Search

Accurate detection and resilience of object detectors in structural damage detection are important in ensuring the continuous use of civil infrastructure. However, achieving robustness in object detectors remains a persistent challenge, impacting their ability to generalize effectively. This study proposes DetectorX, a robust framework for structural damage detection coupled with a micro drone. DetectorX addresses the challenges of object detector robustness by incorporating two innovative modules: a stem block and a spiral pooling technique. The stem block introduces a dynamic visual modality by leveraging the outputs of two Deep Convolutional Neural Network (DCNN) models. The framework employs the proposed event-based reward reinforcement learning to constrain the actions of a parent and child DCNN model leading to a reward. This results in the induction of two dynamic visual modalities alongside the Red, Green, and Blue (RGB) data. This enhancement significantly augments DetectorX's perception and adaptability in diverse environmental situations. Further, a spiral pooling technique, an online image augmentation method, strengthens the framework by increasing feature representations by concatenating spiraled and average/max pooled features. In three extensive experiments: (1) comparative and (2) robustness, which use the Pacific Earthquake Engineering Research Hub ImageNet dataset, and (3) field-experiment, DetectorX performed satisfactorily across varying metrics, including precision (0.88), recall (0.84), average precision (0.91), mean average precision (0.76), and mean average recall (0.73), compared to the competing detectors including You Only Look Once X-medium (YOLOX-m) and others. The study's findings indicate that DetectorX can provide satisfactory results and demonstrate resilience in challenging environments.

Multivariate anomaly detection finds its importance in diverse applications. Despite the existence of many detectors to solve this problem, one cannot simply define why an obtained anomaly inferred by the detector is anomalous. This reasoning is required for network operators to understand the root cause of the anomaly and the remedial action that should be taken to counteract its occurrence. Existing solutions in explainable AI may give cues to features that influence an anomaly, but they do not formulate generalizable rules that can be assessed by a domain expert. Furthermore, not all outliers are anomalous in a business sense. There is an unfulfilled need for a system that can interpret anomalies predicted by a multivariate anomaly detector and map these patterns to actionable rules. This paper aims to fulfill this need by proposing a semi-autonomous anomaly rule miner. The proposed method is applicable to both discrete and time series data and is tailored for radio access network (RAN) anomaly detection use cases. The proposed method is demonstrated in this paper with time series RAN data.

Increasing educational attainment is one of the most important and effective tools for health and economic improvements. The extent to which extreme climate events disrupt education, resulting in fewer years of schooling and reduced educational attainment, remains under-studied. Children in low- and middle-income countries may be uniquely vulnerable to loss of schooling after such disasters due to the poor physical condition of schools and the lack of resources to rebuild and mitigate unexpected household shocks. Our analysis assesses this overlooked social cost of tropical cyclones on schooling attainment. We study the education records of nearly 5.1 million people living in 13 low- and middle-income countries that were exposed to tropical cyclones between 1954-2010. We find that exposure to tropical cyclones during preschool age is associated with a 2.7 percentage point decrease in primary school enrollment on average (14.2% decrease), with larger effects from more intense storms (up to 28% decrease for the most intense storms). These effects are more pronounced among school-age girls compared to boys and are greater in areas less accustomed to experiencing tropical cyclones. We estimate that, across all LMICs, tropical cyclone exposure has resulted in more than 410,000 children not attending primary school in the last 20 years, leading to a reduction of more than 4.1 million total years of schooling. These impacts, identified among some of the world's poorest populations, may grow in importance as exposure to severe tropical cyclones is projected to increase with climate change.

Most students who begin at a community college do not complete their desired credential. Many former students fail to graduate due to various barriers rather than their academic performance. To encourage previously successful non-completers to re-enroll and eventually graduate, a growing number of community colleges have implemented re-enrollment campaigns focused on former students who have already made substantial progress toward graduation. In this study, we randomly assigned over 27,000 former community college students to a control group, "information-only" treatment group, or "information and one-course waiver" treatment group to examine whether re-enrollment campaigns can improve their likelihood of long-term persistence and credential completion. Although we showed in earlier work that the "information and one-course waiver" treatment had a positive impact on former students' likelihood of re-enrollment, our findings reveal the re-enrollment intervention has no effect on students' likelihood of long-term persistence or credential completion for the pooled sample or any subgroup of interest, including low-income students, racially minoritized students, or adult students. Simply put, this particular re-enrollment intervention including one-time, one-course tuition waivers increased former students' likelihood of re-enrollment but was not an effective lever to increase "long-term" academic outcomes among previously successful community college students who departed early without earning a credential.

In an $[n] \times [n]$ integer grid, a monochromatic $L$ is any set of points $\{(i, j), (i, j+t), (i+t, j+t)\}$ for some positive integer $t$, where $1 \leq i, j, i+t, j+t \leq n$. In this paper, we investigate the upper bound for the smallest integer $n$ such that a $3$-colored $n \times n$ grid is guaranteed to contain a monochromatic $L$. We use various methods, such as counting intervals on the main diagonal and using Golomb rulers, to improve the upper bound. This bound originally sat at 2593, and we improve it first to 1803, then to 1573, then to 772, and finally to 493. In the latter part of this paper, we discuss the lower bound and our attempts to improve it using SAT solvers., Comment: This version of the paper includes discussion on the lower bound and SAT solvers that was not deemed relevant for the main paper

Linkography -- the analysis of links between the design moves that make up an episode of creative ideation or design -- can be used for both visual and quantitative assessment of creative activity traces. Traditional linkography, however, is time-consuming, requiring a human coder to manually annotate both the design moves within an episode and the connections between them. As a result, linkography has not yet been much applied at scale. To address this limitation, we introduce fuzzy linkography: a means of automatically constructing a linkograph from a sequence of recorded design moves via a "fuzzy" computational model of semantic similarity, enabling wider deployment and new applications of linkographic techniques. We apply fuzzy linkography to three markedly different kinds of creative activity traces (text-to-image prompting journeys, LLM-supported ideation sessions, and researcher publication histories) and discuss our findings, as well as strengths, limitations, and potential future applications of our approach.

