arXiv:2606.14679v1 Announce Type: new Abstract: Online inventory optimization (OIO) is online convex optimization with physical memory: inventory carryover makes the feasible action set depend on the past. A natural principle, used in stochastic inventory learning and recently in OIO under a single linear capacity constraint, is to maintain a hidden target chosen by an online learner and implement its projection onto the currently feasible order-up-to set. We prove that this simple principle is optimal for OIO on arbitrary bounded convex capacity sets. With online gradient descent as the base learner, the method improves the best known regret guarantee for OIO on general convex sets from inverse to inverse-square-root dependence on the common-demand probability, and we prove a matching lower bound. The same principle gives the first polylogarithmic regret guarantee for strongly convex losses and the first dynamic regret guarantee adapting to Euclidean path variation on general convex capacity sets. The analysis introduces a norm alignment principle: the right state variable is the distance from the hidden target to the feasible set, measured in the same norm as the projection. Under norm alignment, this distance evolves pathwise as a scalar queue, with target movement as arrival and common demand as service. This reduction to one-dimensional queue control resolves the state dependence and extends the guarantees to general convex capacity sets, beyond the reach of prior productwise approaches. Experiments on synthetic and real-world inventory data corroborate the theory.
Science Journals
arXiv:2606.14684v1 Announce Type: new Abstract: Real-time fire classification systems require models that are simultaneously accurate, computationally efficient, and deployable on resource-constrained hardware. This work proposes \textbf{HumP-KD}, a Hybrid Uncertainty-aware Multi-stage Progressive Knowledge Distillation framework for efficient fire classification. Two datasets, FlameVision and Dataset-II, containing 8,600 and 31,309 images, are used. Various CNN and transformer baselines are applied under standard preprocessing, online augmentation, Gaussian noise and motion blur robustness conditions. The proposed HumP-KD model distills knowledge from two frozen heterogeneous transformer teachers, Swin-Tiny and ViT-Base, along with their Meta-MLP ensemble, into a lightweight MobileViT-S student via three tightly integrated components. Hierarchical Progressive Knowledge Distillation employs a Hierarchical Feature Builder. It generates a fused spatial attention mask to guide distillation toward discriminative regions selectively. Multi-Stage Knowledge Distillation progressively activates three distillation stages across training. On Dataset-II, HumP-KD achieves a mean F1 score of $0.9876 \pm 0.0063$ across 10 independent trials, significantly outperforming the MobileViT-S baseline trained without distillation ($0.9537 \pm 0.0351$), with statistical significance confirmed by both independent t-test ($p = 0.0195$) and Wilcoxon signed-rank test ($W = 1$, $p = 0.0039$). The proposed method also demonstrates strong generalization across datasets and robustness under degraded visual conditions. The student model retains only 4.94M parameters and 19.01Mb model size, representing a $5.7\times$ parameter reduction over Swin-Tiny and a $17.5\times$ reduction over ViT-Base, while achieving 37.72 CPU FPS, making it suitable for real-time deployment.
arXiv:2412.15545v2 Announce Type: replace Abstract: Social media is an important space for interactions between climate policy actors, with a burgeoning literature recognizing them as critical platforms of political contestation. We identified Twitter accounts associated with 904 climate change policy actors across nine countries, and collected their activities from 2017--2022, totalling 40 million activities from 16,086 accounts at different organizational levels. We studied these actors and their interactions as a polycentric governance system, emphasizing how boundary blurring between the public and private on social media platforms uniquely shapes the online policy process. Initial results show there is considerable temporal and cross-national variation in how prominent climate-related activities were, but all national policy systems generally responded to climate-related events, such as climate protests, in a similar manner. Examining patterns of interaction within and across countries, we find that these national policy systems rarely directly interact with one another, but are connected through consistently engaging with the same content produced by accounts of international organizations, climate activists, and researchers.
