Artificial Intelligence
Right Words, Wrong Moment: A Clinician-Grounded Analysis of Distress in 19,930 Conversations between Young People and ChatGPT
Marx Wang, Ella Zhang, Cameron Tan et al. (+11 more)
Young people increasingly turn to General-Purpose Conversational
Agents (GPCAs), such as ChatGPT, in moments of distress. We examine young adults' (ages 18-25) experiences using ChatGPT. We first collected 19,930 ChatGPT conversations and survey data from 158 young adults. We then selected five example conversations reflecting user distress. Finally, we asked ten clinicians to review those five conversations. We found distressed participants reported greater emotional engagement with ChatGPT and greater behavioral change from using it than their peers. When they turned to ChatGPT in moments of acute distress, ChatGPT was quick to give overly dramatic responses and excessive action-oriented suggestions. Clinicians endorsed ChatGPT's availability and much of its wording, but identified seven process failures, such as prematurely jumping to solutions. We translated clinicians' feedback into design guidelines following three stages: 1) asking about safety, 2) de-escalating intensity to restore emotional regulation, and 3) exploring concerns without agreeing with them.
Read the full abstract ▾
All authors Marx Wang, Ella Zhang, Cameron Tan, Andrea Mock, Songling Ngo, Zijing Wang, Robert Wolfe, Shirin Amouei, Rachel A. Hanebutt, Desmond C. Ong, Caroline Figueroa, Katie Davis, Anind K. Dey, Alexis Hiniker
Sound
HEAR: Real Voices, Real Bias: A Large-Scale Human-Recorded, Demographically Diverse Benchmark for Audio Language Models
Shen Yan, Duc Le and Irina-Elena Veliche
We introduce HEAR (Human-recorded Evaluation of Audio-LLM bias by Real speakers), a large-scale, ecologically valid
benchmark comprising 87k real human audio samples from 843 demographically diverse participants. HEAR enables comprehensive evaluation through Multiple Choice Question Answering (MCQA) and open-ended long-form tasks. To our knowledge, this is the first large-scale voice
benchmark grounded entirely in authentic human
speech. We evaluate model behavior across both real-time
speech-to-
speech and
speech-to-text architectures. Our results reveal that voice-conditioned bias is a model-specific property. Furthermore, we demonstrate that personalization instructions consistently exacerbate demographic disparities. Our findings establish that voice bias is a controllable model characteristic, providing a foundational framework for future bias mitigation and evaluation in Audio-LLM development.
Read the full abstract ▾
Cryptography and Security
When Privacy Becomes a Weapon: Understanding Doxxing and Privacy Vulnerabilities in Mainland China's Social Media Ecosystem
Xiao Zhan, Shijing He, Chi Zhang et al. (+1 more)
Doxxing, the malicious disclosure of personal information, has become a pervasive
privacy threat. Yet existing research remains predominantly Western-centric, limiting our understanding of how doxxing unfolds in contexts where mandatory identity systems, platform governance, and cultural logics fundamentally reshape
privacy risks and harm trajectories. We address this gap through semi-structured interviews with 18 doxxing survivors in mainland China, synthesizing their experiences into a framework conceptualizing how doxxing operates in this context. Our findings reveal both patterns echoing prior Western findings, such as platform amplification mechanisms that resonate with Western findings, and China-specific dynamics shaped by the interplay of regulatory mandates (compulsory identity linkage) and cultural logics including nationalist discourse, fandom culture, Confucian values, and low
privacy literacy. Survivors' experiences further reveal how doxxing reshapes understanding of
privacy: from preference to precondition, from momentary disclosure to temporal vulnerability, and from individual control to structural powerlessness. These insights challenge agency-centered
privacy frameworks and suggest that effective protection requires constraining systemic vulnerabilities rather than relying solely on user empowerment. We conclude by proposing multifaceted recommendations spanning legal reform, platform design, and social initiatives.
Read the full abstract ▾
All authors Xiao Zhan, Shijing He, Chi Zhang, Jose Such
Number Theory
Primes in arithmetic progressions and Siegel zeroes
Thomas Wright
Let $\chi$ be a Dirichlet character mod $D$ with $L(s,\chi)$ its associated $L$-function, and let $\psi(x,q,a)$ be, as usual, Chebyshev's
prime-counting function for the
primes of the arithmetic progression $a$ (mod $q$) with $(a,q)=1$. Let $\chi$ be a primitive character modulo $D$, and let $\nu>0$ be small. We prove that if $L(s,\chi)$ has a Siegel zero at $s=\beta=1-\frac{1}{\eta\log D}$ with $\eta>\eta_0(\nu)$ for some large $\eta_0(\nu)$, there exists a range of $x$ for which the asymptotic $$\psi(x,q,a)=\frac{\psi(x)}{\phi(q)}\left[1+O(\varepsilon_{\eta_0})-\chi\left(\frac{aD}{(q,D)}\right)\right]$$ holds for $q<x^{\frac{30}{59}-\nu}$. We also show slightly better bounds for $q$ if we take an average over a range of $q$, finding an Elliott-Halberstam-type result for $q\sim Q$ on the range $Q<x^{\frac{16}{31}-\nu}$. This improves on a 2003 result of Friedlander and Iwaniec that requires $q<x^{\frac{233}{462}}$ and builds on recent work of Sachpazis.