Having the ability to infer characteristics of autonomous agents would profoundly revolutionize defense, security, and civil applications. Our previous work was the first to demonstrate that supervised neural network time series classification (NN TSC) could rapidly predict the tactics of swarming autonomous agents in military contexts, providing intelligence to inform counter-maneuvers. However, most autonomous interactions, especially military engagements, are fraught with uncertainty, raising questions about the practicality of using a pretrained classifier. This article addresses that challenge by leveraging expected operational variations to construct a richer dataset, resulting in a more robust NN with improved inference performance in scenarios characterized by significant uncertainties. Specifically, diverse datasets are created by simulating variations in defender numbers, defender motions, and measurement noise levels. Key findings indicate that robust NNs trained on an enriched dataset exhibit enhanced classification accuracy and offer operational flexibility, such as reducing resources required and offering adherence to trajectory constraints. Furthermore, we present a new framework for optimally deploying a trained NN by the defenders. The framework involves optimizing defender trajectories that elicit adversary responses that maximize the probability of correct NN tactic classification while also satisfying operational constraints imposed on the defenders.

Maintaining the integrity of solar power plants is a vital component in dealing with the current climate crisis. This process begins with analysts creating a detailed map of a plant with the coordinates of every solar panel, making it possible to quickly locate and mitigate potential faulty solar panels. However, this task is extremely tedious and is not scalable for the ever increasing capacity of solar power across the globe. Therefore, we propose an end-to-end deep learning framework for detecting individual solar panels using a rotated object detection architecture. We evaluate our approach on a diverse dataset of solar power plants collected from across the United States and report a mAP score of 83.3%.

Hyperentanglement, which refers to entanglement across more than one degree of freedom (DoF), is a valuable resource in photonic quantum information technology. However, the lack of efficient characterization schemes hinders its quantitative study and application potential. Here, we present a rapid quantitative characterization of spatial-polarization hyperentangled biphoton state produced from spontaneous parametric down-conversion. We first demonstrate rapid certification of the hyperentanglement dimensionality with a cumulative acquisition time of only 17 minutes. In particular, we verify transverse spatial entanglement through a violation of the Einstein-Podolsky-Rosen criterion with a minimum conditional uncertainty product of $(0.11\pm0.05)\hbar$ and certify the entanglement dimensionality to be at least 148. Next, by performing spatially-resolved polarization state tomography of the entire field, we demonstrate the generation of an entire class of near-maximally polarization-entangled states with an average concurrence of $0.8303\pm0.0004$. Together, the results reveal a total dimensionality of at least 251, which is the highest dimensionality reported for hyperentanglement. These measurements quantitatively resolve the influence of the spatial correlations of the down-converted photons and the angular spectrum of the pump beam on the polarization entanglement. Our study lays important groundwork for further exploiting the high dimensionality and cross-DoF correlations in hyperentangled states for future quantum technologies.

The aim of this article is to give several results related to Walsh's spider diffusions living on a star-shaped network that have a spinning measure selected from the own local time of the motion at the vertex (cf.[17]). We prove the corresponding It\^o's formula and give some global trajectory properties such as $L^1$-approximation of the local time and the Markov property. Regarding the behavior of the process at the vertex, we show that that the distribution of the process is non atomic at the junction point and we characterize the instantaneous scattering distribution along some ray with the aid of the probability coefficients of diffraction. We obtain also a Feynmann-Kac representation for linear parabolic systems posed on star-shaped networks that where introduced in [18] possessing a so-called local-time Kirchhoff's boundary condition.

Designing novel materials for impact applications and predicting material behavior using data-driven techniques require extensive datasets, which can be challenging to obtain through conventional experimental methods. While high throughput characterization is feasible at the sample micro and nanoscale, there remains a critical need for an experimental tool capable of facilitating high-throughput measurements at the macroscale. This paper presents the design and development of a fully automated split Hopkinson pressure bar (SHPB) with full-field diagnostics for performing dynamic compression experiments on different materials. The automated SHPB experiments consist of four main components: 1) striker launch and retrieval system, 2) bar repositioning mechanism, 3) automated sample placement system, and 4) diagnostics. Each of these systems employs various electromechanical devices and actuators programmed to work collaboratively to automate the SHPB experimental setup. A data analysis tool has been developed to automate the post-processing of the data obtained from the setup. The setup also incorporates automated high-speed imaging, enabling full-field strain measurement capabilities. To benchmark the setup, fully automated dynamic compression experiments were conducted on 45 Copper 101 samples and 20 3D-printed resin samples. The stress-strain curve was extracted from the raw data using the automated data analysis tool. The efficiency of the data analysis tool is validated by comparing the data with existing data analysis tools for SHPB. The stress-strain response obtained from the copper experiments aligned well with existing data on copper. Furthermore, the full-field strain measurements in experiments show strain values comparable to those of the strain gauge measurements. This advancement has enabled dynamic compression experiments on materials at a rate of 60 samples per hour.

We introduce STRING: Separable Translationally Invariant Position Encodings. STRING extends Rotary Position Encodings, a recently proposed and widely used algorithm in large language models, via a unifying theoretical framework. Importantly, STRING still provides exact translation invariance, including token coordinates of arbitrary dimensionality, whilst maintaining a low computational footprint. These properties are especially important in robotics, where efficient 3D token representation is key. We integrate STRING into Vision Transformers with RGB(-D) inputs (color plus optional depth), showing substantial gains, e.g. in open-vocabulary object detection and for robotics controllers. We complement our experiments with a rigorous mathematical analysis, proving the universality of our methods., Comment: Videos of STRING-based robotics controllers can be found here: https://sites.google.com/view/string-robotics

Plasticity loss, a critical challenge in neural network training, limits a model's ability to adapt to new tasks or shifts in data distribution. This paper introduces AID (Activation by Interval-wise Dropout), a novel method inspired by Dropout, designed to address plasticity loss. Unlike Dropout, AID generates subnetworks by applying Dropout with different probabilities on each preactivation interval. Theoretical analysis reveals that AID regularizes the network, promoting behavior analogous to that of deep linear networks, which do not suffer from plasticity loss. We validate the effectiveness of AID in maintaining plasticity across various benchmarks, including continual learning tasks on standard image classification datasets such as CIFAR10, CIFAR100, and TinyImageNet. Furthermore, we show that AID enhances reinforcement learning performance in the Arcade Learning Environment benchmark.

