arXiv:2509.22015v2 Announce Type: replace
Abstract: Standard Sparse Autoencoders (SAEs) excel at discovering a dictionary of a model's learned features, providing a powerful lens for passive feature discovery. However, this passive nature makes it difficult to systematically evaluate or analyze concepts that users explicitly care about. We introduce Concept-SAE, a framework that augments SAEs with a structured and controllable interface for probing user-defined concepts. Concept-SAE decomposes an activation subspace into two orthogonal components: Concept Tokens, which are aligned to externally specified semantics through dual supervision on both concept existence and spatial localization, and Free Tokens, which operate like standard SAEs to capture all remaining information. This hybrid disentanglement strategy ensures that Concept Tokens are faithful, spatially grounded, and cleanly separated from the residual subspace while preserving the ability of SAEs for open-ended concept discovery. We conduct extensive experiments demonstrating that Concept-SAE yields high-fidelity, well-localized, and strongly disentangled concept representations, outperforming alternatives in interface quality. Finally, we validate the utility of this conceptual interface through three diagnostic evaluations: a detection test on classifying adversarial image samples, a controllability test focusing on controlled counterfactual editing and a stability test using adversarial perturbations. Together, these results show that Concept-SAE equips SAEs with a reliable mechanism for evaluating, probing, and diagnosing user-defined concepts.
Science Journals
arXiv:2509.07833v3 Announce Type: replace
Abstract: We perform a theoretical study of a continuous superradiant laser supporting its experimental realization at FEMTO-ST using two sequentially-emitting ensembles of ${}^{171}\mathrm{Yb}$ atoms coupled to the same Fabry-Perot cavity. Using an open quantum system approach, we identify for the simplest case the parameter space where the laser reaches tens of picowatts of power with a sub-millihertz linewidth. Studying the impact of inhomogeneous frequency broadening and variations in atom-cavity coupling on the superradiant emission, we find the laser properties robust with respect to such perturbations, also thanks to the occurrence of synchronization of the atomic dipoles. We then consider a two-site configuration, in which atoms in each site are equally coupled to the cavity and have equal detunings, with different values for the two ensembles. We find for balanced and imbalanced atom numbers that synchronization leads in a certain parameter space to a single narrow spectral line whose central frequency follows the weighted average frequency. This result indicates that sequential loading can enable continuous superradiant emission for metrological applications, provided that the relative frequencies of the two ensembles are controlled to the level required by the target stability.
arXiv:2605.24470v3 Announce Type: replace
Abstract: Video-text retrieval has witnessed remarkable progress driven by large-scale vision-language pretraining, yet most existing approaches inherit an implicit assumption from image-text retrieval: that visual semantics can be captured frame-by-frame. This assumption overlooks the temporal dynamics of egocentric videos. The EPIC-KITCHENS-100 Multi-Instance Retrieval (MIR) challenge further raises the bar by providing soft-label relevance matrices rather than binary labels, demanding models that can resolve graded semantic correspondences across modalities. In this report, we present our solution, termed TempRet, to the CVPR 2026 EPIC-KITCHENS-100 MIR challenge. Our approach builds upon a CLIP-based dual-encoder backbone and introduces two key components to address the temporal and cross-modal challenges. First, a temporal transformer operates exclusively on the video side, modeling inter-frame dependencies through learnable positional encodings and multi-head self-attention over frame-level CLIP features. Second, a two-stage reranking pipeline first retrieves Top-K candidates via the dual-encoder, then refines their scores using a cross-encoder equipped with an Image-Text Matching (ITM) head. The entire system is trained with Symmetric Multi-Similarity Loss to exploit the soft-label relevance matrices provided by the challenge. Our method achieves 67.97% average mAP and 82.92% average nDCG on the EK-100 MIR benchmark, demonstrating the effectiveness of temporal modeling and cross-modal refinement for egocentric video retrieval.
arXiv:2605.17116v2 Announce Type: replace
Abstract: Post-Quantum cryptography is about to substitute current cryptographic schemes as being resilient in attacks from quantum computers. McEleiece and Bit Flip Key Encapsulation (BIKE) are two delight representatives based on coding theory where classical structural attacks against these algorithms can be successfully phased out by selecting the appropriate key size. Using low cost equipment, the method of Simple Power Analysis (SPA) is used in this paper to evaluate whether or not there is significant information leakage during the decapsulation phase where the shared secret key is generated. Executing a related experiment it is shown that correlation between electromagnetic emissions and secret values exists. In the aftermath, with only 200 power traces collected, machine learning models can predict secret bits of the shared session key, produced during the decapsulation.
