Search

The rise of generative image models leads to privacy concerns when it comes to the huge datasets used to train such models. This paper investigates the possibility of inferring if a set of face images was used for fine-tuning a Latent Diffusion Model (LDM). A Membership Inference Attack (MIA) method is presented for this task. Using generated auxiliary data for the training of the attack model leads to significantly better performance, and so does the use of watermarks. The guidance scale used for inference was found to have a significant influence. If a LDM is fine-tuned for long enough, the text prompt used for inference has no significant influence. The proposed MIA is found to be viable in a realistic black-box setup against LDMs fine-tuned on face-images., Comment: In Proceedings of the 20th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2025) - Volume 2: VISAPP, pages 439-446

Epithelial cells form diverse structures from squamous spherical organoids to densely packed pseudostratified tissues. Quantification of cellular properties in these contexts requires high-resolution deep imaging and computational techniques to achieve truthful three-dimensional (3D) structural features. Here, we describe a detailed step-by-step protocol for sample preparation, imaging and deep-learning-assisted cell segmentation to achieve accurate quantification of fluorescently labelled individual cells in 3D within live tissues. We share the lessons learned through troubleshooting 3D imaging of Drosophila wing discs, including considerations on the choice of microscopy modality and settings (objective, sample mounting) and available segmentation methods. In addition, we include a computational pipeline alongside custom code to assist replication of the protocol. While we focus on the segmentation of cell outlines from membrane labelling, this protocol applies to a wide variety of samples, and we believe it be valuable for studying other tissues that demand complex analysis in 3D.

Ontology matching (OM) plays a key role in enabling data interoperability and knowledge sharing, but it remains challenging due to the need for large training datasets and limited vocabulary processing in machine learning approaches. Recently, methods based on Large Language Model (LLMs) have shown great promise in OM, particularly through the use of a retrieve-then-prompt pipeline. In this approach, relevant target entities are first retrieved and then used to prompt the LLM to predict the final matches. Despite their potential, these systems still present limited performance and high computational overhead. To address these issues, we introduce MILA, a novel approach that embeds a retrieve-identify-prompt pipeline within a prioritized depth-first search (PDFS) strategy. This approach efficiently identifies a large number of semantic correspondences with high accuracy, limiting LLM requests to only the most borderline cases. We evaluated MILA using the biomedical challenge proposed in the 2023 and 2024 editions of the Ontology Alignment Evaluation Initiative. Our method achieved the highest F-Measure in four of the five unsupervised tasks, outperforming state-of-the-art OM systems by up to 17%. It also performed better than or comparable to the leading supervised OM systems. MILA further exhibited task-agnostic performance, remaining stable across all tasks and settings, while significantly reducing LLM requests. These findings highlight that high-performance LLM-based OM can be achieved through a combination of programmed (PDFS), learned (embedding vectors), and prompting-based heuristics, without the need of domain-specific heuristics or fine-tuning.

Point defects dictate the properties of many functional materials. The standard approach to modelling the thermodynamics of defects relies on a static description, where the change in Gibbs free energy is approximated by the internal energy. This approach has a low computational cost, but ignores contributions from atomic vibrations and structural configurations that can be accessed at finite temperatures. We train a machine learning force field (MLFF) to explore dynamic defect behaviour using $\mathrm{Te_i^{+1}}$ and $\textit{V}{\mathrm{_{Te}^{+2}}}$ in CdTe as exemplars. We consider the different entropic contributions (e.g., electronic, spin, vibrational, orientational, and configurational) and compare methods to compute the defect free energies, ranging from a harmonic treatment to a fully anharmonic approach based on thermodynamic integration. We find that metastable configurations are populated at room temperature and thermal effects increase the predicted concentration of $\mathrm{Te_i^{+1}}$ by two orders of magnitude -- and can thus significantly affect the predicted properties. Overall, our study underscores the importance of finite-temperature effects and the potential of MLFFs to model defect dynamics at both synthesis and device operating temperatures.

Cognitive and neurological impairments are very common, but only a small proportion of affected individuals are diagnosed and treated, partly because of the high costs associated with frequent screening. Detecting pre-illness stages and analyzing the progression of neurological disorders through effective and efficient intelligent systems can be beneficial for timely diagnosis and early intervention. We propose using Large Language Models to extract features from free dialogues to detect cognitive decline. These features comprise high-level reasoning content-independent features (such as comprehension, decreased awareness, increased distraction, and memory problems). Our solution comprises (i) preprocessing, (ii) feature engineering via Natural Language Processing techniques and prompt engineering, (iii) feature analysis and selection to optimize performance, and (iv) classification, supported by automatic explainability. We also explore how to improve Chatgpt's direct cognitive impairment prediction capabilities using the best features in our models. Evaluation metrics obtained endorse the effectiveness of a mixed approach combining feature extraction with Chatgpt and a specialized Machine Learning model to detect cognitive decline within free-form conversational dialogues with older adults. Ultimately, our work may facilitate the development of an inexpensive, non-invasive, and rapid means of detecting and explaining cognitive decline.