While conventional Transformers generally operate on sequence data, they can be used in conjunction with structure models, typically SE(3)-invariant or equivariant graph neural networks (GNNs), for 3D applications such as protein structure modelling. These hybrids typically involve either (1) preprocessing/tokenizing structural features as input for Transformers or (2) taking Transformer embeddings and processing them within a structural representation. However, there is evidence that Transformers can learn to process structural information on their own, such as the AlphaFold3 structural diffusion model. In this work we show that Transformers can function independently as structure models when passed linear embeddings of coordinates. We first provide a theoretical explanation for how Transformers can learn to filter attention as a 3D Gaussian with learned variance. We then validate this theory using both simulated 3D points and in the context of masked token prediction for proteins. Finally, we show that pre-training protein Transformer encoders with structure improves performance on a downstream task, yielding better performance than custom structural models. Together, this work provides a basis for using standard Transformers as hybrid structure-language models.

In this paper, we consider a new transmission eigenvalue problem derived from the scattering by a clamped cavity in a thin elastic material. Scattering in a thin elastic material can be modeled by the Kirchhoff--Love infinite plate problem. This results in a biharmonic scattering problem that can be handled by operator splitting. The main novelty of this transmission eigenvalue problem is that it is posed in all of $\mathbb{R}^2$. This adds analytical and computational difficulties in studying this eigenvalue problem. Here, we prove that the eigenvalues can be recovered from the far field data as well as discreteness of the transmission eigenvalues. We provide some numerical experiments via boundary integral equations to demonstrate the theoretical results. We also conjecture monotonicity with respect to the measure of the scatterer from our numerical experiments.

This work derives two basic engagement zone models, describing regions of potential risk or capture for a mobile vehicle by a pursuer. The pursuer is modeled as having turn-constraints rather than simple motion. Turn-only (C-Paths) and turn-straight (CS-Paths) paths are considered for the pursuer of limited range. Following the derivation, a simulation of a vehicle avoiding the pursuer's engagement zone is provided., Comment: 6 pages 11 figures

(ABRIDGED) We use galaxy-galaxy lensing to investigate how the dark matter (DM) haloes and stellar content of galaxies with $0.012 \leq z \leq 0.32$ and $10 \leq \log_{10}(M_\star/\mathrm{M}_\odot) \leq 12$ change as a result of the merger process. To this end, we construct two samples of galaxies obtained from the Ultraviolet Near Infrared Optical Northern Survey (UNIONS), comprising 1 623 post-mergers and $\sim$30 000 non-merging controls, that live in low-density environments to use as our lenses. These samples are weighted to share the same distributions of stellar mass, redshift, and geometric mean distance to a galaxy's three nearest neighbours to ensure differences in the lensing signal are due to the merger process itself. We do not detect a significant difference in the excess surface density profile of post-mergers and non-merging controls with current data. Fitting haloes composed of a point-like stellar mass component and an extended DM structure described by a Navarro-Frenk-White profile to the lensing measurements yields, for both samples, halo masses of $M_\text{halo}\sim 4\times10^{12}\,\mathrm{M}_\odot$ and a moderately negative correlation between $M_\text{halo}$ and concentration $c$. . . . Extreme merger-induced starbursts, in which more than 60 percent of the stars are formed in the burst, are ruled out at the 95 per cent confidence level. Although we do not find merger-induced differences in the DM haloes and stellar content of galaxies with our current data, application of our methods to upcoming surveys that are able to provide samples $\sim$10$\times$ larger than our current catalogue are expected to detect the weak-lensing signatures of mergers and further constrain their properties., Comment: 10 pages, 7 figures, 1 table. Submitted

Electronic Health Records (EHRs) hold immense potential for advancing healthcare, offering rich, longitudinal data that combines structured information with valuable insights from unstructured clinical notes. However, the unstructured nature of clinical text poses significant challenges for secondary applications. Traditional methods for structuring EHR free-text data, such as rule-based systems and multi-stage pipelines, are often limited by their time-consuming configurations and inability to adapt across clinical notes from diverse healthcare settings. Few systems provide a comprehensive attribute extraction for terminologies. While giant large language models (LLMs) like GPT-4 and LLaMA 405B excel at structuring tasks, they are slow, costly, and impractical for large-scale use. To overcome these limitations, we introduce GENIE, a Generative Note Information Extraction system that leverages LLMs to streamline the structuring of unstructured clinical text into usable data with standardized format. GENIE processes entire paragraphs in a single pass, extracting entities, assertion statuses, locations, modifiers, values, and purposes with high accuracy. Its unified, end-to-end approach simplifies workflows, reduces errors, and eliminates the need for extensive manual intervention. Using a robust data preparation pipeline and fine-tuned small scale LLMs, GENIE achieves competitive performance across multiple information extraction tasks, outperforming traditional tools like cTAKES and MetaMap and can handle extra attributes to be extracted. GENIE strongly enhances real-world applicability and scalability in healthcare systems. By open-sourcing the model and test data, we aim to encourage collaboration and drive further advancements in EHR structurization.