Read the full abstract ▾
Atmospheric and Oceanic Physics
A neural network-based Universal Thermal Climate Index for reliable global thermal-stress classification across extreme weather
Bikem Pastine, Milan Klöwer, Tianning Tang et al. (+2 more)
Extreme temperatures are the leading cause of
climate-related mortality world-wide.
Climate-health research and operational weather
forecasting require accurate estimates of human thermal stress. The Universal Thermal
Climate Index (UTCI) is among the most sophisticated and widely used feels-like temperature metrics. However, its ubiquitous
polynomial approximation does not generalize well to extreme weather conditions. Here, we introduce Neural-UTCI, a neural network that calculates UTCI with substantially higher accuracy across global conditions at a lower computational cost for operational use. Neural-UTCI reduces the
polynomial approximation RMSE from 2.78{\deg}C to 0.36 {\deg}C, an 87% improvement, and lowers thermal stress misclassification rates from 5.3% to 1.7%, with consistent performance across resampling experiments. These differences affect thermal exposure metrics. For example, during the 2003 European heatwave summer in Rome, Italy, the number of very strong heat stress days increases from 15 to 35 days when using Neural-UTCI compared to operational products like ERA5-HEAT. Simultaneously, Neural-UTCI reliably classifies extreme cold stress conditions, allowing continuous global application. By improving UTCI accuracy, Neural-UTCI can strengthen
climate-health risk assessments and public weather warning systems, especially as global warming increases the incidence of extreme events.
Read the full abstract ▾
All authors Bikem Pastine, Milan Klöwer, Tianning Tang, Sarah Wilson Kemsley, Louise Slater
Machine Learning
Intrinsic Associative Memory on Riemannian Manifolds: Curvature, Capacity, and Emergent Modes
Krishnakumar Balasubramanian and Zhaoyang Shi
Geometry does more than constrain an associative
memory:
curvature determines what it remembers and which states it creates. We develop intrinsic dense associative
memories on Riemannian
manifolds by casting
memory as Epanechnikov kernel-density mode seeking. We compare geodesic and volume-corrected energies and show that
curvature separates their behavior. We prove that geodesic
memory always retains an isolated pattern, while corrected
memory obeys a sharp
Ricci-
curvature threshold: positive
curvature can erase
memories in high dimensions, while negative
curvature reinforces them. We derive geodesic capacity scalings of $q_\beta^{-1/2}$ for retaining every pattern and $q_\beta^{-1}$ for a typical one, where $q_\beta$ is the pairwise kernel-overlap probability. We show how overlap creates novel
memories: designed $N$-pattern configurations realize all $2^N-1$ subset modes, but random data at the storage threshold yield only a Poisson number. We establish exact one-step recall using Riemannian mean shift. In simulations, we recover the predicted
curvature transition and every designed mode. On WordNet's full noun hierarchy, we demonstrate that volume correction improves low-capacity
retrieval. Together, our work shows that
curvature is a design variable for associative
memory, not merely a property of the data.
Read the full abstract ▾
Machine Learning
FluxLite: Inference-Time Proposal Control for Discrete Diffusion Models
Yinuo Ren, Haoxuan Chen, Grant M. Rotskoff et al. (+2 more)
Many inference-time tasks for pretrained discrete
diffusion models and
diffusion language models reduce to drawing samples from a tilted version of the pretrained distribution. Feynman-Kac sequential Monte Carlo (SMC) makes this correction exact in principle, but its prescribed weights routinely degenerate when the proposal dynamics are misaligned with the tilt, capping the practical gains from additional particles. We introduce FluxLite, a lightweight, training-free proposal-control framework for discrete
diffusion. On the sparse directed graph of pretrained reverse rates, any sparse jump-rate perturbation can be exactly compensated by a $q_t$-weighted graph-divergence term in the Feynman-Kac potential; the target path is therefore preserved while the residual reweighting variance becomes a local convex objective. We instantiate this principle as two practical samplers: a one-hop local reallocation rule (HEU) and a small nonnegative quadratic program over pretrained-rate bases (D-VCG). We further prove population stability under the standard score-
entropy training loss, identifying a tilted-path coverage factor that governs robustness to score error, together with finite-particle convergence for a fixed controlled Feynman-Kac recursion. Empirically, FluxLite improves over standard Feynman-Kac SMC baselines by up to two orders of magnitude in terminal KL on an analytically tractable finite-state CTMC
benchmark, and reduces row-correlation MSE on 2D Ising sampling by 5-7x in
geometric mean and up to 55x at peak.
Read the full abstract ▾
All authors Yinuo Ren, Haoxuan Chen, Grant M. Rotskoff, Jiequn Han, Lexing Ying
Optimization and Control
Gradient-Free Methods for Stochastic Convex Optimization with Stochastic Functional Constraints
Vadim Abronin, Alexander Gasnikov and Darina Dvinskikh
We develop accelerated gradient-free methods for
stochastic convex optimization with constraints defined by expectations. Our batched primal-dual sliding method uses two-point evaluations sharing a random sample and guarantees expected objective error and expected maximum constraint violation at most $\varepsilon$. It achieves $O(\varepsilon^{-1/2})$ sequential oracle rounds for smooth data and $O(d^{1/4}/\varepsilon)$ for nonsmooth data in dimension $d$, recovering the accuracy and dimension dependence of the corresponding unconstrained accelerated methods. The smooth rate is optimal in accuracy. Each round collects all samples needed for its inner primal-dual updates. Total evaluations retain quadratic dependence on inverse accuracy, with only polylogarithmic dependence on the number of constraints under vector feedback. Complementary lower bounds distinguish the dimension cost of gradient estimation from the unavoidable logarithmic cost of estimating noisy constraint levels. For smooth strongly convex objectives, restarts give logarithmic round complexity and objective evaluation cost linear in inverse accuracy, while constraint-level estimation necessarily remains quadratic. Known affine constraints require no constraint-oracle calls.