arXiv:2606.05976v1 Announce Type: new
Abstract: Recent work shows that LLM agents struggle to correct errors in their own reasoning traces yet show markedly higher correction rates when identical claims appear under external sources. We ask whether this asymmetry reflects a capability deficit or a role-label artifact: does an agent's willingness to correct a wrong claim depend causally on the chat-template role that carries it, rather than on the claim's content? Our setup keeps the erroneous claim byte-identical across all conditions (SHA-256 verified) and varies only its wrapping role: the agent's own \role{<thought>}, a \role{user} message, a \role{tool} response, or a \role{system <memory>} block. Across 13 model-domain cells covering seven model families and three domains ($n{=}30$ paired tasks per cell), relabeling the claim from \role{<thought>} to an external role lifts the explicit-correction rate by 23 to 93 percentage points, with 10 of 13 cells reaching $p{<}0.001$. Further experiments confirm that the effect is asymmetric, mechanistically decomposable, and robust across domains. The failure to self-correct is not a cognitive deficit; it is a chat-template artifact. We exploit this artifact by designing a prompt-structure-only intervention that requires no training and no model modification, with its strongest role label being domain-dependent: \role{<memory>} dominates on math, while a plain \role{user} message dominates on logical deduction.
arXiv:2605.08318v2 Announce Type: replace
Abstract: We study the problem of \emph{architecture selection} for deep learning models trained to solve partial differential equations (PDEs), asking when transformer-based architectures with learned attention outperform Fourier-domain neural operators. We introduce the \textbf{Multi-Scale Attention Transformer} (\msat{}), a deep learning architecture that encodes spatiotemporal solution histories as token sequences and trains end-to-end via a composite supervised objective with optional physics-informed regularization terms. We conduct a comprehensive empirical evaluation against nine baselines -- including physics-informed neural networks (PINNs), neural operators (FNO, DeepONet, GNOT), and state-space models (Mamba-NO) -- across five benchmark problems from the PINNacle suite, using identical train/test splits and reference data for all methods. \msat{} achieves state-of-the-art generalization on complex geometry problems ($L^2_\mathrm{rel} = 0.0101$ on Heat2D-CG, a $3.7\times$ improvement over FNO) at $34\,\mathrm{s}$ total inference vs.\ $120{,}812\,\mathrm{s}$ for Mamba-NO. Ablation studies over the physics regularization component reveal a precise inductive bias tradeoff: physics priors reduce test error on diffusion-dominated problems but degrade generalization on chaotic and recirculating-flow regimes, directly characterizing the prior misspecification boundary. Approximation error bounds as a function of domain boundary complexity $\kappa$ provide a theoretical basis for these empirical findings and a principled rule for architecture selection.
arXiv:2605.24059v2 Announce Type: replace
Abstract: We present a three-step recipe for identifying attention-head circuits in pretrained transformers. A per-head spectral signal -- the time-integrated participation ratio of each head's attention output -- ranks heads doing sustained content-dependent computation without labels or attribution gradients. A task-pattern screen filters this general indicator into a task-specific candidate circuit, and group ablation against a matched-random control completes the causal claim. We validate across an 8x parameter range (51M to 1B-active / 7B-total), two architecture families (dense, mixture-of-experts), and four pretraining pipelines. The recipe ports: a 2-6 head induction circuit is causally necessary in every model tested, with a 94-100% drop in synthetic-induction top-1 after ablation. The spectral signal is predictive without supervision: on six independent seeds of a 51M-parameter probe model, the same computation identifies the seed-specific circuit on each seed. The fraction of heads doing identifiable specialized computation is conserved at 17-19% across the Pythia family (124M to 410M), while specific induction circuits stay 3-11 heads -- sublinear in total head count. This paper is the methodology anchor of a three-paper program; companion papers extend the recipe to developmental trajectories during pretraining and to composed-task circuits where pattern selectivity decouples from task-causal structure.