Utilising JWST, ALMA and the VLA we present high angular resolution (0.06''- 0.42''), multi-wavelength (4 micron - 3cm) observations of the VLA 1623-2417 protostellar system to characterise the origin, morphology and, properties of the continuum emission. JWST observations at 4.4 micron reveal outflow cavities for VLA 1623 A and, for the first time, VLA 1623 B, as well as scattered light from the upper layers of the VLA 1623 W disk. We model the millimetre-centimetre spectral energy distributions to quantify the relative contributions of dust and ionised gas emission, calculate dust masses, and use spectral index maps to determine where optical depth hinders this analysis. In general, all objects appear to be optically thick down to ~90 GHz, show evidence for significant amounts (10's - 100's M_Earth) of large (>1 mm) dust grains, and are dominated by ionised gas emission for frequencies ~<15 GHz. In addition, we find evidence of unsettled millimetre dust in the inclined disk of VLA 1623 B possibly attributed to instabilities within the circumstellar disk, adding to the growing catalogue of unsettled Class 0/I disks. Our results represent some of the highest resolution observations possible with current instrumentation, particularly in the case of the VLA. However, our interpretation is still limited at low frequencies (~<22 GHz) and thus motivates the need for next-generation interferometers operating at centimetre wavelengths., Comment: Accepted for publication in ApJ. 35 pages, 10 figures

Tailoring the degrees of freedom (DoF) of light for a desired purpose, so-called structured light, has delivered numerous advances over the past decade, ranging from communications and quantum cryptography to optical trapping, and microscopy. The shaping toolkit has traditionally been linear in nature, only recently extended to the nonlinear regime, where input beams overlap in a nonlinear crystal to generate a structured output beam. Here we show how to enhance the fidelity of the structured output by aligning light with light. Using orbital angular momentum modes and difference frequency generation as an example, we demonstrate precise control of the spatial overlap in both the transverse and longitudinal directions using the structure of one mode as a virtual structured (in amplitude and phase) light-based aperture for the other. Our technique can easily be translated to other structured light fields as well as alternative nonlinear processes such as second harmonic generation and sum frequency generation, enabling advancements in communication, imaging, and spectroscopy., Comment: 11 pages, 6 figures

Light's spatial degree of freedom is emerging as a potential resource for a myriad of applications, in both classical and quantum domains, including secure communication, sensing and imaging. However, it has been repeatedly shown that a complex medium (atmosphere, optical fibre, turbid media, etc.) can perturb the spatial amplitude, phase and polarization of the structured light fields leading to a degradation in their performance. A promising solution to this is the use of invariant modes to whom the medium appears transparent. While the creation and robustness of these modes has been experimentally demonstrated, they are difficult to implement in many important applications due to large channel matrices, a susceptibility to numerical artefacts, non-physical solutions and unreliable performance. In this work, we outline a procedure for determining these invariant modes using a modal basis, which results in a set of eigenmodes that are free of these issues, are consistently realisable and require a much smaller channel matrix to compute. Using atmospheric turbulence and LG modes as the underlying basis as an illustrative example, we find robust modes for a variety of turbulence strengths with a basis of only 231 modes, one order of magnitude smaller than previous approaches. These modes consistently show a fidelity of above 80% after propagating through the complex channel, a significant improvement over sending the individual LG modes themselves and reveal an invariant modal spectrum through the channel. Our approach will work for any complex medium and modal basis, paving the way for the effective implementation of the eigenmode approach in real-world situations., Comment: 21 pages, 8 figures. Submitted to Optics Express

The purpose of this article is to study a new problem of stochastic control, related to Walsh's spider diffusion, named: stochastic optimal scattering control. The optimal scattering control of the spider diffusion at the junction point is governed by an appropriate and highly non-trivial condition of the Kirchhoff Law type, involving an optimal diffraction probability measure selected from the own local time of the spider process at the vertex. In this work, we prove first the weak dynamic programming principle in the spirit of [32], adapted to the new class of spider diffusion introduced recently in [37]-[38]. Thereafter, we show that the value function of the problem is characterized uniquely in terms of a Hamilton Jacobi Bellman (HJB) system posed on a star-shaped network, having a new boundary condition at the vertex called : non linear local-time Kirchhoff's transmission. The key main point is to use the recent comparison theorem obtained in [40], that has significantly unlocked the study of this type of problem. We conclude by discussing the formulation of stochastic scattering control problems, where there is no dependency w.r.t. the local-time variable, for which their well-posedness appear as a simpler consequence of the results of this work and the advances contained in [40]., Comment: 53 pages

B-gallium oxide (B-Ga2O3) is of high interest for power electronics because of its unique combination of melt growth, epitaxial growth, n-type dopability, ultrawide bandgap, and high critical field. Optimization of crystal growth processes to promote beneficial defects and suppress harmful ones requires accurate quantitative modelling of both native and impurity defects. Here we quantitatively model defect concentrations as a function of bulk crystal growth conditions and demonstrate the necessity of including effects such as bandgap temperature dependence, chemical potentials from thermochemistry, and defect vibrational entropy in modelling based on defect formation energies computed by density functional theory (DFT) with hybrid functionals. Without these contributions, grossly-erroneous and misleading predictions arise, e.g. that n-type doping attempts would be fully compensated by Ga vacancies. Including these effects reproduces the experimental facts that melt-grown Sn-doped B-Ga2O3 crystals are conductive with small compensation while annealing the same crystals in O2 at intermediate temperatures renders them insulating. To accomplish this modeling, we developed a comprehensive modelling framework (KROGER) based on calculated defect formation energies and flexible thermodynamic conditions. These capabilities allow KROGER to capture full and partial defect equilibria amongst native defects and impurities occurring during specific semiconductor growth or fabrication processes. We use KROGER to model 873 charge-states of 259 defects involving 19 elements in conditions representing bulk crystal growth by edge-fed growth (EFG) and annealing in O2. Our methodology is transferrable to a wide range of materials beyond B-Ga2O3. Integration of thermodynamic and first-principles modelling of point defects provides insight into optimization of point defect populations in growth and processing., Comment: 39 pages, 6 figures, references

