of unlabeled data, typically via self-supervised learning (SSL). However, many of these models rely on architectures that offer limited interpretability, a critical issue in high-stakes domains such as medical imaging. We propose \model, a foundation model that is interpretable-by-design via a BagNet backbone whose small receptive fields generate class evidence maps that are faithful to the model's decision-making process. Additionally, \model{} incorporates a $2D$ projection layer during pretraining that enables direct visualization of the representation space, providing a dataset-level view of the learned structure including meaningful clinical clusters as well as potential spurious correlations. We trained \model{} on over 800,000 color fundus photographs from various sources to learn generalizable representations for different downstream tasks. Our model achieves performance comparable to RETFound, which has $16\times$ more parameters, while providing interpretable predictions on out-of-distribution data. These results suggest that large-scale SSL pretraining paired with inherent interpretability can lead to robust representations for retinal imaging. Code and pretrained models are available at \href{https://anonymous.4open.science/r/dual-ifm-3D5A/README.md}{www.anonymous.4open.science/dual-IFM}.
arXiv:2607.06125v1 Announce Type: new Abstract: Neural decompilation is increasingly studied as a code-generation problem, yet its evaluation methodology remains underdeveloped for modern languages. We present a systematic empirical study of fine-tuning effectiveness and metric validity for Dart Ahead-of-Time (AOT) neural decompilation. We evaluate six fine-tuned model variants across three base architectures (4B-8B parameters) using three metrics: CodeBLEU, compile@k, and pass@k on a new 154-task HumanEval-Dart benchmark. Our study yields three principal findings grounded in paired task-level statistical tests. First, no fine-tuning configuration produces a statistically significant pass@k improvement. The sole positive case yields +0.71 pp (McNemar p=0.21), while fine-tuning the strongest base (Qwen3-8B) causes a highly significant regression of -5.65 pp (p<0.001). This capacity-dependent trend is consistent across architectures but needs broader scale sweeps. Second, cross-lingual interference from Swift training is highly significant at 4B (-2.66 pp, p<0.001) but statistically indistinguishable from zero at 8B, consistent with the scaling hypothesis. Third, we demonstrate metric divergence: CodeBLEU and compile@k can improve significantly while pass@k moves in the opposite direction. This has implications for any LLM code generation task where fine-tuning targets superficial similarity. Error analysis reveals assembly sequence length is the strongest predictor of task difficulty (p=0.001), with a capability cliff at 200 instructions. We contribute the HumanEval-Dart benchmark, a Dart-adapted CodeBLEU, and empirical evidence that pass@k must be the primary evaluation metric for neural decompilation.
arXiv:2607.06495v1 Announce Type: new Abstract: Live sports commentary is grounded generation under a deadline: statements concern real, named athletes, the grounding state changes every few seconds, and no reference text exists at generation time. We present Pitwall, a production system that generates natural-language Formula 1 strategy briefings in English, Spanish, and Portuguese, treating faithfulness as an architectural property rather than an aspiration: every published sentence is decomposed into typed factual claims (positions, gaps, tyres, pace, overtakes, race control) and each claim is verified against the probabilistic race state that prompted it. The same verifier gates the fine-tuning data: of 3,045 model-written targets, only the 81.9% whose every claim is state-supported are retained, the rest falling back to a provably faithful template, so the generator never sees an ungrounded target. Verification is meaningful because of the grounding substrate: a vectorized Monte Carlo engine (N=2,000 per-lap race continuations) calibrated on 126 races (2018-2024) and validated on fully held-out 2025-2026 seasons (winner-in-top-3 90.3% over 155 backtests; held-out Brier 0.0745). A recurring finding spans both halves of the system: virtues trade off and must be gated separately. In simulation, calibration-optimal is not decision-optimal; in generation, fine-tuning on richer targets buys vividness that collapses into hallucination when the grounding state is sparse -- a failure a four-base replication traces to base-model instruction adherence, not scale, and that sparse-context auditing removes from the production model. End-to-end operation -- live timing to verified trilingual briefings -- was confirmed at two consecutive live Grands Prix (Austria and Britain, 2026); at Silverstone a timestamped probability trace, committed to disk before the outcome was known, locked onto the eventual winner ten laps before the flag.