arXiv:2506.11152v4 Announce Type: replace-cross
Abstract: Single-cell transcriptomics and proteomics have become a great source for data-driven insights into biology, enabling the use of advanced deep learning methods to understand cellular heterogeneity and gene expression at the single-cell level. With the advent of spatial-omics data, we have the promise of characterizing cells within their tissue context as it provides both spatial coordinates and intra-cellular transcriptional or protein counts. Proteomics offers a complementary view by directly measuring proteins, which are the primary effectors of cellular function and key therapeutic targets. However, existing models either ignore the spatial information or the complex genetic and proteomic programs within cells. Thus they cannot infer how cell internal regulation adapts to microenvironmental cues. Furthermore, these models often utilize fixed gene vocabularies, hindering their generalizability unseen genes. In this paper, we introduce HEIST, a hierarchical graph transformer foundation model for spatial transcriptomics and proteomics. HEIST models tissues as hierarchical graphs. The higher level graph is a spatial cell graph, and each cell in turn, is represented by its lower level gene co-expression network graph. HEIST achieves this by performing both intra-level and cross-level message passing to utilize the hierarchy in its embeddings and can thus generalize to novel datatypes including spatial proteomics without retraining. HEIST is pretrained on 22.3M cells from 124 tissues across 15 organs using spatially-aware contrastive and masked autoencoding objectives. Unsupervised analysis of HEIST embeddings reveals spatially informed subpopulations missed by prior models. Downstream evaluations demonstrate generalizability to proteomics data and state-of-the-art performance in clinical outcome prediction, cell type annotation, and gene imputation across multiple technologies.
arXiv:2604.01349v4 Announce Type: replace
Abstract: Reservoir simulation workflows face a fundamental data asymmetry: input parameter fields (geostatistical permeability realizations, porosity distributions) are free to generate in arbitrary quantities, yet existing neural operator surrogates require large corpora of expensive labeled simulation trajectories and cannot exploit this unlabeled structure. We introduce \textbf{PI-JEPA} (Physics-Informed Joint Embedding Predictive Architecture), a surrogate pretraining framework that trains \emph{without any completed PDE solves}, using masked latent prediction on unlabeled parameter fields under per-sub-operator PDE residual regularization. The predictor bank is structurally aligned with the Lie--Trotter operator-splitting decomposition of the governing equations, dedicating a separate physics-constrained latent module to each sub-process (pressure, saturation transport, reaction), enabling fine-tuning with as few as 100 labeled simulation runs. On single-phase Darcy flow, PI-JEPA achieves $1.9\times$ lower error than FNO and $2.4\times$ lower error than DeepONet at $N_\ell{=}100$, with 24\% improvement over supervised-only training at $N_\ell{=}500$, demonstrating that label-free surrogate pretraining substantially reduces the simulation budget required for multiphysics surrogate deployment.
arXiv:2506.12308v4 Announce Type: replace-cross
Abstract: In this article, we introduce a novel low-altitude wireless network (LAWN), which is a reconfigurable, three-dimensional (3D) layered architecture. In particular, the LAWN integrates connectivity, sensing, control, and computing across aerial and terrestrial nodes that enable seamless operation in complex, dynamic, and mission-critical environments. Different from the conventional aerial communication systems, LAWN's distinctive feature is its tight integration of functional planes in which multiple functionalities continually reshape themselves to operate safely and efficiently in the low-altitude sky. With the LAWN, we discuss several enabling technologies, such as integrated sensing and communication (ISAC), semantic communication, and fully-actuated control systems. Finally, we identify potential applications and key cross-layer challenges. This article offers a comprehensive roadmap for future research and development in the low-altitude airspace.
arXiv:2605.01844v2 Announce Type: replace
Abstract: Steering is a widely used technique for controlling large language models, yet its effects are often unstable and hard to predict. Existing theoretical accounts are largely based on the Linear Representation Hypothesis (LRH). While LRH assumes that concepts can be orthogonalized for lossless control, this idealized mapping fails in real representations and cannot account for the observed unpredictability of steering. By relaxing LRH's orthogonality assumption while preserving linear representations, we show that overlapping concept contributions naturally yield a sample-specific axis-orthogonal structure. We formalize this as the Cylindrical Representation Hypothesis (CRH). In CRH, a central axis captures the main difference between concept absence and presence and drives concept generation. A surrounding normal plane controls steering sensitivity by determining how easily the axis can activate the target concept. Within this plane, only specific sensitive sectors strongly facilitate concept activation, while other sectors can suppress or delay it. While the surrounding normal plane can be reliably identified from difference vectors, the sensitive sector cannot, introducing intrinsic uncertainty at the sector level. This uncertainty provides a principled explanation for why steering outcomes often fluctuate even when using well-aligned directions. Our experiments verify the existence of the cylindrical structure and demonstrate that CRH provides a valid and practical way to interpret model steering behavior in real settings: https://github.com/mbzuai-nlp/CRH.