arXiv:2412.15707v3 Announce Type: replace Abstract: This paper investigates whether online learning algorithms in pricing produce competitive outcomes or tacit collusion. This issue has drawn considerable attention from competition regulators as algorithmic pricing becomes more common in digital markets. Understanding when such algorithms lead to equilibrium or supra-competitive prices is critical for buyers, sellers, and policymakers. We study the behavior of multi-armed bandit (MAB) online learning algorithms in repeated price competition. These algorithms require little information to learn, making them realistic models of automated pricing. Our analysis shows that mean-based algorithms, a special variant of online learning algorithms, converge to correlated rationalizable actions. In the Bertrand environments considered, this implies convergence to the Nash equilibrium or adjacent prices. Numerical experiments reveal that most MAB algorithms, including those that are not mean-based, also converge. We observe supra-competitive prices only in specific cases where all sellers implement the same symmetric version of certain algorithms, such as UCB. This effect diminishes as the number of competitors increases. Our results suggest that, even in a stylized repeated Bertrand competition, sustained supra-competitive prices may be less of a concern when independent agents use different online learning algorithms. Our insights are relevant for regulators and managers considering the use of algorithmic pricing algorithms.
arXiv:2505.11577v5 Announce Type: replace Abstract: Recent application programming interface (API) restrictions on major social media platforms challenge compliance with the EU Digital Services Act [20], which mandates data access for algorithmic transparency. We develop a structured audit framework to assess the growing misalignment between regulatory requirements and platform implementations. Our comparative analysis of X/Twitter, Reddit, TikTok, and Meta identifies critical ``audit blind-spots'' where platform content moderation and algorithmic amplification remain inaccessible to independent verification. Our findings reveal an ``accountability paradox'': as platforms increasingly rely on AI systems, they simultaneously restrict the capacity for independent oversight. We propose targeted policy interventions aligned with the AI Risk Management Framework of the National Institute of Standards and Technology [80], emphasizing federated access models and enhanced regulatory enforcement.
arXiv:2606.14402v1 Announce Type: cross Abstract: Electrochemical doping offers dynamic control of the electronic properties of organic semiconductors, and it is the enabling feature of a range of technologies, including electrochemical transistors, energy-storage devices, light-emitting electrochemical cells (LECs), and bioelectronics. Electrochemical doping is commonly controlled by the selection of the constituents in the active material of the device or the applied voltage bias, but herein we report that the incorporation of Au nanoparticles (Au-NPs) at an electrode interface can constitute an alternative control parameter. The LEC features balanced p- and n-type electrochemical doping that forms a p-n junction doping structure in its active material, and we find that it is possible to reshape this doping profile by incorporating Au-NPs at an electrode interface. Specifically, we establish that the inclusion of neat non-capped Au-NPs at the anodic interface shifts the p-n junction (i.e., the emission zone) away from the anode. In contrast, the inclusion of Au-NPs capped with sodium citrate is found to reverse this behavior, so that the emission zone is instead moved towards the anode. We utilize this control parameter to shift the emission zone towards a position of constructive (destructive) interference, as manifested in a strong increase (decrease) of the LEC emission efficiency. Our findings establish an interfacial strategy for modulating the spatial profile of electrochemical doping and tuning device performance without altering the chemistry of the active material, relying instead on the surface modification of one electrode. This approach is important because it provides a versatile and minimally invasive route to optimize electrochemical devices while preserving the intrinsic properties and formulation of the active material.
arXiv:2606.14568v1 Announce Type: cross Abstract: Glioma grading and survival prediction require the integration of heterogeneous information collected at different spatial and biological scales. Histopathology describes tissue morphology, mRNA expression captures molecular activity, and magnetic resonance imaging provides a non-invasive view of tumor extent and radiological heterogeneity. Existing glioma prognosis models often combine only two of these sources, while their alignment objectives remain mostly pairwise. This paper introduces GLORIA, a novel trimodal framework for GLioma Omics - Radiology - hIstopathology Alignment. GLORIA processes whole-slide image regions, gene-expression profiles, and 3D MRI volumes through modality-specific encoders, projects them into a shared latent space, and aligns them with a Gramian contrastive loss that measures the volume spanned by the three modality embeddings. The aligned representations are fused through a cross-modal gating module and optimized jointly for three-class glioma grading and overall survival prediction. We evaluate GLORIA on a matched TCGA-GBM/LGG and BraTS21 cohort, comprising 132 patients with all three modalities. On the shared trimodal test set, GLORIA improves over the bimodal WSI-mRNA baseline in all the metrics considered.