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve over 90\% accuracy on popular benchmarks like MMLU, limiting informed measurement of state-of-the-art LLM capabilities. In response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. HLE consists of 3,000 questions across dozens of subjects, including mathematics, humanities, and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable, but cannot be quickly answered via internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a significant gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai., Comment: 26 pages, 6 figures

As cyber threats grow increasingly sophisticated, reinforcement learning (RL) is emerging as a promising technique to create intelligent and adaptive cyber defense systems. However, most existing autonomous defensive agents have overlooked the inherent graph structure of computer networks subject to cyber attacks, potentially missing critical information and constraining their adaptability. To overcome these limitations, we developed a custom version of the Cyber Operations Research Gym (CybORG) environment, encoding network state as a directed graph with realistic low-level features. We employ a Graph Attention Network (GAT) architecture to process node, edge, and global features, and adapt its output to be compatible with policy gradient methods in RL. Our GAT-based approach offers key advantages over flattened alternatives: robust policies capable of handling dynamic network topology changes, generalisation to networks of varying sizes beyond the training distribution, and interpretable defensive actions grounded in tangible network properties. We demonstrate the effectiveness of our low-level directed graph observations by training GAT defensive policies that successfully adapt to changing network topologies. Evaluations across networks of different sizes, but consistent subnetwork structure, show our policies achieve comparable performance to policies trained specifically for each network configuration. Our study contributes to the development of robust cyber defence systems that can better adapt to real-world network security challenges.

We introduce the notion of a totally ($K$-) bounded element of a W*-probability space $(M, \varphi)$ and, borrowing ideas of Kadison, give an intrinsic characterization of the $^*$-subalgebra $M_{tb}$ of totally bounded elements. Namely, we show that $M_{tb}$ is the unique strongly dense $^*$-subalgebra $M_0$ of totally bounded elements of $M$ for which the collection of totally $1$-bounded elements of $M_0$ is complete with respect to the $\|\cdot\|_\varphi^\#$-norm and for which $M_0$ is closed under all operators $h_a(\log(\Delta))$ for $a \in \mathbb{N}$, where $\Delta$ is the modular operator and $h_a(t):=1/\cosh(t-a)$ (see Theorem 4.3). As an application, we combine this characterization with Rieffel and Van Daele's bounded approach to modular theory to arrive at a new language and axiomatization of W*-probability spaces as metric structures. Previous work of Dabrowski had axiomatized W*-probability spaces using a smeared version of multiplication, but the subalgebra $M_{tb}$ allows us to give an axiomatization in terms of the original algebra operations. Finally, we prove the (non-)axiomatizability of several classes of W*-probability spaces., Comment: 30 pages, comments welcome

Prior case retrieval (PCR) is crucial for legal practitioners to find relevant precedent cases given the facts of a query case. Existing approaches often overlook the underlying semantic intent in determining relevance with respect to the query case. In this work, we propose LeCoPCR, a novel approach that explicitly generate intents in the form of legal concepts from a given query case facts and then augments the query with these concepts to enhance models understanding of semantic intent that dictates relavance. To overcome the unavailability of annotated legal concepts, we employ a weak supervision approach to extract key legal concepts from the reasoning section using Determinantal Point Process (DPP) to balance quality and diversity. Experimental results on the ECtHR-PCR dataset demonstrate the effectiveness of leveraging legal concepts and DPP-based key concept extraction., Comment: Accepted to NAACL 2025

Consider the problem of deciding, for a particular multipartite quantum state, whether or not it is realizable in a quantum network with a particular causal structure. This is a fully quantum version of what causal inference researchers refer to as the problem of causal discovery. In this work, we introduce a fully quantum version of the inflation technique for causal inference, which leverages the quantum marginal problem. We illustrate the utility of this method using a simple example: testing compatibility of tripartite quantum states with the quantum network known as the triangle scenario. We show, in particular, how the method yields a complete classification of pure three-qubit states into those that are and those that are not compatible with the triangle scenario. We also provide some illustrative examples involving mixed states and some where one or more of the systems is higher-dimensional. Finally, we examine the question of when the incompatibility of a multipartite quantum state with a causal structure can be inferred from the incompatibility of a joint probability distribution induced by implementing measurements on each subsystem., Comment: 20+6 pages; 3 figures

The objective of this paper is to provide a baseline for performing multi-modal data classification on a novel open multimodal dataset of hepatocellular carcinoma (HCC), which includes both image data (contrast-enhanced CT and MRI images) and tabular data (the clinical laboratory test data as well as case report forms). TNM staging is the classification task. Features from the vectorized preprocessed tabular data and radiomics features from contrast-enhanced CT and MRI images are collected. Feature selection is performed based on mutual information. An XGBoost classifier predicts the TNM staging and it shows a prediction accuracy of $0.89 \pm 0.05$ and an AUC of $0.93 \pm 0.03$. The classifier shows that this high level of prediction accuracy can only be obtained by combining image and clinical laboratory data and therefore is a good example case where multi-model classification is mandatory to achieve accurate results.

A functional $k$-batch code of dimension $s$ consists of $n$ servers storing linear combinations of $s$ linearly independent information bits. These codes are designed to recover any multiset of $k$ requests, each being a linear combination of the information bits, by $k$ disjoint subsets of servers. A recent conjecture suggests that for any set of $k = 2^{s-1}$ requests, the optimal solution requires $2^s-1$ servers. This paper shows that the problem of functional $k$-batch codes is equivalent to several other problems. Using these equivalences, we derive sufficient conditions that improve understanding of the problem and enhance the ability to find the optimal solution.