arXiv:2607.06134v1 Announce Type: new Abstract: Google Safe Browsing (GSB) and DNS resolution operate concurrently during browser navigation, yet their packet-level synchronization remains understudied. This work characterizes the timing gap (\(\Delta_{time}\)) between GSB-related query close events and parallel DNS resolution responses, identifying a consistent temporal offset with potential security relevance. Using packet-capture analysis across general and CNAME-domain datasets, we observe positive timing gaps in approximately 79\% of measurements. In these instances, DNS responses lag behind GSB-related query closures with median delays of 67-79 ms and maximum delays surpassing 2,400 ms. These empirical results highlight a measurable window within the browsing workflow. We suggest that such temporal inconsistencies, particularly in complex CNAME-domain resolutions, may create a security-relevant timing precondition under DNS-manipulation threat models. These results provide a foundation for further research into timing-based risks and mitigations in browser safety mechanisms.
arXiv:2607.06132v1 Announce Type: new Abstract: Multi-Pool Chemical Exchange Saturation Transfer (CEST) MRI provides valuable metabolic information but is clinically limited by long acquisition times. Although sparse sampling reduces scanning time, reconstructing high-resolution Z-spectra from limited data remains an ill-posed inverse problem. Conventional interpolation and generic Implicit Neural Rep-resentations (INRs) often lack physical constraints, leading to spectral artifacts and physically invalid signals. To address this, we propose Lorentz Encoding (LE), a physics-informed framework that formulates CEST reconstruction as a self-supervised reconstruction task via implicit continuous coordinate learning. Unlike generic positional encodings, LE regularizes the continuous spectral mapping by projecting sparse coordinates into a physically constrained space governed by a combination of parametric Lorentzian profiles with learnable basis functions. This mechanism effectively reduces noise and enforces consistency with physical models. Experiments on in vivo human brain data demonstrate that LE significantly outperforms state-of-the-art methods. Specifically, under a 39-point sampling strategy, LE achieves a PSNR of 57.58 dB and an SSIM of 0.9994. Furthermore, the learned physics-informed encodings form a continuous, geometrically ordered trajectory in the latent space, ensuring accurate quantitative metabo-lite mapping (APT, NOE, MT).
arXiv:2607.06240v1 Announce Type: new Abstract: A drawing of a graph is {\em $x$-monotone} if every vertical line intersects each edge of the graph at most once. We present an $O(n^5)$ time algorithm for deciding whether a simple drawing of the complete graph $K_n$ is weakly isomorphic to an $x$-monotone drawing. We note that this algorithm can also decide whether a drawing of $K_n$ is strongly isomorphic to an $x$-monotone drawing.
arXiv:2607.06175v1 Announce Type: new Abstract: Large language models (LLMs) can generate BPMN process models from natural-language descriptions, yet supervised fine-tuning (SFT) limits their output quality to the patterns present in the training data. Reinforcement learning (RL) can optimize beyond this ceiling using external quality measures, but how the reward function should be designed when quality is multi-dimensional remains unexplored. We present a systematic investigation of reward function design for RL-based process model generation, training two LLM families (Llama~3.1 8B, Qwen~2.5 14B) under 48 configurations using Group Sequence Policy Optimization with rewards derived from an automated evaluation framework comprising 38 metrics across syntactic, pragmatic, and semantic quality. Three findings emerge. First, RL significantly improves pragmatic and syntactic quality while preserving semantic fidelity, reducing output variability by more than sixfold. Second, equal reward weighting consistently outperforms targeted weighting: emphasizing a specific dimension fails to improve it and can collapse the model into a low-quality mode. Third, design choices interact with model architecture in non-trivial ways: the invalidity penalty is essential for one model but irrelevant for the other, and SFT initialization is indispensable for one architecture but counterproductive for another. These results demonstrate that reward composition is a primary determinant of optimization outcomes, with effects as large as the decision to apply RL itself. The findings generalize to any structured generation task where quality is assessed along multiple automated dimensions. We release our implementation and experimental code at https://github.com/chlauer99/RL_for_process_modeling.