arXiv:2606.14590v1 Announce Type: cross Abstract: Raman and infrared anomalies associated with H-bond symmetrization in $\delta$-AlOOH, including mode softening and linewidth broadening at 5-10 GPa, occur at significantly lower pressures than predicted by static harmonic theory. To resolve this discrepancy, we combine harmonic phonon calculations with strongly constrained and appropriately normed (SCAN)-based deep-potential molecular dynamics and phonon quasiparticle analysis at 300 K. This framework extracts temperature- and pressure-dependent frequencies and lifetimes from long-time trajectories, capturing the branch reorganization and rapid linewidth growth characteristic of the disordering regime. Incorporating quasiparticle renormalization and directional longitudinal-optical-transverse-optical (LO-TO) splitting further yields near-quantitative agreement with the ambient-pressure OH-stretching Raman multiplet. These results identify finite-temperature dynamical effects and the progressive loss of spectral coherence as the origin of the spectroscopic signatures of H-bond symmetrization.
arXiv:2511.05017v3 Announce Type: replace Abstract: Hallucinations in Large Vision-Language Models (LVLMs) remain a persistent challenge, often stemming from inadequate integration of visual information during multimodal reasoning. A key cause is the model's over-reliance on textual priors and underutilization of visual cues, leading to outputs that are linguistically fluent but visually inaccurate. For example, given an image of an empty kitchen countertop, an LVLM might hallucinate a "bowl of fruit" or "cup of coffee", relying on language associations rather than visual evidence. Most LVLMs incorporate visual features by appending them to the input stream of a pre-trained LLM and training on large-scale vision-language datasets. Our systematic analysis reveals that this strategy often leads to over-dependence on textual information due to the inherent bias of LLMs towards language-dominant representations. This imbalance skews attention towards the text over visual content, weakening the model's ability to ground outputs in visual inputs. To address this, we propose a simple yet effective visual feature incorporation method that encourages the model to learn visually-informed textual embeddings distinct from those of the base LLM and promotes a more balanced attention distribution. Experimental results across multiple hallucination benchmarks demonstrate that our method significantly reduces hallucinations and fosters more balanced multimodal reasoning. Notably, our approach achieves substantial gains, including +9.33% on MMVP-MLLM, +2.99% on POPE-AOKVQA, up to +3.4% on Merlin, and +3% on the hard-data split of HallusionBench.
arXiv:2511.14897v2 Announce Type: replace Abstract: We present an unsupervised single image bidirectional Magnetic Resonance Image (MRI) synthesizer that synthesizes an Ultra-Low Field (ULF) like image from a High-Field (HF) magnitude image and vice-versa. Unlike existing MRI synthesis models, our approach is inspired by the physics that drives contrast changes between HF and ULF MRIs. Our forward model simulates a HF to ULF transformation by estimating the tissue-type Signal-to-Noise ratio (SNR) values based on target contrast values. For the Super-Resolution task, we used an Implicit Neural Representation (INR) network to synthesize HF image by simultaneously predicting tissue-type segmentations and image intensity without observed HF data. The proposed method is evaluated using synthetic ULF-like data from generated from standard 3T T$_1$-weighted images for qualitative assessments and paired 3T-64mT T$_1$-weighted images for validation experiments. WM-GM contrast improved by 52% in synthetic ULF-like images and 37% in 64mT images. Sensitivity experiments demonstrated the robustness of our forward model to variations in target contrast, noise and initial seeding.
arXiv:2602.13421v2 Announce Type: replace-cross Abstract: Computation in biological systems is fundamentally energy-constrained, yet standard theories of computation treat energy as freely available. Here, we argue that variational free energy minimization under a Poisson assumption offers a principled path toward an energy-aware theory of computation. Our key observation is that the Kullback-Leibler (KL) divergence term in the Poisson free energy objective becomes proportional to the prior firing rates of model neurons, yielding an emergent metabolic cost term that penalizes high baseline activity. This structure couples an abstract information-theoretic quantity -- the *coding rate* -- to a concrete biophysical variable -- the *firing rate* -- which enables a trade-off between coding fidelity and energy expenditure. Such a coupling arises naturally in the Poisson variational autoencoder (P-VAE) -- a brain-inspired generative model that encodes inputs as discrete spike counts and recovers a spiking form of *sparse coding* as a special case -- but is absent from standard Gaussian VAEs. To demonstrate that this metabolic cost structure is unique to the Poisson formulation, we compare the P-VAE against Grelu-VAE, a Gaussian VAE with ReLU rectification applied to latent samples, which controls for the non-negativity constraint. Across a systematic sweep of the KL term weighting coefficient $\beta$ and latent dimensionality, we find that increasing $\beta$ monotonically increases sparsity and reduces average spiking activity in the P-VAE. In contrast, Grelu-VAE representations remain unchanged, confirming that the effect is specific to Poisson statistics rather than a byproduct of non-negative representations. These results establish Poisson variational inference as a promising foundation for a resource-constrained theory of computation.