Read the full abstract ▾
Astrophysics of Galaxies
GA-NIFS view of LAP1: Assembly of Pop~III stellar clusters during EoR
J. Scholtz, R. Maiolino, S. Carniani et al. (+28 more)
Lensed and Pristine 1 (LAP1) is one of the most promising candidates for the detection of so-called Pop~III
stars at $z\sim6.6$. In this work, we present new high-spectral-resolution ($R \sim 2700$) \jwst NIRSpec-IFS observations of the system along with a detailed re-analysis of all archival \jwst and VLT/MUSE data. We confirm the very low
metallicity of the system, with $12+\log(\text{O/H})=6.5\text{--}6.8$ (1–2% Z$_{\odot}$), and its low
stellar mass, with $M_{*}= 10^{3\text{--}4}$ \Msun, based on the non-detection in \jwst/NIRCam filter stacks and the \Hbeta luminosity. The high-resolution IFS data allow us to constrain the upper limits on the dynamical masses ($\log_{10}(M_{*}/M_{\odot})$) of the A and B components to $<6.6\text{--}7.6$ and $<5.6\text{--}6.6$, respectively. Based on the
metallicity and the high equivalent width of \Halpha (EW $>1200$\,Å), we estimate the possible range of the LAP1
stellar population ages to be $2\text{--}5$~Myr, consistent with the first burst of
star formation in a pristine environment. Based on the size ($<10$~pc) and mass of the system, we discuss the nature of LAP1 as representing the formation of the first
stellar clusters in a larger undetected
galaxy, similar to other low-
metallicity Pop~III candidates in the literature.
Read the full abstract ▾
All authors J. Scholtz, R. Maiolino, S. Carniani, E. Vanzella, R. J. Terlevich, H. Übler, M. Perna, S. Arribas, E. Bertola, P. Bergamini, T. Böker, A. Bolamperti, A. Bunker, S. Charlot, M. Curti, F. D'Eugenio, C. Grillo, L. R. Ivey, X. Ji, G. C. Jones, M. Meneghetti, M. M. Messa, R. Pascalau, B. Rodríguez Del Pino, P. Rosati, E. Rusta, S. Salvadori, E. Terlevich, G. Venturi, S. Zamora, A. Zanella
Machine Learning
EnJoi: Ensemble Joint Score Filter for Generative Data Assimilation
Julien Moreau and Marc Lelarge
Data Assimilation (DA) aims to recover the full state of a dynamical system that is only partially observed. A solution is to use Score-based models to generate physically consistent trajectories that agree with the observations. These Autoregressive
Diffusion models are trained by conditioning on the previous state; however, they do not take into account the uncertainty of their past predictions. We propose a new
diffusion-based assimilation
algorithm that dynamically balances the confidence in the current state and the new observations. Crucially, we choose to learn the distribution of the joint state containing both the past and future. This allows us to use a modified version of En4DVar, a classical DA
algorithm that relies on the covariance of an ensemble of particles. Experiments on fluid and traffic flow simulations show improved reconstruction performance, especially in situations where observations are sparse and non-homogeneous.
Read the full abstract ▾
Computer Vision
HERO: Histology Encoder for Robust Representation in Oncology
Zhi Li, Eghbal Amidi, Yating Cheng et al. (+11 more)
Foundation models trained on large pathology image corpora now provide strong, transferable representations for computational pathology. Over the past few years a series of such models has been released, each trained on more slides than the last; on standard classification and
segmentation benchmarks, the leading models are now separated by small margins. In clinical use, however, the foundation model is applied to images from hospitals, scanners, and staining protocols outside its training data. Encoders generally embed these acquisition factors alongside biological information, which may introduce downstream errors and hinder safe clinical adoption. A pathology foundation model should therefore be robust to acquisition shift without giving up representation quality, yet robustness is seldom the axis along which models are compared. In this report, we introduce HERO (Histology Encoder for Robust Representation in Oncology), a ViT-G/14 pathology foundation model trained with the DINO and iBOT objectives and refined with high-resolution Gram anchoring on a morphology-balanced corpus of 500 million tiles from approximately 575,000 clinical whole-slide images. Across the evaluated public
benchmarks, HERO shows the strongest robustness to center, scanner, and stain variation among the compared state-of-the-art foundation models, performs comparably on tile-level classification,
segmentation, and gene-expression prediction, ranks first on average across 39 evaluated slide-level clinical tasks, and, under an equal-weighted framework-level analysis, has the best average rank across the six
benchmark frameworks.