Orchestrating next gen applications over hterogeneous resources along the Cloud-IoT continuum calls for new strategies and tools to enable scalable and application-specific managements. Inspired by the self-organisation capabilities of bacteria colonies, we propose a declarative, fully decentralised application management solution, targeting pervasive opportunistic Cloud-IoT infrastructures. We present acustomisable declarative implementation of the approach and validate its scalability through simulation over motivating scenarios, also considering end-user's mobility and the possibility to enforce application-specific management policies for different classes of applications.

This study compares three evolutionary algorithms for the problem of fog service placement: weighted sum genetic algorithm (WSGA), non-dominated sorting genetic algorithm II (NSGA-II), and multiobjective evolutionary algorithm based on decomposition (MOEA/D). A model for the problem domain (fog architecture and fog applications) and for the optimization (objective functions and solutions) is presented. Our main concerns are related to optimize the network latency, the service spread and the use of the resources. The algorithms are evaluated with a random Barabasi-Albert network topology with 100 devices and with two experiment sizes of 100 and 200 application services. The results showed that NSGA-II obtained the highest optimizations of the objectives and the highest diversity of the solution space. On the contrary, MOEA/D was better to reduce the execution times. The WSGA algorithm did not show any benefit with regard to the other two algorithms.

We investigate the perception of visual variables on wall-sized tiled displays within an immersive environment. We designed and conducted two formal user studies focusing on elementary visualization reading tasks in VR. The first study compared three different virtual display arrangements (Flat, Cylinder, and Cockpit). It showed that participants made smaller errors on virtual curved walls (Cylinder and Cockpit) compared to Flat. Following that, we compared the results with those from a previous study conducted in a real-world setting. The comparative analysis showed that virtual curved walls resulted in smaller errors than the real-world flat wall display, but with longer task completion time. The second study evaluated the impact of four 3D user interaction techniques (Selection, Walking, Steering, and Teleportation) on performing the elementary task on the virtual Flat wall display. The results confirmed that interaction techniques further improved task performance. Finally, we discuss the limitations and future work.

Within the rapidly evolving domain of Electronic Design Automation (EDA), Large Language Models (LLMs) have emerged as transformative technologies, offering unprecedented capabilities for optimizing and automating various aspects of electronic design. This survey provides a comprehensive exploration of LLM applications in EDA, focusing on advancements in model architectures, the implications of varying model sizes, and innovative customization techniques that enable tailored analytical insights. By examining the intersection of LLM capabilities and EDA requirements, the paper highlights the significant impact these models have on extracting nuanced understandings from complex datasets. Furthermore, it addresses the challenges and opportunities in integrating LLMs into EDA workflows, paving the way for future research and application in this dynamic field. Through this detailed analysis, the survey aims to offer valuable insights to professionals in the EDA industry, AI researchers, and anyone interested in the convergence of advanced AI technologies and electronic design., Comment: 21 pages, 2 figures, 3 tables, accepted by TODAES

Medium-timescale (minutes to hours) radio transients are a relatively unexplored population. The wide field-of-view and high instantaneous sensitivity of instruments such as MeerKAT provides an opportunity to probe this class of sources, using image-plane detection techniques. The previous letter in this series describes our project and associated TRON pipeline designed to mine archival MeerKAT data for transient and variable sources. In this letter, we report on a new transient, a radio flare, associated with Gaia DR3 6865945581361480448, a G type star, whose parallax places it at a distance of 1334 pc. Its duration and high degree of circular polarization suggests electron cyclotron maser instability as the mechanism, consistent with an RS CVn variable., Comment: 4 pages, 4 figures, submitted to MNRAS Letters

Medium-timescale (minutes to hours) radio transients are a relatively unexplored population. The wide field-of-view and high instantaneous sensitivity of instruments such as MeerKAT provides an opportunity to probe this class of sources, using image-plane detection techniques. We aim to systematically mine archival synthesis imaging data in order to search for medium-timescale transients and variables that are not detected by conventional long-track image synthesis techniques. We deploy a prototype blind transient and variable search pipeline named TRON. This processes calibrated visibility data, constructs high-time cadence images, performs a search for variability on multiple timescales, and extracts lightcurves for detected sources. As proof of concept, we apply it to three MeerKAT observations of globular clusters, known to host transient or variable sources. We detect a previously known eclipsing MSP suspected to be a `black widow' system, in the globular cluster Omega Centauri, with a light curve confirming the eclipsing nature of the emission. We detect a previously known `red back' eclipsing MSP in the globular cluster Terzan 5. Using observations of the globular cluster 47 Tucanae, we detect two known millisecond pulsars (MSPs), and one previously reported MSP candidate, with hints of eclipsing behaviour., Comment: 6 pages, 8 figures, published in MNRAS Letters

We initiate the study of the effective content of $K$-theory for $\mathrm{C}^*$-algebras. We prove that there are computable functors which associate, to a computably enumerable presentation of a $\mathrm{C}^*$-algebra $\boldA$, computably enumerable presentations of the abelian groups $K_0(\boldA)$ and $K_1(\boldA)$. When $\boldA$ is stably finite, we show that the positive cone of $K_0(\boldA)$ is computably enumerable. We strengthen the results in the case that $\boldA$ is a UHF algebra by showing that the aforementioned presentation of $K_0(\boldA)$ is actually computable. In the UHF case, we also show that $\boldA$ has a computable presentation precisely when $K_0(\boldA)$ has a computable presentation, which in turn is equivalent to the supernatural number of $\boldA$ being lower semicomputable; we give an example that shows that this latter equivalence cannot be improved to requiring that the supernatural number of $\boldA$ is computable. Finally, we prove that every UHF algebra is computably categorical.

Remote sensing imagery is dense with objects and contextual visual information. There is a recent trend to combine paired satellite images and text captions for pretraining performant encoders for downstream tasks. However, while contrastive image-text methods like CLIP enable vision-language alignment and zero-shot classification ability, vision-only downstream performance tends to degrade compared to image-only pretraining, such as MAE. In this paper, we propose FLAVARS, a pretraining method that combines the best of both contrastive learning and masked modeling, along with geospatial alignment via contrastive location encoding. We find that FLAVARS significantly outperforms a baseline of SkyCLIP for vision-only tasks such as KNN classification and semantic segmentation, +6\% mIOU on SpaceNet1, while retaining the ability to perform zero-shot classification, unlike MAE pretrained methods.