arXiv:2605.26179v2 Announce Type: replace-cross
Abstract: Density functional theory (DFT) serves as the basis for computational discovery in materials science and chemistry, yet each calculation demands extensive human effort: adjusting algorithms when convergence stalls, revising plans when unexpected physics emerges, and inserting steps as intermediate results reshape the problem. Existing LLM-based agents automate only the initial planning stage, producing a full execution plan upfront and leaving all subsequent adaptation to hand-crafted rules. As a result, these workflows remain fragile, do not generalize well beyond pre-planned scenarios, and often require expert intervention when failures or unexpected intermediate results require changes to the calculation path. Here, we introduce AutoDFT, a closed-loop multi-agent framework that embeds LLM reasoning into every stage of the DFT lifecycle, where a strategic planner produces a skeletal plan of step objectives; a step planner generates numerical parameters just in time from preceding results; and a monitor-recover-reflect cycle diagnoses failures, repairs them, and revises the plan when the evidence justifies it. We demonstrate both breadth and depth: breadth on VASPBench, a purpose-built benchmark spanning 34 tasks and 9 DFT calculation types, where AutoDFT achieves 94.1% task-level success with GPT-5.2; and depth on established materials databases, where AutoDFT produces quantitatively reliable property predictions across electronic, magnetic, and energetic properties. By closing the loop between planning and execution, AutoDFT enables experimentalists without deep computational expertise to obtain reliable first-principles results.
arXiv:2605.24175v2 Announce Type: replace
Abstract: Ultracold neutral atoms in an optical lattice and an optical tweezer array offer highly-controllable quantum many-body systems, utilized for various quantum science and technology such as quantum computing, quantum metrology, and quantum simulation. By combining high-fidelity imaging of individual atoms, one can further enhance the capability of such experimental platforms as quantum gas microscopes, tweezer clocks, and tweezer-array-based quantum computers. In this work, we propose a minimally destructive single-atom imaging by deterministic coherent excitation of atoms with alternately applied pi-pulses from counter-propagating directions, mitigating the fundamental heating effect associated with the stochastic absorption process. Using ytterbium-174 atoms trapped in an optical tweezer array, we experimentally demonstrate fast and low-loss single-atom imaging with a discrimination fidelity of 99.89(5) % and a survival probability of 98.80(44) % in 17.6 microseconds. Importantly, our scheme exhibits the lower heating rate, about half of that of the former scheme utilizing the incoherent excitation. This fast and minimally destructive imaging scheme is beneficial for relaxing the requirement on the trap depth, thereby enabling scalable atom imaging across a wide range of quantum science platforms.
arXiv:2506.22728v2 Announce Type: replace
Abstract: Let $w$ be a string of length $n$. The problem of counting factors crossing a position -- Problem 64 from the textbook ``125 Problems in Text Algorithms'' [Crochemore, Lecroq, and Rytter, 2021] -- asks to count the number $\mathcal{C}(w,k)$ (resp. $\mathcal{N}(w,k)$) of distinct substrings in $w$ that have occurrences containing (resp. not containing) a position $k$ in $w$. The solutions provided in their textbook compute $\mathcal{C}(w,k)$ and $\mathcal{N}(w,k)$ in $O(n)$ time for a single position $k$ in $w$, and thus a direct application would require $O(n^2)$ time for all positions $k = 1, \ldots, n$ in $w$. Their solution is designed for constant-size alphabets. In this paper, we present new algorithms which compute $\mathcal{C}(w,k)$ in $O(n)$ total time for general ordered alphabets, and $\mathcal{N}(w,k)$ in $O(n)$ total time for linearly sortable alphabets,for all positions $k = 1, \ldots, n$ in $w$. We further derive model-dependent optimal bounds by separating the algorithms into preprocessing and linear-time postprocessing: for $\mathcal{C}$ the preprocessing is run reporting, and for $\mathcal{N}$ it is preprocessing based on longest previous non-overlapping factors (LPnF) and longest next factors (LNF). In particular, all values $\mathcal{C}(w,k)$ can be computed in $O(n\log n)$ time over general unordered alphabets in which direct accesses to alphabet characters are restricted to equality tests, and in $O(n\log\sigma)$ time in the word RAM model, where $\sigma$ denotes the number of distinct characters occurring in $w$. For $\mathcal{N}(w,k)$, the equality-testing complexity over general unordered alphabets is $\Theta(n^2)$. We also show that our upper bounds are optimal for all of the aforementioned alphabet assumptions and computation models.