Read the full abstract ▾
All authors Zhi Li, Eghbal Amidi, Yating Cheng, Tyson Dawson, Gorkem Can Ates, Shuzhen Kuang, Norsang Lama, Md Ashequr Rahman, Zhiying Lu, Elisabeth K. Kong, Milan Radovich, David Spetzler, George W. Sledge, Ming Chen
Computation and Language
Question-Specific Knowledge Graphs for Efficient Visual Reasoning
Ting-Chih Chen, Emile van Krieken, Shujian Yu et al. (+1 more)
Recent work in visual question answering has shown that
vision-
language models can exhibit strong
reasoning capabilities by translating visual inputs into textual representations. The effectiveness of this translation depends on how well visual details are retained; models need to surface and align both explicit and implicit knowledge sufficient to support
reasoning, without introducing spurious assumptions. Existing methods that leverage detailed image captions introduce visual details unrelated to the
reasoning task, inflating input
token counts and increasing computational cost. To address these challenges, we propose VisKG, a reinforcement learning (RL) framework in which models learn to translate visual content into question-specific knowledge graph (KG) representations. This process filters out perceptual noise while preserving the entity-relation structure needed for chain-of-thought
reasoning, following the principle of minimum sufficient information. To ensure stable RL post-training, VisKG adopts Group reward-Decoupled Normalization Policy Optimization (GDPO). In addition, we strengthen the supervision stage with negative rationale samples, exposing the model to incorrect
reasoning paths before RL post-training. Experimental results across science, mathematics, and general visual understanding
benchmarks show that VisKG achieves performance comparable to or better than baselines, while requiring fewer
tokens than caption-based representations. Moreover, training VisKG with GDPO improves accuracy by 2% over its GRPO-trained counterpart on average. These results suggest that KG representations are a promising approach for supporting multi-step
reasoning and open up future work on adaptively selecting the most suitable representation for a given task.
Read the full abstract ▾
All authors Ting-Chih Chen, Emile van Krieken, Shujian Yu, Filip Ilievski
Astrophysics of Galaxies
Massive Population III Galaxies Near a GN-z11 Analog: The Role of the Baryon-Dark Matter Streaming Velocity
Eli Visbal, Greg L. Bryan and Ryan Hazlett
Recently the James Webb Space
Telescope (JWST) detected Hebe, a Population III (Pop III) candidate located ${\sim}3$ kpc in projection from the massive $z=10.6$
galaxy GN-z11. This raises the question of whether Pop III
star formation is expected in the vicinity of massive high-
redshift galaxies or whether it would be prevented by early metal enrichment. We address this with a semi-analytic model of Pop III and metal-enriched
star formation applied to a zoom-in N-body simulation of a highly overdense region surrounding a GN-z11 analog. The zoom-in region resolves the minihalos where the first
stars form. We find that Pop III and metal-enriched
star formation both occur earlier, and with a higher
star formation rate density, than in a typical mean-density environment. The majority of Pop III
star formation in minihalos ends by $z\sim15$; however, we identify a distinct population of late-forming Pop III
galaxies that form near the central
galaxy. These systems have their
star formation delayed due to ionizing feedback from the central GN-z11 analog and experience a Lyman-Werner (LW) intensity of ${\sim}10$ - $100~J_{21}$, potentially leading to massive Pop III starbursts similar to Hebe. Interestingly, the abundance of these Pop III
galaxies depends strongly on the
baryon-dark matter streaming velocity, $v_{\rm bc}$, because a higher streaming velocity allows
halos to remain chemically pristine for longer. None of these late-forming Pop III sources occur for $v_{\rm bc}=0$, while many form at high $v_{\rm bc}$. Averaging over the expected distribution of streaming velocities, we predict that of order four GN-z11-like
galaxies would need to be surveyed at $z\sim11$ to expect to find one such source. We also note these environments may provide a natural formation site for heavy black hole seeds.
Read the full abstract ▾
Disordered Systems and Neural Networks
Discrete Scale Invariance of Ising Mesons
Denis V. Vasilyev and Abbas Ali Saberi
Discrete scale invariance often arises from resonant few-body or singular scale-invariant dynamics. We show that it can instead emerge in the internal motion of a composite excitation in an ordered
quantum magnet. In the marginal $1/r^4$ transverse-field Ising chain, long-range bonds generate an asymptotic inverse-square attraction between dressed domain walls, with strength fixed by the spontaneous magnetization. The dressed two-kink threshold
curvature sets the relative kinetic scale. Together, these independently accessible quantities determine the scale-anomaly exponent and hence the
meson hierarchy. In the supercritical regime, the same exponent governs
geometric binding-energy and size ratios, logarithmic level accumulation, and log-periodic threshold scattering. Full-
spin exact diagonalization and a fourth-order weak-field
Hamiltonian support the coupled energy–size scaling in the finite window between core and ring effects. This provides a microscopic many-body realization of
quantum limit-cycle physics, with its universal scaling fixed by
magnetic order and kink mobility.
Read the full abstract ▾
Statistical Computation
Ready for the Clinic? A Survey of Open-Source Software for Response-Adaptive Randomization in Clinical Trials
Stina Zetterstrom, David S. Robertson and Sofía S. Villar
Response-adaptive randomization (RAR) modifies treatment allocation probabilities during a clinical trial as response/outcome data accumulate, with the aim of improving patient benefit, statistical efficiency, or both. Despite substantial methodological development, adoption of RAR in clinical practice has remained limited, and the software available to support its design and implementation has not previously been reviewed. We identified 16 publicly available, open-source software packages implementing RAR methods, spanning urn-based, target-allocation,
Bayesian, Markov decision process (MDP)-based, and dose-finding approaches, and evaluated them with respect to their methodological and practical characteristics. We found that while several software packages exist, most are method-specific, and only a small number provide broader, general-purpose adaptive-trial design and analysis capabilities. Explicit support for practically relevant features, including delayed and missing outcome data, temporal trends in response rates, flexible operating-characteristic evaluation, and platform or multi-arm multi-stage trial designs, is rare or absent across the identified software. These findings suggest that while a diverse set of tools exists for exploring RAR designs, gaps remain between the methodological literature and the software available to implement it in practice. Continued development of flexible, practically oriented, and validated software is important for the wider adoption of RAR in clinical research, and we highlight interesting areas of further work.