We lift the Lefschetz number from an algebraic invariant of maps between spaces to an invariant of morphisms of data over the spaces., Comment: 21 pages. All comments are welcome!

Lead-halide perovskites (LHPs) have shot to prominence as efficient energy conversion materials that can be processed using cost-effective fabrication methods. A widely-quoted reason for their exceptional performance is their ability to tolerate defects, enabling long charge-carrier lifetimes despite high defect densities. Realizing defect tolerance in broader classes of materials would have a substantial impact on the semiconductor industry. Significant effort has been made over the past decade to unravel the underlying origins of defect tolerance to design stable alternatives to LHPs comprised of nontoxic elements. However, it has become clear that understanding defect tolerance in LHPs is far from straightforward. This review discusses the models proposed for defect tolerance in halide perovskites, evaluating the experimental and theoretical support for these models, as well as their limitations. We cover attempts to apply these models to identify materials beyond the lead-halide system that could also exhibit defect tolerance, and the successes and pitfalls encountered over the past decade. Finally, a discussion is made of some of the important missing pieces of information required for a deeper understanding and predictive models that enable the inverse design of defect tolerant semiconductors., Comment: 64 pages, 4 figures, 3 boxes, 1 table

We theoretically investigate the magnetic response on a three-dimensional superconducting nanobridge system, which is compound of two parallel parallelepiped (samples) connected through a nanobridge of size $\mathbf{L}$ and thickness $\mathbf{x}$, which mediates interactions between them. This study is conducted in the presence of a magnetic field $\mathbf{H}$ and the transport of a direct current $\mathbf{J}$. We use the well-know time dependent Ginzburg-Landau theory ($\mathbf{TDGL}$) for analyzed the possible effects on the density Gibbs free energy $\mathbf{F}$, magnetization $\mathbf{M}$, and superconducting electronic Cooper pair density $|\psi|^{2}$. We are interested in studying two cases: varying the $\mathbf{L}$ and $\mathbf{x}$ of the nanobridge in the absence of induced $\mathbf{J}$, and including the induction of external $\mathbf{J}$ for fixed $\mathbf{L}$ and $\mathbf{x}$. We find that $\mathbf{L}$ and $\mathbf{x}$ play an essential role in stabilizing (controlling) vortex states in the nanobridge, and the presence of induced $\mathbf{J}$ ($\mathbf{J}>0$ and $\mathbf{J}<0$), with a fixed $\mathbf{L}$ and $\mathbf{x}$, causes the movement of vortex states in the nanobridge just when $\mathbf{J}$ is induced at both faces of superconducting nanobridge system., Comment: 11 pages, 10 figures

We consider experimentally and computationally the phenomenon of the reverse Janssen effect, involving the counterintuitive finding that the force on the base of a column containing granular particles may be larger than the weight of the granular material itself. This finding is in contrast to the common Janssen effect, for which the force on the base is smaller than the particle weight, illustrating one of the best-known differences between granular and liquid systems. We find that the reverse Janssen effect is strongly influenced by the pouring protocol: under Earth's gravitational field, we find that the reverse Janssen effect is strongly and consistently influenced by the pouring height, as well as by (to somewhat lesser degree) pouring flux. Pouring grains from the height measured in tens of particle diameters leads to a reverse Janssen effect that is an order of magnitude stronger than the one found for small pouring heights of few particle diameters. This influence of the particles' delivery protocol allows for the development of a better understanding of the general features of the reverse Janssen effect and of the comparison between experiments and simulations reported in this and previous works.

We present a numerical experiment that demonstrates the possibility to capture topological phase transitions via an x-ray absorption spectroscopy scheme. We consider a Chern insulator whose topological phase is tuned via a second-order hopping. We perform time-dynamics simulations of the out-of-equilibrium laser-driven electron motion that enables us to model a realistic attosecond spectroscopy scheme. In particular, we use an ultrafast scheme with a circularly polarized IR pump pulse and an attosecond x-ray probe pulse. A laser-induced dichroism-type spectrum shows a clear signature of the topological phase transition. We are able to connect these signatures with the Berry structure of the system. This work extend the applications of attosecond absorption spectroscopy to systems presenting a non-trivial topological phase.