arXiv:2602.16221v2 Announce Type: replace-cross Abstract: This paper re-evaluates the relationship between the observed sunward electron cutoff energy and the depth of the Sun's global electrostatic potential. It investigates whether taking into account the effects of local traps formed by magnetic fluctuations provides an alternative explanation for the observed electron deficit. The fluctuations of the highly variable interplanetary magnetic field form a series of shallow magnetic mirror traps that move at approximately the speed of the solar wind. The study investigates the dynamics of electrons as they move outward against an attractive solar electrostatic potential and interact with these traps. By following the motion of the electrons using first-principles calculations, we assess the effect of the traps on the velocity distribution of the particles. Electrons that escape the local trap continue to lose energy as they move outward until they are eventually captured by subsequent traps, preventing them from returning to the observer as sunward-moving particles. We derive a mathematical expression for the cutoff velocity, defined as the limit beyond which particles can no longer overtake the outer endpoint of a local trap. It turns out that the observed cutoff energy characterizes only the local potential drop within a trap, rather than the total depth of the Sun's potential well. The true potential well can be significantly deeper, scaled by the ratio of the radial distance from the Sun to the trap size. Furthermore, electrons captured by these moving traps contribute to the formation of the solar wind core population. The Sun's electrostatic potential is a more significant factor in solar wind acceleration than previously interpreted from cutoff data. The interaction between electrostatic deceleration and moving magnetic traps provides a new framework for understanding the origin and behavior of the solar wind core electrons.
arXiv:2602.18023v4 Announce Type: replace-cross Abstract: We present \texttt{warpax}, a JAX-based toolkit that certifies the energy condition structure of warp drive spacetimes for all observers, in a frame-independent way, from the eigenstructure of the mixed stress-energy tensor $T^a{}_b$. At a Hawking--Ellis Type~I point, an eigenvalue inequality decides each energy condition exactly and for every observer. A Type~IV point has no rest frame and violates all of them unconditionally. Because the eigenvalues are boost invariants and no Eulerian normal is ever constructed, the test is well defined at all warp speeds, including $v_s \ge 1$. Across the warp drive family, this yields a clean dichotomy through the luminal transition: the irrotational Rodal geometry is globally Type~I at every speed, while the Alcubierre, Nat\'ario, and Van~den~Broeck bubble walls are dominated by Type~IV regions. The split is controlled by the vorticity of the ADM shift, with the Type~IV imaginary eigenvalue growing linearly, $f=\kappa\omega$, in a controlled limit. Single-frame analyses understate this structure. For the Rodal drive, the Eulerian frame misses about $72\%$ of the wall weak-energy violations seen by boosted observers, a statement made analytic by a closed-form worst observer. A boost-invariant exoticity ranking places the irrotational drive roughly seventy times below the bubble-wall drives, and the wall null-energy deficit follows a universal $-C v_s^2$ law that makes the Santiago--Schuster--Visser no-go result quantitative. A geodesic-integrated averaged null energy condition and a Ford--Roman comparison preserve the ordering: every drive violates, and the Rodal geometry is the mildest.