Read the full abstract ▾
Machine Learning
KernelOnet: An Interpretable Neural Operator Based on Kernel Functions
Yuan Guo, Hanshu Chen, Qiang Xi et al. (+2 more)
This paper proposes an interpretable neural operator framework, the Kernel Operator Network (KernelOnet), which incorporates kernel functions explicitly into the neural operator architecture, so that the operator structure matches the kernel-expansion form used in boundary-type kernel-expansion methods. Unlike traditional neural operators such as DeepONet, which learn basis functions implicitly through deep networks, KernelOnet replaces the trunk network with explicit kernels and offers three complementary kernels: a data-driven learnable kernel, in which a neural network parameterizes a radial basis function learned from data, and which for constant-coefficient linear problems can be regarded as a non-singular fundamental solution; a physics-informed kernel, which embeds physical information such as analytic fundamental solutions into the network structure, so that the expansion satisfies the governing equation automatically and can be trained without supervision on boundary conditions alone, with no interior solution data; and a hybrid kernel, which splits the solution, according to the linear principal part of the governing equation, into a homogeneous part spanned by analytic fundamental solutions and a source part carried by low-rank learned correction kernels, thereby balancing physical priors against data fitting on nonlinear problems lacking an analytic fundamental solution. On three
benchmarks and one engineering problem in a shallow-water waveguide, KernelOnet attains high accuracy; where comparable with DeepONet, it is more accurate with fewer learnable parameters. Its unsupervised configuration needs no interior solution labels, and its per-query inference cost is far below that of per-instance solvers, offering an effective route to acoustic propagation in unbounded exterior domains that general-purpose neural operators struggle to handle.
Read the full abstract ▾
All authors Yuan Guo, Hanshu Chen, Qiang Xi, Timon Rabczuk, Zhuojia Fu
Cryptography and Security
PrivacySkills: How Privacy Guidance Shapes Source Selection in LLM Agents
Lucas Biechy, Cédric Eichler, Héber H. Arcolezi et al. (+1 more)
While prior work has documented
privacy failures in LLM
agents, it remains unclear how the presentation of
privacy guidance influences their choice of information sources. We introduce PrivacySkills, a controlled framework for evaluating how
agents choose among acquisition pathways that provide the same task-relevant value: consulting publicly available personal information, accessing confidential sources, or interacting with the user. The evaluation framework comprises 55 synthetic tasks spanning 11 categories of personal information, with 169 associated skills that describe the available acquisition pathways. We consider
privacy guidance through system-level instructions, skill-level metadata labels, or both. Separately, we vary user availability and urgency framing. With users available and no
privacy guidance,
agents access confidential sources in 30% of valid runs on average across five open-weight models, despite sufficient alternatives. This rate increases to 45% when users are unavailable, whereas urgency framing has no detectable effect. System-level
privacy instructions alone have limited effects on confidential access, while skill-level intrusiveness labels produce a modest reduction (24% on average), but combining the two roughly halves confidential access. Our findings motivate incorporating
privacy annotations into skill specifications and evaluating their effectiveness alongside system-level instructions.
Read the full abstract ▾
All authors Lucas Biechy, Cédric Eichler, Héber H. Arcolezi, Nicolas Anciaux
Multiagent Systems
Embodied Semantic Communication for Collective Autonomous Agents: A Tutorial on Representation, Wireless Delivery, and Closed-Loop Coordination
Yizheng Huang, Wensheng Lin, Lixin Li et al. (+3 more)
As autonomous systems and
embodied intelligence enter the dynamic physical world, multi-
agent collaboration calls for a paradigm shift in communication design. However, existing communication paradigms overlook that
agents form action understanding from their own states, environmental observations, and collaboration relations through a process that evolves as a task unfolds. Consequently, reliable bit delivery, general semantic recovery, or single-task utility optimization alone cannot ensure that heterogeneous
agents form coordinated actions compatible with their own conditions from shared information during task execution. To address this gap, this paper proposes
embodied semantic communication (ESC) as a paradigm that transforms information transmission into action-oriented semantic interaction. Specifically, ESC characterizes how an explicit communication link can encapsulate
multimodal perceptual states, intrinsic hardware capabilities, and collaborative intents into unified actionable semantic representations, thereby enabling heterogeneous receiving
agents to parse, align, and ground them in local motor control. This paper clarifies the conceptual boundary, system characteristics, and environment-constrained technical pathways of ESC. It maps the underlying mathematical tools, including semantic information theory, world models, and multi-
agent decision theory. Finally, this paper summarizes key open challenges, including measurable semantic reliability, ambiguity-triggered interaction under dynamic environments and tasks, and bandwidth-adaptive semantic transmission, outlining a roadmap for collective
embodied networks.