New technologies such as Machine Learning (ML) gave great potential for evaluating industry workflows and automatically generating key performance indicators (KPIs). However, despite established standards for measuring the efficiency of industrial machinery, there is no precise equivalent for workers' productivity, which would be highly desirable given the lack of a skilled workforce for the next generation of industry workflows. Therefore, an ML solution combining data from manufacturing processes and workers' performance for that goal is required. Additionally, in recent times intense effort has been devoted to explainable ML approaches that can automatically explain their decisions to a human operator, thus increasing their trustworthiness. We propose to apply explainable ML solutions to differentiate between expert and inexpert workers in industrial workflows, which we validate at a quality assessment industrial workstation. Regarding the methodology used, input data are captured by a manufacturing machine and stored in a NoSQL database. Data are processed to engineer features used in automatic classification and to compute workers' KPIs to predict their level of expertise (with all classification metrics exceeding 90 %). These KPIs, and the relevant features in the decisions are textually explained by natural language expansion on an explainability dashboard. These automatic explanations made it possible to infer knowledge from expert workers for inexpert workers. The latter illustrates the interest of research in self-explainable ML for automatically generating insights to improve productivity in industrial workflows., Comment: IEEE Industrial Electronics Magazine (2023)

Human feedback is central to the alignment of Large Language Models (LLMs). However, open questions remain about methods (how), domains (where), people (who) and objectives (to what end) of feedback processes. To navigate these questions, we introduce PRISM, a dataset that maps the sociodemographics and stated preferences of 1,500 diverse participants from 75 countries, to their contextual preferences and fine-grained feedback in 8,011 live conversations with 21 LLMs. With PRISM, we contribute (i) wider geographic and demographic participation in feedback; (ii) census-representative samples for two countries (UK, US); and (iii) individualised ratings that link to detailed participant profiles, permitting personalisation and attribution of sample artefacts. We target subjective and multicultural perspectives on value-laden and controversial issues, where we expect interpersonal and cross-cultural disagreement. We use PRISM in three case studies to demonstrate the need for careful consideration of which humans provide what alignment data.

We introduce two novel complementary notions of the Lefschetz number for a functor from a finite acyclic category to itself and we prove a Lefschetz fixed-object theorem and a Lefschetz fixed-morphism theorem. In order to do so, we use the connection between these type of categories and simplicial structures, such as trisps or delta complexes. Through the use of a pair of functors that, when composed, form the barycentric subdivision, we are not only able to identify fixed objects but also fixed chains of morphisms., Comment: 8 pages with 3 figures

As the field of AI continues to evolve, a significant dimension of this progression is the development of Large Language Models and their potential to enhance multi-agent artificial intelligence systems. This paper explores the cooperative capabilities of Large Language Model-augmented Autonomous Agents (LAAs) using the well-known Meltin Pot environments along with reference models such as GPT4 and GPT3.5. Preliminary results suggest that while these agents demonstrate a propensity for cooperation, they still struggle with effective collaboration in given environments, emphasizing the need for more robust architectures. The study's contributions include an abstraction layer to adapt Melting Pot game scenarios for LLMs, the implementation of a reusable architecture for LLM-mediated agent development - which includes short and long-term memories and different cognitive modules, and the evaluation of cooperation capabilities using a set of metrics tied to the Melting Pot's "Commons Harvest" game. The paper closes, by discussing the limitations of the current architectural framework and the potential of a new set of modules that fosters better cooperation among LAAs.

Defects are a universal feature of crystalline solids, dictating the key properties and performance of many functional materials. Given their crucial importance yet inherent difficulty in measuring experimentally, computational methods (such as DFT and ML/classical force-fields) are widely used to predict defect behaviour at the atomic level and the resultant impact on macroscopic properties. Here we report doped, a Python package for the generation, pre-/post-processing, and analysis of defect supercell calculations. doped has been built to implement the defect simulation workflow in an efficient and user-friendly -- yet powerful and fully-flexible -- manner, with the goal of providing a robust general-purpose platform for conducting reproducible calculations of solid-state defect properties.