arXiv:2512.15153v2 Announce Type: replace
Abstract: Evaluating whether human action is standard or not and providing reasonable feedback to improve action standardization is very crucial but challenging in real-world scenarios. However, current video understanding methods are mainly concerned with what and where the action is, which is unable to meet the requirements. Meanwhile, most of the existing datasets lack the labels indicating the degree of action standardization, and the action quality assessment datasets lack explainability and detailed feedback. Therefore, we define a new Human Action Form Assessment (AFA) task, and introduce a new diverse dataset CoT-AFA, which contains a large scale of fitness and martial arts videos with multi-level annotations for comprehensive video analysis. We enrich the CoT-AFA dataset with a novel Chain-of-Thought explanation paradigm. Instead of offering isolated feedback, our explanations provide a complete reasoning process--from identifying an action step to analyzing its outcome and proposing a concrete solution. Furthermore, we propose a framework named Explainable Fitness Assessor, which can not only judge an action but also explain why and provide a solution. This framework employs two parallel processing streams and a dynamic gating mechanism to fuse visual and semantic information, thereby boosting its analytical capabilities. The experimental results demonstrate that our method has achieved improvements in explanation generation (e.g., +16.0% in CIDEr), action classification (+2.7% in accuracy) and quality assessment (+2.1% in accuracy), revealing great potential of CoT-AFA for future studies. Our dataset and source code is available at https://github.com/MICLAB-BUPT/EFA.
arXiv:2606.05977v1 Announce Type: new
Abstract: We consider the inverse medium scattering problem for the Helmholtz equation in two dimensions, i.e., the task to recover a compactly supported penetrable two-dimensional scatterer from full knowledge of the associated far field data or, equivalently, the far field operator. Although this problem is uniquely solvable, it is severely ill-posed and nonlinear. In the regime of weak scattering, the Born approximation yields a linearized relation between the contrast and the far field data, thus overcoming the second difficulty. This linear setting allows to build on recent work on linearized electrical impedance tomography, which relies on triangular Zernike decompositions, to derive an explicit reconstruction formula that expresses the expansion coefficients of the contrast in terms of those of the far field data. By choosing the expansion functions appropriately, the resulting system matrix decouples into separate (infinite) triangular systems for the spatial angular frequencies in the contrast. Consequently, each of these systems can be solved independently by performing forward substitutions. Our numerical experiments indicate that this approach, combined with an adequate regularization method, remains effective even when applied to full nonlinear far field data beyond the Born regime.
arXiv:2605.27887v2 Announce Type: replace
Abstract: Large language models (LLMs) have shown strong performance across diverse financial tasks, yet portfolio management (PM), a critical financial decision-making task, remains poorly benchmarked. Existing benchmarks exhibit two main gaps: they ignore cross-asset correlation structures, thereby failing to distinguish genuinely diversified portfolios from concentrated ones, and fail to evaluate the complete PM decision pipeline in real-world scenarios. We introduce PortBench, a benchmark spanning six heterogeneous asset classes over ten years. PortBench consists of two complementary layers: a static QA dataset of 6,269 correlation-based questions across seven task templates, and a dynamic five-stage allocation pipeline that mirrors the full PM decision cycle. To evaluate these layers, we introduce two dedicated metrics: a dual-layer correlation score that measures whether proposed portfolios exploit inter-class hedging and avoid intra-class concentration, and CEPS, a metric that quantifies how reasoning errors compound across pipeline stages. We further assess strategy robustness and investor alignment under three historical stress regimes and risk profiles. Evaluating ten frontier LLMs, we find that despite strong performance on static financial QA, 90\% of model-profile combinations fail to outperform a basic equal-weight allocation, and models that satisfy every procedural constraint still suffer catastrophic drawdowns under stress. Our source code is available at \href{https://github.com/AgenticFinLab/portbench}{this https URL}.