Read the full abstract ▾
All authors Yizheng Huang, Wensheng Lin, Lixin Li, Qinghe Du, Wenchi Cheng, Zhu Han
Robotics
Passive-Dynamic-Walking-Inspired Dynamics Guidance for Energy-Efficient Humanoid Locomotion
Hyeonjin Choi, Joongheon Kim and Daekyum Kim
Learning energy-efficient
humanoid locomotion requires discovering mechanically economical gait coordination, not merely reducing actuator effort. Reinforcement learning promotes efficiency through effort-related reward penalties, which guide the step-to-step mechanics of walking only indirectly. This article proposes a framework inspired by passive dynamic walking (PDW) that temporarily creates slope-equivalent conditions favorable to economical gait discovery and removes all PDW-specific guidance before nominal-dynamics optimization. During early training, a tilted-
gravity field assists sagittal progression on flat collision geometry, complemented by curriculum-coupled reward terms. The core framework requires no reference trajectories, gait phases, or contact schedules. In a five-seed forward-locomotion study on a 29-DoF Unitree G1, the framework reduces mechanical cost of transport by 6.8-15.2% over commanded speeds of 0.5-2.0m/s without degrading velocity tracking. Mechanical-work decomposition attributes the reduction to positive actuator work, and reward-matched comparisons separate the guided regime's faster gait acquisition from the tilt's additional benefit to converged economy. The framework extends to unassisted omnidirectional locomotion, where its benefit persists once a walking-specific motion prior supplies kinematic coordination, the combination reducing speed-matched cost of transport by 18.7%. On hardware, forward cost of transport falls by 16.3% with the motion prior and by 4.5% without it, the latter within the trial-to-trial spread.
Read the full abstract ▾
Machine Learning
Graph neural networks for sampling-invariant embeddings of organized signal sets
Martin Bauw, Santiago Velasco-Forero and Jesus Angulo
Sensor networks and radars can deliver signals as organized sets, e.g. ordered signals, signals describing range
cells within a grid or signals perceived as graph nodes. Within such sets, individual signals may be characterized by distinct sampling parameters. This paper investigates organized signal sets neural network encoders. In the context of this work, the purpose of such encoders is to project heterogeneously sampled signal sets into an arbitrary fixed-size vectors space. This new representation space is designed so that signal sets can be processed as vectors rid of sampling differences to allow for arbitrary
topology-aware processing with no signal processing constraints. Within this representation space designed to reduce the influence of heterogeneous sampling parameters, the relevance of signal sets representations is evaluated by considering signal sets discrimination potential with a focus on waveforms separation. The encoding and
embeddings discrimination experiments conducted rely exclusively on synthetic complex-valued radiofrequency signals.
Read the full abstract ▾
Populations and Evolution
Failure of the fittest: limits on asexual adaptation to changing environments
Rebekah Hall, Raphael Forien, Matthew M Osmond et al. (+1 more)
The dynamics of adaptation in a population are informed by its mode of reproduction. Whether individuals reproduce sexually or asexually, then, can have large effects on population persistence and extinction in changing environments. Classical theory explains this dependence via genetic variance without further accounting for a fundamentally different mode of inheritance. Recent theoretical advances capture these mechanisms, highlighting how they dramatically change the nature of adaptation in gradually changing environments. Here we show that these recent predictions, which assume infinite-population sizes, are overly optimistic about the fate of finite asexual populations when environmental change is rapid. This is because every individual in an infinite asexual population descends from a once optimal ancestor and the likelihood of finding this optimal individual in the population declines rapidly with population size. To predict the fate of finite asexual populations we derive the
equilibrium phenotypic distribution assuming there exist no individuals beyond some unknown trait. As expected, we find that when the phenotypic distribution contains the optimal trait, the population's fate is consistent with previous predictions. When the optimal trait is lost, however, an additional fitness load is incurred. We further derive a prediction for the typical trajectory of an ancestral lineage, which explains the additional load: all individuals descend from a common ancestor carrying the fittest trait, which may be suboptimal. Our work uncovers a finite-population effect that is specific to asexual adaptation and distinct from genetic drift, which significantly hinders the ability of small asexual populations to persist in changing environments.
Read the full abstract ▾
All authors Rebekah Hall, Raphael Forien, Matthew M Osmond, Vincent Calvez
Cryptography and Security
Same Bytes, Different Authority: Reserved-Token Representations in Chat-Template Prompt Injection
Yan Zhan, Yunze Song, Mengkai Hou et al. (+3 more)
Prompt injection against LLM
agents becomes much stronger when the injected instruction is wrapped in the model's own chat template. A forged template marker such as <|im_start|> can reach the model either as a single reserved control
token or as a sequence of ordinary subword
tokens. The two decode to exactly the same text, and because tokenization runs on the server, the defender rather than the attacker decides which one the model receives. We use this to measure how much of the injected instruction's authority comes from the reserved
token's learned representation. Encoding the forged markers as subwords, with the text held fixed and a control for the extra
tokens this adds, lowers attack success on the InjecAgent
benchmark by 39 to 66 percentage points on three of four open-weight families, and the gap carries over to multi-turn
agent tasks in AgentDojo. On Qwen3-8B the gap is 8 points, because without reserved ids the model still recognises the forged turn from its text by
reasoning; suppressing the
reasoning block widens the gap to 50. The authority sits in the single learned vector at the marker position: the mean of the marker's subword vectors does not reproduce it, the vector of the nearest ordinary
token restores the attack on Llama-3.1, and an adaptive attacker who searches for non-reserved markers finds such
embedding neighbours on three of four families. In every base and instruction-tuned pair we test, instruction tuning strengthens the model's preference for reserved markers. The standard mitigation, a tokenizer option that encodes special
tokens as ordinary subwords, applies only to
tokens a configuration declares special, so in 33 of 67 distinct tokenizer configurations, covering 255 of the 400 most-downloaded chat models on Hugging Face, it leaves intact the tool-protocol
tokens through which
agents read untrusted tool output, and the gap persists on that channel.