arXiv:2607.05563v1 Announce Type: new Abstract: Interpretable explanation methods in Artificial Intelligence aim to uncover the underlying causes and their effects, enabling a deeper understanding of why a system behaves in a certain way under different inputs. Unlike traditional explainability methods, which mainly highlight correlations between input and output variables, causal explanation focuses on interventional questions. By doing so, it provides more robust insights, helping users understand automated decisions, especially in high-risk domains. Recovering an explicit directed causal structure, however, is often impractical in large-scale, hybrid cyber-physical systems with feedback loops and partial observability. This paper introduces a novel framework inspired by statistical mechanics that instead models variable dependencies through an undirected, energy-based representation of cyber-physical IoT systems. Our approach enables rigorous dependency-aware attribution by analysing how variations in the energy landscape reflect the influence of individual components, without recovering a directed causal graph. It also supports reasoning about perturbation effects across hybrid interactions, providing reliable explanations of abnormal behaviours. We empirically examined our framework through simulations on an industrial IoT testbed with hybrid continuous and discrete variables, demonstrating higher attribution accuracy, improved robustness and better scalability than state-of-the-art graph-based approaches. While the attributions are not intended to fully recover the system's generative dynamics, they provide valuable, dependency-aware explanations supporting both human interpretation and downstream predictive and diagnostic tasks. Although demonstrated in industrial IoT security, our framework also applies to other high-dimensional cyber-physical and socio-technical systems requiring principled, structural explanations.
arXiv:2607.05749v1 Announce Type: new Abstract: The support structure is an integral part of the design of nuclear fusion reactors, especially 3D stellarator devices. A practical reactor's coils and support structures must have three competing qualities: an accurate magnetic field for good confinement, sufficient rigidity to protect the brittle high-temperature superconductor (HTS) from damage, and a simple geometry for low-cost construction. In existing devices, the coil geometry is often optimized without knowledge of the support structures' design and the coils' true stress and deformation. The support structures are then placed by hand through repeated finite element analyses (FEA) until engineering requirements are met. This makes the structural design of stellarator coil systems lengthy and labor-intensive. Using new developments in differentiable structural mechanics, we present coil-fem, an open-source software tool that integrates support differentiable FEA into the stellarator coil optimization loop. It enables the integrated optimization of coil geometry and support clamp locations to simultaneously reduce magnetic field errors and stresses in the coil body. We also present the first combined coil-support optimization in the stellarator literature. Using a penalty term based on coil-fem, we produced a coil set with 2.4x lower RMS von Mises stress and similar field error compared to an unoptimized baseline.
arXiv:2607.06155v1 Announce Type: new Abstract: Modern sequence models are increasingly deployed as agents that interleave token generation with calls to external tools. We give an exact, architecture-level account of when such tool access increases computational expressivity. We model any fixed finite-precision recurrent sequence model, including finite-precision state-space models (SSMs) with $B$ bits of internal state, as a deterministic finite-state controller interacting with an oracle through a finite command/observation interface. Our results form a sharp dichotomy. First, tools that are themselves finite-state add essentially nothing: a product-state simulation internalizes any finite-state bounded-interface oracle with finite memory set $M$ at a cost of only $\log_2 |M| + O(1)$ additional bits, so the augmented system remains finite-state. Second, a single minimal infinite-state tool, namely a tape supporting only local $\mathtt{read}$, $\mathtt{write}$, and $\mathtt{move}$ commands, makes the system Turing complete: for every single-tape Turing machine with state set $Q$ and tape alphabet $\Gamma$, a controller with $O(\log |Q| + \log |\Gamma|)$ bits of internal memory simulates it, and we exhibit a concrete exponential separation: $\mathrm{EQ}_n$ requires $2^n$ states without tools but a single constant-size controller with the tape tool. Third, we show that this construction is realized exactly by a natural one-layer finite-precision selective affine SSM controller with binary one-hot hidden states, $\{0,1\}$ transition matrices, and zero biases. Selectivity is essential to the construction. In the supplementary material, we make all constants explicit, prove a logarithmic oracle-assisted universal simulation, where $O(\log B)$ recurrent bits suffice to simulate any $B$-state Turing machine, and prove a matching impossibility result.