Histopathology is a gold standard for cancer diagnosis under a microscopic examination. However, histological tissue processing procedures result in artifacts, which are ultimately transferred to the digitized version of glass slides, known as whole slide images (WSIs). Artifacts are diagnostically irrelevant areas and may result in wrong deep learning (DL) algorithms predictions. Therefore, detecting and excluding artifacts in the computational pathology (CPATH) system is essential for reliable automated diagnosis. In this paper, we propose a mixture of experts (MoE) scheme for detecting five notable artifacts, including damaged tissue, blur, folded tissue, air bubbles, and histologically irrelevant blood from WSIs. First, we train independent binary DL models as experts to capture particular artifact morphology. Then, we ensemble their predictions using a fusion mechanism. We apply probabilistic thresholding over the final probability distribution to improve the sensitivity of the MoE. We developed DL pipelines using two MoEs and two multiclass models of state-of-the-art deep convolutional neural networks (DCNNs) and vision transformers (ViTs). DCNNs-based MoE and ViTs-based MoE schemes outperformed simpler multiclass models and were tested on datasets from different hospitals and cancer types, where MoE using DCNNs yielded the best results. The proposed MoE yields 86.15% F1 and 97.93% sensitivity scores on unseen data, retaining less computational cost for inference than MoE using ViTs. This best performance of MoEs comes with relatively higher computational trade-offs than multiclass models. The proposed artifact detection pipeline will not only ensure reliable CPATH predictions but may also provide quality control., Comment: Submitted to BMC Medical Informatics and Decision Making Journal

With the rise of text-to-image (T2I) generative AI models reaching wide audiences, it is critical to evaluate model robustness against non-obvious attacks to mitigate the generation of offensive images. By focusing on ``implicitly adversarial'' prompts (those that trigger T2I models to generate unsafe images for non-obvious reasons), we isolate a set of difficult safety issues that human creativity is well-suited to uncover. To this end, we built the Adversarial Nibbler Challenge, a red-teaming methodology for crowdsourcing a diverse set of implicitly adversarial prompts. We have assembled a suite of state-of-the-art T2I models, employed a simple user interface to identify and annotate harms, and engaged diverse populations to capture long-tail safety issues that may be overlooked in standard testing. The challenge is run in consecutive rounds to enable a sustained discovery and analysis of safety pitfalls in T2I models. In this paper, we present an in-depth account of our methodology, a systematic study of novel attack strategies and discussion of safety failures revealed by challenge participants. We also release a companion visualization tool for easy exploration and derivation of insights from the dataset. The first challenge round resulted in over 10k prompt-image pairs with machine annotations for safety. A subset of 1.5k samples contains rich human annotations of harm types and attack styles. We find that 14% of images that humans consider harmful are mislabeled as ``safe'' by machines. We have identified new attack strategies that highlight the complexity of ensuring T2I model robustness. Our findings emphasize the necessity of continual auditing and adaptation as new vulnerabilities emerge. We are confident that this work will enable proactive, iterative safety assessments and promote responsible development of T2I models., Comment: 10 pages, 6 figures

In this paper we introduce and analyse, from a game theoretical perspective, several multi-agent or multi-item continuous review inventory models in which the buyers are exempted from ordering costs if the price of their orders is greater than or equal to a certain amount. For all models we obtain the optimal ordering policy. We first analyse a simple model with one firm and one item. Then, we study a model with one firm and several items, for which we design a procedure based on cooperative game theory to evaluate the impact of each item on the total cost. Then, we deal with a model with several firms and one item for each firm, for which we characterise a rule to allocate the total cost among the firms in a coalitionally stable way. Finally, we discuss a model with several firms and several items, for which we characterise a rule to allocate the total cost among the firms in a coalitionally stable way and to evaluate the impact of each item on the cost that would be payable to each firm when using the allocation rule. All the concepts and results of this article are illustrated using data from a case study.

Based on Holler (1982) and Armijos-Toro et al. (2021) we propose two power indices to measure the influence of the players in the class of weighted majority games in partition function form. We compare these new power indices with their original versions on the class of games in characteristic function form. Finally, we use both pairs of power indices for games in partition function form to study the distribution of power in the National Assembly of Ecuador that emerged after the elections of February 7, 2021.

In this paper, we introduce a notion of mergeable weighted majority games with the aim of providing the first characterization of the Colomer-Mart\'inez power index (Colomer and Mart\'inez in J Theor Polit 7(1):41-63, 1995). Furthermore, we define and characterize a new power index for the family of weighted majority games that combines ideas of the Public Good (Holler in Polit Stud 30(2):262-271, 1982) and Colomer-Mart\'inez power indices. Finally, we analyze the National Assembly of Ecuador using these and some other well-known power indices., Comment: This article was published on April 15, 2023 in Open Access in the journal Annals of Operations Research

In this paper we deal with production situations where a cap or limit to the amount of greenhouse gas emissions permitted is imposed. Fixing a tax for each ton of pollutant emitted is also considered. We use bankruptcy rules to define cooperative games with externalities associated with these situations and analyze the existence of coalitionally stable allocations of the emission permits. We prove that the constrained equal awards ( CEA ) rule provides stable allocations and as a direct mechanism, it is incentive compatible. These two facts have interesting managerial implications to control pollution emissions.