Read the full abstract ▾
All authors Yan Zhan, Yunze Song, Mengkai Hou, Wanting Zhang, Shaobo Liu, Zhijun Gao
Geometric Topology
On intersection forms of spin $4$-manifolds with AR homology sphere boundary
Eiichiro Hakamada
In this paper, building on the work of Dai–Sasahira–Stoffregen, we derive constraints on the intersection forms of smooth, compact, indefinite,
spin $4$-
manifolds bounded by AR homology spheres using two methods. First, we use the constraints on the existence of $\operatorname{Pin}(2)$-equivariant maps between representation spheres, established by Hopkins–Lin–Shi–Xu in their proof of the $10/8 + 4$-inequality. Second, we compute Lin's $\kappa o$-invariant for AR homology spheres. Combined with Lin's version of the relative $10/8$-inequality, this computation yields constraints different from those based on Hopkins–Lin–Shi–Xu. As applications, we obtain new lower bounds on the minimal genus of surfaces bounded by
knots whose double branched covers are AR homology spheres in various $4$-
manifolds, and provide an obstruction to smooth $4$-
manifold decompositions related to the $11/8$-
conjecture.
Read the full abstract ▾
Multiagent Systems
Bregman Consensus
Andrei N. Soklakov
Consider a community of
agents who are seeking consensus on a set of parameters. The
agents agree to use the same Bregman-type divergence to quantify disagreement between their individual estimates of the parameters but have varying confidence in each other's abilities. Each
agent is happy to revise their estimate by moving to the weighted barycenter of all individual estimates with higher weights applied to more trusted
agents. We show that such revisions naturally lead to an iterative
algorithm which converges to a unique consensus estimate of the parameters. Furthermore, since the consensus estimate is itself a barycenter with computable weights, the group emerges as a collective super-
agent with a well-formed opinion regarding the ability of each individual
agent.
Read the full abstract ▾
Machine Learning
CADENCE: A Confidence-Adaptive Dual-Expert Network for Fast and Accurate Time Series Classification
Onisa Mapunda
Time series classification (TSC) exhibits a sharp trade-off between accuracy and computational scalability. Meta-ensembles like HIVE-COTE 2.0 reach state-of-the-art accuracy but require extensive compute, whereas ultra-fast random convolutional transforms (e.g., MiniRocket, Hydra) run in seconds but struggle with phase-independent distributions, signal kinematics, and decision tree fragmentation on large class counts. In this work, we present CADENCE (Confidence-Adaptive Dual-Expert Network for time series Classification Excellence), a unified, CPU-native dual-expert architecture. CADENCE decouples representation learning into two specialized pathways: (i) a Convolutional Linear Expert pairing 10,000 deterministic dilated features with closed-form L2-regularized Woodbury ridge classification, and (ii) a Distributional Interval Expert pairing competing dilated kernels (Hydra) with dyadic Cornish-Fisher moment approximations across signal kinematics and FFT spectral bands, fitted with an ExtraTrees ensemble. An internal validation meta-router with rare-class preservation dynamically selects between pure expert routing and confidence-weighted soft blending, followed by a full refit on 100% of training data. Evaluated across all 109 equal-length UCR Archive
datasets over 30 resamples (3,270 total runs), CADENCE achieves a grand mean accuracy of 0.8864. This ranks #2 across the archive, surpassed only by HIVE-COTE 2.0 (0.8895, p_Holm = 0.295, no statistically significant difference), while outperforming Hydra+MultiRocket (0.8818), MultiRocket (0.8797), and HIVE-COTE 1.0 (0.8786, p_Holm = 0.048). CADENCE closes the gap to HIVE-COTE 2.0 to 0.31 percentage points while taking an average of only 17.53 seconds per
dataset on a dual-core CPU. Source code and evaluation scripts: https://github.com/onisa-jr/CADENCE.git
Read the full abstract ▾
Multiagent Systems
Prompted Identity Degrades Cooperation in Multi-Agent LLM Systems
Xavier Del Giudice, Alessio Palma, Matteo Migliarini et al. (+2 more)
Multi-
agent LLM systems increasingly mix models from several providers, yet exposing each
agent's underlying model identity to its peers significantly impairs cooperation. We show that when
agents are aware of each other's model family, the group splits into clusters, where
agents prefer interacting with others carrying their same label, although nothing in the task rewards or asks for such a split. We argue that the label itself causes this split, which we define as $\textit{factionalism}$. We show and measure this phenomenon in two cooperative games and on a
reasoning benchmark, with nine to twenty-five
agents drawn from up to five open-weight model families. We further show that when the announced families are shuffled, or replaced by arbitrary labels, the factions still follow this information; when the label is removed, this behavior disappears. In strictly cooperative tasks, labeled groups spend on average $30\%$ more rounds and $55\%$ more
tokens to reach a decision, and their success rate drops from $96\%$ to $81\%$. The effect replicates across tasks, group sizes and model families. Withholding identity labels from the
agents is simple and effective mitigation.