Neutrino self-interactions beyond the standard model have profound implications in astrophysics and cosmology. In this work, we study an uncharted scenario in which one of the three neutrino species has a mass much smaller than the temperature of the cosmic neutrino background. This results in a relativistic component that significantly broadens the absorption feature on the astrophysical neutrino spectra, in contrast to the sharply peaked absorption expected in the extensively studied scenarios assuming a fully nonrelativistic cosmic neutrino background. By solving the Boltzmann equations for neutrino absorption and regeneration, we demonstrate that this mechanism provides novel sensitivity to sub-keV mediator masses, well below the traditional $\sim 1$--100 MeV range. Future observations of the diffuse supernova neutrino background with Hyper-Kamiokande could probe coupling strengths down to $g \sim 10^{-8}$, surpassing existing constraints by orders of magnitude. These findings open new directions for discoveries and offer crucial insights into the interplay between neutrinos and the dark sector., Comment: 11 pages, 4 figures

Quantum computing and quantum sensing represent two distinct frontiers of quantum information science. In this work, we harness quantum computing to solve a fundamental and practically important sensing problem: the detection of weak oscillating fields with unknown strength and frequency. We present a quantum computing enhanced sensing protocol that outperforms all existing approaches. Furthermore, we prove our approach is optimal by establishing the Grover-Heisenberg limit -- a fundamental lower bound on the minimum sensing time. The key idea is to robustly digitize the continuous, analog signal into a discrete operation, which is then integrated into a quantum algorithm. Our metrological gain originates from quantum computation, distinguishing our protocol from conventional sensing approaches. Indeed, we prove that broad classes of protocols based on quantum Fisher information, finite-lifetime quantum memory, or classical signal processing are strictly less powerful. Our protocol is compatible with multiple experimental platforms. We propose and analyze a proof-of-principle experiment using nitrogen-vacancy centers, where meaningful improvements are achievable using current technology. This work establishes quantum computation as a powerful new resource for advancing sensing capabilities., Comment: 9 pages, 4 figures + 44 pages, 3 figures

Plasmonic nanostructures have wide applications in photonics including pathogen detection and diagnosis via Surface-Enhanced Raman Spectroscopy (SERS). Despite major role plasmonics play in signal enhancement, electrostatics in SERS is yet to be fully understood and harnessed. Here, we perform a systematic study of electrostatic interactions between 785 nm resonant gold nanorods designed to harbor zeta potentials of +29, +16, 0 and -9 mV spanning positive neutral and negative domains. SERS activity is tested on representative Gram-negative Escherichia coli and Gram-positive Staphylococcus epidermidis bacteria with zeta potentials of -30 and -23 mV respectively in water. Raman spectroscopy and Cryo-Electron microscopy reveal that +29, +16, 0 and -9 mV nanorods give SERS enhancement of 7.2X, 3.6X, 4.2X, 1.3X to Staphylococcus epidermidis and 3.9X, 2.8X, 2.9X, 1.1X to Escherichia coli. Theoretical results show that electrostatics play the major role among all interaction forces in determining cell-nanorod proximity and signal enhancement. We identify steric repulsion due to cell protrusions to be the critical opposing force. Finally, a design principle is proposed to estimate the electrostatic strength in SERS. Our work provides new insights into the principle of bacteria-nanorod interactions, enabling reproducible and precise biomolecular readouts, critical for next-generation point-of-care diagnostics and smart healthcare applications.

Colloidal perovskite quantum dots (PQDs) are an exciting platform for on-demand quantum, and classical optoelectronic and photonic devices. However, their potential success is limited by the extreme sensitivity and low stability arising from their weak intrinsic lattice bond energy and complex surface chemistry. Here we report a novel platform of buried perovskite quantum dots (b-PQDs) in a three-dimensional perovskite thin-film, fabricated using one-step, flash annealing, which overcomes surface related instabilities in colloidal perovskite dots. The b-PQDs demonstrate ultrabright and stable single-dot emission, with resolution-limited linewidths below 130 {\mu}eV, photon-antibunching (g^2(0)=0.1), no blinking, suppressed spectral diffusion, and high photon count rates of 10^4/s, consistent with unity quantum yield. The ultrasharp linewidth resolves exciton fine-structures (dark and triplet excitons) and their dynamics under a magnetic field. Additionally, b-PQDs can be electrically driven to emit single photons with 1 meV linewidth and photon-antibunching (g^2(0)=0.4). These results pave the way for on-chip, low-cost single-photon sources for next generation quantum optical communication and sensing., Comment: 26 pages, 4 figures

We introduce the \textit{crossing profile} of a drawing of a graph. This is a sequence of integers whose $(k+1)^{\text{th}}$ entry counts the number of edges in the drawing which are involved in exactly $k$ crossings. The first and second entries of this sequence (which count uncrossed edges and edges with one crossing, respectively) have been studied by multiple authors. However, to the best of our knowledge, we are the first to consider the entire sequence. Most of our results concern crossing profiles of rectilinear drawings of the complete graph $K_n$. We show that for any $k\leq (n-2)^2/4$ there is such a drawing for which the $k^{\text{th}}$ entry of the crossing profile is of magnitude $\Omega(n)$. On the other hand, we prove that for any $k \geq 1$ and any sufficiently large $n$, the $k^{\text{th}}$ entry can also be made to be $0$. As our main result, we essentially characterize the asymptotic behavior of both the maximum and minimum values that the sum of the first $k$ entries of the crossing profile might achieve. Our proofs are elementary and rely mostly on geometric constructions and classical results from discrete geometry and geometric graph theory., Comment: Paper: 19 pages. Appendix: 6 pages. Many figures