arXiv:2605.21557v2 Announce Type: replace-cross
Abstract: Conventional wisdom holds that large-batch training is fundamentally incompatible with Reinforcement Learning (RL) - beyond a modest threshold, increasing batch sizes typically yields diminishing returns or performance degradation due to the inherent non-stationarity of the data distribution. We challenge this view by observing that non-stationarity is not a fixed property of RL, but evolves throughout training: early stages exhibit rapid behavioral shifts that demand small batches for plasticity, whereas late stages approach a quasi-stationary regime where large batches enable precise convergence. Motivated by this observation, we propose Adaptive Batch Scaling (ABS), that dynamically adjusts the effective batch size according to the stability of the learning policy. Central to ABS is Behavioral Divergence, a novel metric that quantifies policy non-stationarity by measuring action-level shifts between consecutive updates, which we use to scale batch size inversely to policy volatility. Integrated with the Parallelised Q-Network (PQN) algorithm and evaluated on the ALE benchmark, ABS seamlessly reconciles early-stage plasticity with late-stage stable convergence. Strikingly, contrary to conventional wisdom, our results reveal that the combination of larger networks and larger batch sizes achieves the best performance - a scaling behavior previously thought to be unattainable in RL, now unlocked through adaptive batch control.
arXiv:2605.02192v2 Announce Type: replace
Abstract: Should a single collision necessarily terminate an entire navigation episode? In most deep reinforcement learning (DRL) frameworks for robot navigation, this remains the standard practice: every collision immediately triggers a global environment reset and is penalized as a complete task failure. While a collision during deployment naturally indicates task failure, applying the same treatment during training prevents the agent from exploring challenging obstacle configurations, which slows learning progress in the early training phase. In this work, we challenge this convention and propose a Multi-Collision reset Budget (MCB) framework that decouples local collision termination from global environment resets, allowing the agent to retry difficult configurations within the same episode. Simulation experiments show that MCB improves early-stage learning efficiency by reaching target success-rate levels with fewer interactions, with small collision budgets producing the most consistent gains. Real-world experiments on heterogeneous robot platforms further validate the deployability of the learned policies in cluttered environments.
arXiv:2606.06434v1 Announce Type: cross
Abstract: Restriction site-associated DNA sequencing (RAD-seq) is widely used to discover sex-linked markers in non-model organisms, but large studies produce marker tables with millions of RAD tags. RADSex provides the reference workflow for building marker-by-individual depth tables and testing sex-biased marker distributions, but its depth, merge, and related table-building commands grow memory-hungry, and its standard output reports frequentist calls with no posterior evidence and no direct Python or C integration. We present rsx, a Rust implementation of the complete RADSex command set that preserves marker-table semantics and command-line compatibility. rsx combines 2-bit DNA keys, parallel ingestion, memory-mapped marker tables, external sorting, bitset group counts, and streamed Gram-matrix PCA so that memory stays bounded by the number of individuals or by explicit buffers. It adds conjugate Beta-Binomial Bayes factors and posterior probabilities under XY and ZW hypotheses, returning strict, posterior-supported, and Bayes-factor-only evidence grades. A portable, libm-independent minimax approximation of the error function keeps the chi-squared tail reproducible across platforms without changing the underlying Yates test. On four real RAD-seq datasets comprising 41.9 billion bases and 29 million markers, rsx reproduced published RADSex v1.2.0 calls, achieved an 8.38-fold geometric-mean speedup across 56 paired timings (2.77-fold for FASTQ processing), and recovered every Bonferroni-significant positive-control marker. In Danio albolineatus, treated as null in the source publication, the posterior layer surfaced 30 W-linked marker hypotheses; in Notothenia rossii it withheld 400 Bayes-factor-only rows compatible with a low-prevalence null. Python bindings, a C API, and a reproducibility archive provide the workflows used for all reported numbers. rsx is released under GPL-3.0-or-later.