arXiv:2504.16288v3 Announce Type: replace Abstract: It is well-known that checking whether a given string $w$ matches a given regular expression $r$ can be done in quadratic time $O(|w|\cdot |r|)$ and that this cannot be improved to a truly subquadratic running time of $O((|w|\cdot |r|)^{1-\epsilon})$ assuming the strong exponential time hypothesis (SETH). We study the related problem that asks whether $w$ has a \emph{subsequence} that matches $r$, and we show that surprisingly this task admits an algorithm that runs in linear time, i.e., in $O(|w| + |r|)$. We further show that the same holds if we ask for a supersequence instead of a subsequence. Moreover, we show that the \emph{quantitative} problems of computing a longest subsequence or shortest supersequence of $w$ that matches $r$ can be solved with the same complexity as the classical longest common subsequence or shortest common supersequence problems, i.e., in $O(|w|\cdot |r|)$, and conditionally not in $O((|w|\cdot|r|)^{1 - \epsilon})$. By contrast, if instead of subsequences or supersequences we consider other string relations like the infix, prefix, left-extension, or extension relations, then all the corresponding problems (both quantitative and non-quantitative) have the same complexity as classical regex matching, i.e., they can also be solved in $O(|w|\cdot |r|)$, but not in $O((|w|\cdot|r|)^{1 - \epsilon})$ assuming SETH. We last study the complexity of the \emph{universal} problem that asks if \emph{all} subsequences (or supersequences, infixes, prefixes, left-extensions or extensions) of an input string satisfy a given regular expression. For these problems, we show polynomial upper bounds (along with matching conditional lower bounds) for the infix and prefix relations, but PSPACE-completeness for the extension, left-extension and supersequence relations, and coNP-completeness for the subsequence relation.
arXiv:2507.00481v2 Announce Type: replace Abstract: Although software engineering research has focused on optimizing processes and technology, there is a growing recognition that human factors, particularly teamwork, also significantly impact optimization. Recent research suggests that developer personality has a strong influence on teamwork. In fact, personality considerations may have a greater impact on software development than processes and tools. This paper aims to design a study that measures the impact of HEXACO personality traits on the Teamwork Quality (TWQ) of software teams. A preliminary data collection (n=54) was conducted for this purpose. The analysis showed that several personality traits, as well as their composition, had a significant impact on TWQ. Additionally, other variables, such as the proportion of women and age distribution, also affected TWQ. The study's initial results demonstrate the usefulness and validity of the study design. The results also suggest several opportunities to improve teamwork in IT organizations and avenues for further research.
arXiv:2507.05919v2 Announce Type: replace Abstract: We axiomatically characterize the orderings of pairs of sets induced by several distances: Hamming, Jaccard, S{\o}rensen-Dice and Overlap. We also axiomatically characterize these distances. Our axioms are properties describing how a distance changes when we perform elementary modifications of the sets, like adding one element to one of the sets, to both sets, swapping both sets, permuting some elements, etc.
arXiv:2507.10834v4 Announce Type: replace Abstract: Assortment optimization seeks to select a subset of substitutable products, subject to constraints, to maximize expected revenue. The problem is NP-hard due to its combinatorial and nonlinear nature and arises frequently in industries such as e-commerce, where platforms must solve thousands of such problems each minute. We propose a graph convolutional network (GCN) framework to efficiently solve constrained assortment optimization problems. Our approach constructs a graph representation of the problem, trains a GCN to learn the mapping from problem parameters to optimal assortments, and develops three inference policies based on the GCN's output. Owing to the GCN's ability to generalize across instance sizes, patterns learned from small-scale samples can be transferred to large-scale problems. Theoretical results are established to show the expressive power of the proposed GCN, and explain the underlying mechanism of the size generalization ability. Numerical experiments show that a GCN trained on instances with 20 products achieves over 85% of the optimal revenue on problems with up to 2,000 products within seconds, outperforming existing heuristics in both accuracy and efficiency. We further extend the framework to settings with an unknown choice model using transaction data and demonstrate similar performance and scalability.