Quick Merkle Database (QMDB) addresses longstanding bottlenecks in blockchain state management by integrating key-value (KV) and Merkle tree storage into a single unified architecture. QMDB delivers a significant throughput improvement over existing architectures, achieving up to 6X over the widely used RocksDB and 8X over NOMT, a leading verifiable database. Its novel append-only twig-based design enables one SSD read per state access, O(1) IOs for updates, and in-memory Merkleization on a memory footprint as small as 2.3 bytes per entry, enabling it to run on even modest consumer-grade PCs. QMDB scales seamlessly across both commodity and enterprise hardware, achieving up to 2.28 million state updates per second. This performance enables support for 1 million token transfers per second (TPS), marking QMDB as the first solution achieving such a milestone. QMDB has been benchmarked with workloads exceeding 15 billion entries (10X Ethereum's 2024 state) and has proven the capacity to scale to 280 billion entries on a single server. Furthermore, QMDB introduces historical proofs, unlocking the ability to query its blockchain's historical state at the latest block. QMDB not only meets the demands of current blockchains but also provides a robust foundation for building scalable, efficient, and verifiable decentralized applications across diverse use cases., Comment: 11 pages, 3 figures

We establish an infinite hierarchy of finite-time gradient catastrophes for smooth solutions of the 1D Euler equations of gas dynamics with non-constant entropy. Specifically, for all integers $n\geq 1$, we prove that there exist classical solutions, emanating from smooth, compressive, and non-vacuous initial data, which form a cusp-type gradient singularity in finite time, in which the gradient of the solution has precisely $C^{0,\frac{1}{2n+1}}$ H\"older-regularity. We show that such Euler solutions are codimension-$(2n-2)$ stable in the Sobolev space $W^{2n+2,\infty}$.

We investigate the single-site von Neumann entropy along a harmonically confined superfluid chain, as described by the one-dimensional fermionic Hubbard model with strongly attractive interactions. We find that by increasing the confinement (or equivalently the particle filling) the system undergoes a quantum phase transition from a superfluid (SF) to a non-trivial topological insulator (TI) phase, which is characterized by an insulating bulk surrounded by highly entangled superfluid edges. These highly entangled states are found to be robust against perturbations and topologically protected by the particle-hole symmetry, which is locally preserved. We also find a semi-quantitative agreement between entanglement and superconducting order parameter profiles, confirming then that entanglement can be used as a topological marker and an order parameter in these systems. The charge gap not only confirms the SF-TI transition, but also shows that this transition is mediated by a metallic intermediate regime within the bulk. Using the gap we could depict a phase diagram for which the non-trivial topological insulator can be found in such harmonically confined attractive systems.

What is appropriateness? Humans navigate a multi-scale mosaic of interlocking notions of what is appropriate for different situations. We act one way with our friends, another with our family, and yet another in the office. Likewise for AI, appropriate behavior for a comedy-writing assistant is not the same as appropriate behavior for a customer-service representative. What determines which actions are appropriate in which contexts? And what causes these standards to change over time? Since all judgments of AI appropriateness are ultimately made by humans, we need to understand how appropriateness guides human decision making in order to properly evaluate AI decision making and improve it. This paper presents a theory of appropriateness: how it functions in human society, how it may be implemented in the brain, and what it means for responsible deployment of generative AI technology., Comment: 115 pages, 2 figures

Recent advancements in Artificial Intelligence (AI) have transformed decision-making in aeronautics and aerospace. These advancements in AI have brought with them the need to understand the reasons behind the predictions generated by AI systems and models, particularly by professionals in these sectors. In this context, the emergence of eXplainable Artificial Intelligence (XAI) has helped bridge the gap between professionals in the aeronautical and aerospace sectors and the AI systems and models they work with. For this reason, this paper provides a review of the concept of XAI is carried out defining the term and the objectives it aims to achieve. Additionally, the paper discusses the types of models defined within it and the properties these models must fulfill to be considered transparent, as well as the post-hoc techniques used to understand AI systems and models after their training. Finally, various application areas within the aeronautical and aerospace sectors will be presented, highlighting how XAI is used in these fields to help professionals understand the functioning of AI systems and models.

Optical wireless communication (OWC) is a promising technology anticipated to play a key role in the next-generation network of networks. To this end, this paper details the potential of OWC, as a complementary technology to traditional radio frequency communications, in enhancing networking capabilities beyond conventional terrestrial networks. Several usage scenarios and the current state of development are presented. Furthermore, a summary of existing challenges and opportunities are provided. Emerging technologies aimed at further enhancing future OWC capabilities are introduced. Additionally, value-added OWC-based technologies that leverage the unique properties of light are discussed, including applications such as positioning and gesture recognition. The paper concludes with the reflection that OWC provides unique functionalities that can play a crucial role in building convergent and resilient future network of networks., Comment: This work has been submitted to the IEEE for possible publication. 15pp, 16 figures, and one table

In this paper, we consider the inverse scattering problem associated with an anisotropic medium with a conductive boundary condition. We will assume that the corresponding far--field pattern or Cauchy data is either known or measured. The conductive boundary condition models a thin coating around the boundary of the scatterer. We will develop two direct sampling methods to solve the inverse shape problem by numerically recovering the scatterer. To this end, we study direct sampling methods by deriving that the corresponding imaging functionals decay as the sampling point moves away from the scatterer. These methods have been applied to other inverse shape problems, but this is the first time they will be applied to an anisotropic scatterer with a conductive boundary condition. These methods allow one to recover the scatterer by considering an inner--product of the far--field data or the Cauchy data. Here, we will assume that the Cauchy data is known on the boundary of a region $\Omega$ that completely encloses the scatterer $D$. We present numerical reconstructions in two dimensions to validate our theoretical results for both circular and non-circular scatterers.