arXiv:2607.06165v1 Announce Type: new Abstract: Omni-directional (360{\deg}) cameras provide embodied agents with a holistic view of their surroundings, making them suited for directional reasoning in tasks such as navigation and object search. Existing Vision Language Models (VLMs) project 360{\deg} observations to 2D equirectangular projection (ERP) images and process them using architectures designed for perspective images. However, they ignore the spherical nature of 360{\deg} observations, where each pixel represents a viewing direction relative to the agent. Consequently, their direction estimates often become inconsistent under camera view transformations caused by agent motion. This limitation is particularly critical for map-free navigation, where the agent must continuously estimate the target direction in its egocentric frame. We propose EAGOR, a training-free, geometry-aware framework for embodied 360{\deg} directional reasoning. Instead of predicting target directions as ERP image coordinates, EAGOR formulates directional reasoning as recursive Bayesian estimation directly on the sphere. It maintains a continuous belief over target directions and propagates it equivariantly under agent motion without training the backbone VLMs. To achieve this, we introduce the Spherical Harmonic Belief Field (SH-BF), whose spherical harmonic representation provides a globally defined, rotation-aware basis for directional estimation on the spherical manifold. This formulation eliminates ERP seam discontinuities, latitude distortions, and interpolation errors. We evaluate EAGOR on two benchmark datasets and real-world experiments with a legged robot across directional reasoning tasks. EAGOR consistently outperforms existing methods, achieving average relative gains of +34.4% and +45.6% on HOS and OSR-Bench, respectively, while improving navigation success by +14.6%, reducing step count by 17.7%, and lowering mean angular error by 24.5%.
arXiv:2607.05917v1 Announce Type: new Abstract: We develop an analytical framework to constrain Yukawa-type fifth-force interactions using the Lamb shift in hydrogen and deuterium. The sixfold coordinate-space integral for the energy shift is transformed into a one-dimensional momentum-space convolution of the atomic form factor and the Yukawa potential, eliminating the need for approximations. This formulation directly incorporates empirical nuclear charge distributions through their form factors, making the implementation of experimental results straightforward. Applying this method to the latest hydrogen and deuterium spectroscopy data, we derive constraints on the coupling constant $\alpha$ in the interaction range $\lambda \in [10^{-4}, 10^{6}]$~fm. A resonance-like cancellation between the $2s$ and $2p$ orbital shifts is identified at a critical range $\lambda \sim 10^{5}$~fm, where the energy shift vanishes and the constraint on $\alpha$ diverges, marking a transition between attractive and repulsive regimes. The constraints are shown to be independent of the nuclear charge distribution model. Our results indicate that the transition frequency shifts induced by such new forces lie well below current spectroscopic resolution, establishing spectroscopic precision as the primary bottleneck for probing these interactions.
arXiv:2607.05937v1 Announce Type: new Abstract: Sentiment analysis with frozen pre-trained language model (PLM) backbones has become a common paradigm, yet the practical benefit of explicit domain adaptation remains unclear, particularly when backbones encode varying degrees of target-domain knowledge. We present a preliminary case study evaluating a controlled family of frozen embedding backbones (Qwen3-Embedding 0.6B, 4B, 8B), alongside RoBERTa-base and FinBERT. We train a lightweight MLP adapter on consumer reviews using Domain-Adversarial Neural Networks (DANN), Maximum Mean Discrepancy (MMD), and Supervised Contrastive Learning (SCL), and evaluate transfer to movie reviews (SST-2) and a heavily restricted subset of financial news (Financial PhraseBank). Within this constrained sample, we observe two distinct transfer patterns. On SST-2, domain adaptation provides negligible gain regardless of scale. On the financial subset, explicit domain adaptation appears to recover substantial performance for small general-purpose backbones. Notably, we find that adversarial alignment (DANN) is associated with degraded performance for domain-specialized backbones like FinBERT, consistent with erosion of pre-existing domain-specific structure, whereas supervised contrastive loss appears to preserve it. These preliminary findings suggest that the efficacy of explicit domain adaptation is highly contingent on whether the frozen backbone already possesses target-domain coverage.