Read the full abstract ▾
All authors Xavier Del Giudice, Alessio Palma, Matteo Migliarini, Fabio Galasso, Indro Spinelli
Computational Complexity
Pseudo-solutions of polynomial systems and the lower bound problem for $\mbox{AC}^0[p]$-Frege systems
Jan Krajicek
The problem to establish a lower bound for $\mbox{AC}^0[p]$-Frege refutations of a system of
polynomial equations over ${\bf F}_p$ was in K. (2024) reduced to the existence of a pseudo-solution (a notion defined there) for the system. Here we reduce this further, for the system expressing the negation of the PHP, to a property of search trees querying values of linear maps on the vector space of low degree
polynomials over ${\bf F}_p$.
Read the full abstract ▾
Machine Learning
Normative Loss Landscape Navigation: A Trajectory-Based Approach to Mitigating Forgetting in Incremental Learning
Isabelle Aguilar, Zayn Andre Zainal, Luis Fernando Herbozo Contreras et al. (+2 more)
Continual learning models suffer from catastrophic forgetting when trained sequentially on non-stationary data distributions. Previously, this has been addressed through weight regularization. While preconditioning gradients offer a promising alternative to mitigate forgetting, current approaches are myopic. Conversely, standard regularization methods apply rigid, scalar Euclidean penalties that entirely ignore the underlying Riemannian geometry of the parameter space. To overcome this gap, we propose TMLN (Trajectory-Modulatory Landscape
Navigation), a normative
navigation policy that formalizes continual learning as an optimal control problem over a curved loss landscape. TMLN utilizes a
memory-efficient diagonal empirical Fisher Information
Matrix (FIM) to define a localized Riemannian
manifold. To compensate for the spatial limitations of the diagonal approximation, TMLN dynamically modulates a preconditioner using the normalized historical trajectory of the network's parameter values. By integrating this trajectory-based preconditioning directly into the gradient update, we actively shield historically critical parameter directions without relying on additive penalties. Empirical evaluations on class- and domain-incremental
benchmarks demonstrate that our method significantly reduces the loss barrier between consecutive tasks.
Read the full abstract ▾
All authors Isabelle Aguilar, Zayn Andre Zainal, Luis Fernando Herbozo Contreras, Zhaojing Huang, Omid Kavehei
Astrophysics of Galaxies
Photoelectron and electron microscopy investigation of laboratory grown Mg-silicate space dust analogues
Steffen Friis Holleufer, Mie Andersen, Frederik Granzow Doktor et al. (+9 more)
The atomic structure at the surface of interstellar silicate
dust particles likely plays a key role in a variety of chemical processes occurring in
star- and
planet-forming regions of the interstellar medium (ISM). Here, we use the photoelectric effect to characterize, in situ, the local chemical structure of Mg-silicate nanoparticulate films, and ex situ electron microscopy to trace changes to morphology resulting from different starting compositions. Nanoparticulate films are prepared via co-deposition of Si, Mg and O atoms on a graphitic substrate under ultra-high
vacuum (UHV) and are characterised with X-ray photoelectron spectroscopy and near edge X-ray absorption fine structure measurements. Analysis, supported by density functional theory calculations, shows that adjusting the Mg-to-Si ratio from 3.5 to 1.8 changes particle composition from a mixture of MgO and Mg-silicate towards predominantly Mg-silicate. Annealing in UHV also pushes the composition towards Mg-silicate, however, non-stoichiometric chemical motifs are always observed, presenting a distribution of potentially catalytically active sites. Electron microscopy images show composition-dependent film morphologies that restructure differently upon thermal annealing. The Mg-silicate nanoparticulate films presented here are characterised as a 2D version of ISM analogue
dust particles prepared via
laser ablation and are therefore a more readily suitable model for surface science investigations into the catalytic properties of interstellar Mg-silicate
dust.
Read the full abstract ▾
All authors Steffen Friis Holleufer, Mie Andersen, Frederik Granzow Doktor, Lars Eric Borchert, Martin Wibrand Larsen, Duncan S. Sutherland, Zheshen Li, Alexei Preobrajenski, Jeppe Vang Lauritsen, Cornelia Jäger, Liv Hornekær, Andrew Cassidy
Artificial Intelligence
Grab a Coffee: Future-Aware Guidance for Discrete Diffusion with Compiled Objectives
Hua Xu, Dongxin Li, Gwen Yidou-Weng et al. (+3 more)
Discrete
diffusion models generate sequences by iteratively resolving multiple
tokens in parallel, offering a flexible alternative to left-to-right generation. However, guiding this process with a sequence-level objective is difficult because the value of one unresolved
token depends on the other
tokens with which it can form a high-reward sequence. Enumerating all such completions makes the whole guidance computation grow exponentially with the number of unresolved positions. We introduce COFFEE, a plug-and-play framework that avoids this enumeration by separating sequence dependence from the objective. At each
diffusion step, a target-free carrier absorbs the marginal
token distributions predicted by the denoiser to construct a joint model over the unresolved
tokens, while a compiled finite-state model records how their combinations affect the sequence-level preference. Pairing their states allows COFFEE to transfer global preferences to unresolved positions and sample a clean reconstruction without retraining the
diffusion model. The same framework supports explicit hard constraints and learned soft objectives. We evaluate COFFEE across multiple symbolic,
language, and biological
benchmarks, where it achieves strong control results with task-dependent quality and diversity trade-offs. By making objectives available to inference rather than only evaluation, COFFEE brings joint conditioning, completion-weighted guidance, and optimization-based constraints into pretrained neural generation, showing the potential of neural-symbolic methods in
diffusion guidance.
Read the full abstract ▾
All authors Hua Xu, Dongxin Li, Gwen Yidou-Weng, Guy Van den Broeck, Wei Wang, Anji Liu