We introduce a new model involving TU-games and exogenous structures. Specifically, we consider that each player in a population can choose an element in a strategy set and that, for every possible strategy profile, a TU-game is associated with the population. This is what we call a TU-game with strategies. We propose and characterize the maxmin procedure to map every game with strategies to a TU-game. We also study whether or not the relevant properties of TU-games are transmitted by applying the maxmin procedure. Finally, we examine two relevant classes of TU-games with strategies: airport and simple games with strategies.

In order to minimize the generalization error in neural networks, a novel technique to identify overfitting phenomena when training the learner is formally introduced. This enables support of a reliable and trustworthy early stopping condition, thus improving the predictive power of that type of modeling. Our proposal exploits the correlation over time in a collection of online indicators, namely characteristic functions for indicating if a set of hypotheses are met, associated with a range of independent stopping conditions built from a canary judgment to evaluate the presence of overfitting. That way, we provide a formal basis for decision making in terms of interrupting the learning process. As opposed to previous approaches focused on a single criterion, we take advantage of subsidiarities between independent assessments, thus seeking both a wider operating range and greater diagnostic reliability. With a view to illustrating the effectiveness of the halting condition described, we choose to work in the sphere of natural language processing, an operational continuum increasingly based on machine learning. As a case study, we focus on parser generation, one of the most demanding and complex tasks in the domain. The selection of cross-validation as a canary function enables an actual comparison with the most representative early stopping conditions based on overfitting identification, pointing to a promising start toward an optimal bias and variance control., Comment: 26 pages, 6 figures

We report the serendipitous discovery of a large edge-on protoplanetary disk associated with the infrared source IRAS 23077+6707. The disk's apparent size in the Pan-STARRS (PS1) images is ~11", making this one of the largest known disks on the sky. It is likely a young system, still surrounded by the envelope which is very faint but still visible in the PS1 images in the northern part (alternatively this structure could be filaments from the disk itself). We use the PS1 magnitudes and other available photometric data to construct the spectral energy distribution (SED) of the disk. An optical spectrum indicates that the obscured star is hot, most likely late A. We adopt a distance of 300 pc for this object based on Gaia DR3 extinctions. We model the system using the HOCHUNK3D radiative transfer software and find that the system is consistent with a hot star of effective temperature 8000 K surrounded by a disk of size 1650 AU and mass 0.2 M_solar at inclination 82 degrees., Comment: 11 pages, 4 figures, accepted in ApJL

Impregnation treatments are one of the alternatives to protect concrete-based building and monuments from weathering degradation. However, it is important to consider the chemical compatibility of the reaction products with the building material. The impregnation product studied here consists of a silica oligomer able to polymerize, by a simple sol-gel process, inside the pore structure of concrete. In this work, we investigate the ability of this impregnation treatment to produce C-S-H gel in contact with cement paste. A complete characterization of the reaction products demonstrated that the silanol groups from silica oligomers react with the portlandite present in the cement paste generating a material with the chemical, structural and morphological features of CS-H gel. Simultaneously, the 29Si NMR results indicate that the Si-O units are incorporated into the existing C-S-H, increasing its chain length. These results open the way for a simple concrete structures repairing procedure., Comment: 15 pages, 11 figures, 1 table, 44 references. Gold Open Access publication

Defects dictate the properties of many functional materials. To understand the behaviour of defects and their impact on physical properties, it is necessary to identify the most stable defect geometries. However, global structure searching is computationally challenging for high-throughput defect studies or materials with complex defect landscapes, like alloys or disordered solids. Here, we tackle this limitation by harnessing a machine-learning surrogate model to qualitatively explore the defect structural landscape. By learning defect motifs in a family of related metal chalcogenide and mixed anion crystals, the model successfully predicts favourable reconstructions for unseen defects in unseen compositions for 90% of cases, thereby reducing the number of first-principles calculations by 73%. Using CdSe$_x$Te$_{1-x}$ alloys as an exemplar, we train a model on the end member compositions and apply it to find the stable geometries of all inequivalent vacancies for a range of mixing concentrations, thus enabling more accurate and faster defect studies for configurational complex systems.

We develop a theory of integration with respect to the Lefschetz number in the context of o-minimal structures containing the semilinear sets. We prove several results and we apply the theory to the field of object detection using sensors.