arXiv:2606.05957v1 Announce Type: new
Abstract: Singular learning theory and information geometry have studied the same parameter spaces in mostly separate vocabularies: the former computes Bayesian invariants in resolved coordinates, the latter works in original coordinates under a non-degeneracy assumption that overparameterised models routinely violate. We bridge them through one primitive, the dead direction: a unit vector along which the Fisher metric degenerates, equivalently a tangent to the analytic singular set with a definite KL order, set by how fast the KL divergence vanishes. The two readings name the same vector; our central move shows its KL order is recoverable as the decay rate of the directional Fisher curvature approaching the singularity, in original parameter coordinates and without a Hironaka resolution. A selection rule on smooth fibres translates this rate into Watanabe's single-direction contribution to the real log canonical threshold, and we extend the recovery to multi-component crossings, multiplicity $m$, the singular fluctuation $\nu$ (universal in the KL order for 1D directions), prior-RLCT shifts, and tempered posteriors. We then lift this rate to a deep network: a multi-layer K-FAC factorisation writes each Fisher block as a product of activation- and gradient-side rates with a duality between them, instantiated at modern-network primitives (residual streams, layer normalisation, attention). A quotient theorem carries the rate to the gauge quotient $\Theta/G$ under gradient flow on a $G$-invariant metric; SGD qualifies, standard Adam does not, and we construct a $G$-equivariant Adam-family preconditioner (DDCAdam) that does. The bridge yields a parameter-coordinate handle on singular geometry, closed-form per-architecture predictions, and a trajectory-rate readout of Watanabe's triple $(\lambda, m, \nu)$ from one checkpoint's forward and backward passes, without posterior sampling.
arXiv:2606.05795v1 Announce Type: new
Abstract: We propose an AI Agent tailored for link power management in multi-band systems. In S+C+L band span-level study, the agent efficiently solves various optimization objectives. In network-wide evaluation, it delivers 689.0 Tbps gain in total allocated traffic with merely 303 average interactions per power profile.
arXiv:2606.05937v1 Announce Type: new
Abstract: Large Language Models (LLMs) are increasingly used, including in political applications, but their political fairness has been little studied. We assess it using perplexity, posing that a fair model should give equal probability to all political groups. However, we find, across ten LLMs and three datasets covering 37 languages, that LLMs are more perplexed by the texts of far right and nationalist parties than of social-democratic parties. We find this to be consistent with previous work on translation fairness, to the point that perplexity correlates with downstream translation metrics. Our method is applicable to both base LLMs as well as their instruction-tuned counterpart, and we find that both are highly correlated, suggesting that the political fairness of LLMs stems from their pretraining, and is hardly affected by instruction-tuning.
arXiv:2503.10133v2 Announce Type: replace
Abstract: This paper deals with shape irregularity issues in discrete topology optimization algorithms whereby the design is created using the automated distribution of material in the design region. Graph theory is employed to derive appropriate regularity measures for any discrete optimization algorithm. Shape regularity is quantified by scalar figures ready to evaluate design choices in the form of Pareto-frontiers. Developed metrics deal with information concerning material usage, problematic distribution, and features that complicate manufacturing. The theory is verified by several examples demonstrating the treatment of isolated islands of materials, point connections between material segments, or homogeneity.
arXiv:2503.22929v3 Announce Type: replace
Abstract: Face anti-spoofing (FAS) techniques aim to enhance the security of facial identity authentication by distinguishing authentic live faces from deceptive attempts. While two-class FAS methods risk overfitting to training attacks to achieve better performance, one-class FAS approaches handle unseen attacks well but are less robust to domain information entangled within the liveness features. To address this, we propose an Unsupervised Feature Disentanglement and Augmentation Network (\textbf{UFDANet}), a one-class FAS technique that enhances generalizability by augmenting face images via disentangled features. The \textbf{UFDANet} employs a novel unsupervised feature disentangling method to separate the liveness and domain features, facilitating discriminative feature learning. It integrates an out-of-distribution liveness feature augmentation scheme to synthesize new liveness features of unseen spoof classes, which deviate from the live class, thus enhancing the representability and discriminability of liveness features. Additionally, \textbf{UFDANet} incorporates a domain feature augmentation routine to synthesize unseen domain features, thereby achieving better generalizability. Extensive experiments demonstrate that the proposed \textbf{UFDANet} outperforms previous one-class FAS methods and achieves comparable performance to state-of-the-art two-class FAS methods.