arXiv:2507.13263v4 Announce Type: replace Abstract: Bayesian Optimization (BO) is a powerful tool for black-box optimization, but its application to high-dimensional permutation spaces is severely limited by the challenge of defining scalable representations. The current state-of-the-art BO approach for permutation spaces relies on an exhaustive $\Omega(n^2)$ pairwise comparison, inducing a dense representation that is impractical for large-scale permutations. To break this barrier, we introduce a novel framework for generating efficient permutation representations via kernel functions derived from sorting algorithms. Within this framework, the Mallows kernel can be viewed as a special instance derived from enumeration sort. Further, we introduce the \textbf{Merge Kernel} , which leverages the divide-and-conquer structure of merge sort to produce a compact, $\Theta(n\log n)$ to achieve the lowest possible complexity with no information loss and effectively capture permutation structure. Our central thesis is that the Merge Kernel performs competitively with the Mallows kernel in low-dimensional settings, but significantly outperforms it in both optimization performance and computational efficiency as the dimension $n$ grows. Extensive evaluations on various permutation optimization benchmarks confirm our hypothesis, demonstrating that the Merge Kernel provides a scalable and more effective solution for Bayesian optimization in high-dimensional permutation spaces, thereby unlocking the potential for tackling previously intractable problems such as large-scale feature ordering and combinatorial neural architecture search.
arXiv:2509.02426v3 Announce Type: replace Abstract: Electrified roadways (ER) equipped with dynamic wireless power transfer (DWPT) capabilities can patently extend the driving range and reduce the battery size of electric vehicles (EVs). However, due to the spatial arrangement of the transmitter coils in the ER, the DWPT load exhibits frequency content that could excite power system frequency dynamics. In this context, this work aims to study the spectrum of DWPT loads under different traffic conditions. Under simplifying assumptions, we develop statistical models to identify the location and relative magnitude of DWPT load harmonics. Our analysis reveals that the fundamental frequency depends on ER coil spacing and average EV speed. In the worst-case yet unlikely scenario that EVs move in a synchronized fashion, the amplitude of harmonics scales with the EV count. On the contrary, when EVs move freely, harmonics scale with the square root of the EV count. Platoon formations can accentuate harmonics. The spectral content around harmonics decreases in magnitude and increases in bandwidth with the harmonic index. The load of a single EV moving at a time-varying speed can be modeled as a frequency-modulated (FM) signal. Despite the simplifying assumptions, the derived models offer valuable insights for ER planners and grid operators. Dynamic simulations of a modified WECC model with DWPT loads synthesized from realistic EV trajectories and ER specifications corroborate some of these insights.
arXiv:2509.16143v2 Announce Type: replace Abstract: Given an undirected graph G = (V, E) and an integer k, the s-Club asks if Gcontains a vertex subset S of at least k vertices such that G[S] has diameter at most s. Recently, Vertex r-Triangle s-Club, and Edge r-Triangle s-Club that generalize the notion of s-Club have been studied by Garvardt et al. [TOCS-2023, IWOCA-2022] from the perspective of parameterized complexity. Given a graph G and an integer k, the Vertex r-Triangle s-Club asks if there is an s-Club S with at least k vertices such that every vertex u \in S is part of at least r triangles in G[S]. In this paper, we initiate a systematic study of Vertex r-Triangle s-Club for every integer r >= 1 from the perspective of structural parameters of the input graph. In particular, we provide FPT algorithms for Vertex r-Triangle 2-Club when parameterized by the treewidth (tw) of the input graph, and an XP algorithm when parameterized by the h-index of the input graph. Additionally, when parameterized by the feedback edge number (fes) of the input graph. We provide a kernel of O(fes) edges for Vertex r-Triangle s-Club.
arXiv:2601.18707v2 Announce Type: replace Abstract: Machine learning-based surrogate models have emerged as more efficient alternatives to numerical solvers for physical simulations over complex geometries, such as car bodies. Many existing models incorporate the simulation mesh as an additional input, thereby reducing prediction errors. However, generating a simulation mesh for new geometries is computationally costly. In contrast, mesh-free methods, which do not rely on the simulation mesh, typically incur higher errors. Motivated by these considerations, we introduce SMART, a neural surrogate model that predicts physical quantities at arbitrary query locations using only a point-cloud representation of the geometry, without requiring access to the simulation mesh. The geometry and simulation parameters are encoded into a shared latent space that captures both structural and parametric characteristics of the physical field. A physics decoder then attends to the encoder's intermediate latent representations to map spatial queries to physical quantities. Through this cross-layer interaction, the model jointly updates latent geometric features and the evolving physical field. Extensive experiments show that SMART is competitive with and often outperforms existing methods that rely on the simulation mesh as input, demonstrating its capabilities for industry-level simulations.
arXiv:2512.04981v2 Announce Type: replace Abstract: Text-to-image (T2I) systems increasingly rely on Large Language Model (LLM)-based text conditioning to interpret and expand user prompts. While this improves prompt understanding and text-image alignment, we find that it can also introduce implicit demographic assumptions, even when demographic attributes are unspecified. To systematically investigate this behavior across varying levels of prompt ambiguity and complexity, we construct a comprehensive benchmark covering diverse prompt settings. Evaluations on eight recent T2I models show that LLM-based systems consistently exhibit stronger demographic skew than non-LLM-based baselines. We further analyze system prompts, a component unique to LLM-based T2I systems that guides prompt interpretation and expansion. Our analyses show that these instructions strongly influence text embeddings, which subsequently leads to biased image generations. Motivated by these findings, we propose FairPro, a training-free debiasing framework that adaptively generates fairness-aware instructions while preserving user intent. Experiments demonstrate that FairPro substantially reduces demographic disparities while maintaining prompt fidelity.
arXiv:2512.05025v2 Announce Type: replace Abstract: Earth observation (EO) data spans a wide range of spatial, spectral, and temporal resolutions, from high-resolution optical imagery to low resolution multispectral products or radar time series. While recent foundation models have improved multimodal integration for learning meaningful representations, they often expect fixed input resolutions or are based on sensor-specific encoders limiting generalization across heterogeneous EO modalities. To overcome these limitations we introduce RAMEN, a resolution-adjustable multimodal encoder that learns a shared visual representation across EO data in a fully sensor-agnostic manner. RAMEN treats the modality and spatial and temporal resolutions as key input data features, enabling coherent analysis across modalities within a unified latent space. Its main methodological contribution is to define spatial resolution as a controllable output parameter, giving users direct control over the desired level of detail at inference and allowing explicit trade-offs between spatial precision and computational cost. We train a single, unified transformer encoder reconstructing masked multimodal EO data drawn from diverse sources, ensuring generalization across sensors and resolutions. Once pretrained, RAMEN transfers effectively to both known and unseen sensor configurations and outperforms larger state-of-the-art models on the community-standard PANGAEA benchmark, containing various multi-sensor and multi-resolution downstream tasks. Our code and pretrained model are available at https://github.com/nicolashoudre/RAMEN.
arXiv:2603.16499v2 Announce Type: replace-cross Abstract: Clinical thermal ablation outcomes display significant variability that classical bio-heat models cannot fully explain. One reason may lie in the fractal architecture of biological tissues, which has been identified as a robust biomarker directly correlated with cancer grades. This structural heterogeneity, together with memory effects (e.g., thermotolerance), causes heat transfer in living tissues to differ from Fourier diffusion, resulting in anomalous biological transport. In this work, we implemented a realistic fractal-fractional bio-heat model, with non-linear perfusion and PI-controlled power delivery, to quantify the role of tissue fractality in ablation outcomes. Our results reveal that the expansion of coagulation zones is jointly controlled by fractal geometry and its associated topological connectivity. These findings highlight spectral dimension as a key driver of clinical variability, successfully reproducing the reduced ablative efficacy in liver metastases compared to primary carcinomas, and provide evidence for topologically informed treatment strategies for the thermal ablation of malignant neoplasms.
arXiv:2603.04555v2 Announce Type: replace Abstract: AI-driven automation threatens to erode government tax bases, lower living standards, and disempower citizens--risks that mirror the 40-year stagnation of wages during the first industrial revolution. While AI safety research has focused primarily on capability risks, comparatively little work has studied how to mitigate the economic risks of AI. This position paper argues that technical governance researchers should prioritize the study of token taxes: usage-based surcharges on model inference applied at the point of sale. We situate token taxes within previous proposals for robot taxes and identify two key advantages: they are enforceable through existing compute governance infrastructure, and they capture value where AI is used rather than where models are hosted. We then present a research roadmap. For enforcement, we outline a staged audit pipeline--black-box token verification, norm-based tax rates, and white-box audits--and identify open technical problems at each stage. For impact, we highlight the need for economic modeling of cost pass-through and deadweight loss. Finally, we discuss why FLOP taxes may be preferable, token taxes could stifle innovation, and how to prevent AI superpowers from vetoing such measures.