Search

Even though a few initial works have shown on small sets of data some level of bias in the performance of fingerprint recognition technology with respect to certain demographic groups, there is still not sufficient evidence to understand the impact that certain factors such as gender, age or finger-type may have on fingerprint quality and, in turn, also on fingerprint matching accuracy. The present work addresses this still under researched topic, on a large-scale database of operational data containing 10-print impressions of almost 16,000 subjects. The results reached provide further insight into the dependency of fingerprint quality and demographics, and show that there in fact exists a certain degree of performance variability in fingerprint-based recognition systems for different segments of the population. Based on the experimental evaluation, the work points out new observations based on data-driven evidence, provides plausible hypotheses to explain such observations, and concludes with potential follow-up actions that can help to reduce the observed fingerprint quality differences. This way, the current paper can be considered as a contribution to further increase the algorithmic fairness and equality of biometric technology., Comment: Extended journal version submitted to IET Biometrics. 10 pages, 5 figures Reference conference paper: J. Galbally, A. Cepilovs, R. Blanco-Gonzalo, G. Ormiston, O. Miguel-Hurtado, and I. S. Racz, 'Fingerprint quality per individual finger type: A large-scale study on real operational data' in Proc. IEEE Intl. Workshop on Biometrics and Forensics 2023 (IWBF 2023)

Forecasting the semantics and 3D structure of scenes is essential for robots to navigate and plan actions safely. Recent methods have explored semantic and panoptic scene forecasting; however, they do not consider the geometry of the scene. In this work, we propose the panoptic-depth forecasting task for jointly predicting the panoptic segmentation and depth maps of unobserved future frames, from monocular camera images. To facilitate this work, we extend the popular KITTI-360 and Cityscapes benchmarks by computing depth maps from LiDAR point clouds and leveraging sequential labeled data. We also introduce a suitable evaluation metric that quantifies both the panoptic quality and depth estimation accuracy of forecasts in a coherent manner. Furthermore, we present two baselines and propose the novel PDcast architecture that learns rich spatio-temporal representations by incorporating a transformer-based encoder, a forecasting module, and task-specific decoders to predict future panoptic-depth outputs. Extensive evaluations demonstrate the effectiveness of PDcast across two datasets and three forecasting tasks, consistently addressing the primary challenges. We make the code publicly available at https://pdcast.cs.uni-freiburg.de.

Obtaining accurate forecasts for the evolution of epidemic outbreaks from deterministic compartmental models represents a major theoretical challenge. Recently, it has been shown that these models typically exhibit trajectories' degeneracy, as different sets of epidemiological parameters yield comparable predictions at early stages of the outbreak but disparate future epidemic scenarios. Here we use the Doi-Peliti approach and extend the classical deterministic SIS and SIR models to a quantum-like formalism to explore whether the uncertainty of epidemic forecasts is also shaped by the stochastic nature of epidemic processes. This approach allows getting a probabilistic ensemble of trajectories, revealing that epidemic uncertainty is not uniform across time, being maximal around the epidemic peak and vanishing at both early and very late stages of the outbreak. Our results therefore show that, independently of the models' complexity, the stochasticity of contagion and recover processes poses a natural constraint for the uncertainty of epidemic forecasts.

Light can be squeezed by reducing the quantum uncertainty of the electric field for some phases. We show how to use this purely-quantum effect to extract net mechanical work from radiation pressure in a simple quantum photon engine. Along the way, we demonstrate that the standard definition of work in quantum systems is not appropriate in this context, as it does not capture the energy leaked to these quantum degrees of freedom. We use these results to design an Otto engine able to produce mechanical work from squeezing baths, in the absence of thermal gradient. Interestingly, while work extraction from squeezing generally improves for low temperatures, there exists a nontrivial squeezing-dependent temperature for which work production is maximal, demonstrating the complex interplay between thermal and squeezing effects., Comment: 5 pages of main text and 3 more of SM

Deformable image registration is a standard engineering problem used to determine the distortion experienced by a body by comparing two images of it in different states. This study introduces two new DIR methods designed to capture non-affine deformations using Radon transform-based similarity measures and a classical regularizer based on linear elastic deformation energy. It establishes conditions for the existence and uniqueness of solutions for both methods and presents synthetic experimental results comparing them with a standard method based on the sum of squared differences similarity measure. These methods have been tested to capture various non-affine deformations in images, both with and without noise, and their convergence rates have been analyzed. Furthermore, the effectiveness of these methods was also evaluated in a lung image registration scenario.

Time crystals are many-body systems that spontaneously break time-translation symmetry, and thus exhibit long-range spatiotemporal order and robust periodic motion. Recent results have demonstrated how to build time-crystal phases in driven diffusive fluids using an external packing field coupled to density fluctuations. Here we exploit this mechanism to engineer and control on-demand custom continuous time crystals characterized by an arbitrary number of rotating condensates, which can be further enhanced with higher-order modes. We elucidate the underlying critical point, as well as general properties of the condensates density profiles and velocities, demonstrating a scaling property of higher-order traveling condensates in terms of first-order ones. We illustrate our findings by solving the hydrodynamic equations for various paradigmatic driven diffusive systems, obtaining along the way a number of remarkable results, e.g. the possibility of explosive time crystal phases characterized by an abrupt, first-order-type transition. Overall, these results demonstrate the versatility and broad possibilities of this promising route to time crystals.

Humans and many animals show remarkably adaptive behavior and can respond differently to the same input depending on their internal goals. The brain not only represents the intermediate abstractions needed to perform a computation but also actively maintains a representation of the computation itself (task abstraction). Such separation of the computation and its abstraction is associated with faster learning, flexible decision-making, and broad generalization capacity. We investigate if such benefits might extend to neural networks trained with task abstractions. For such benefits to emerge, one needs a task inference mechanism that possesses two crucial abilities: First, the ability to infer abstract task representations when no longer explicitly provided (task inference), and second, manipulate task representations to adapt to novel problems (task recomposition). To tackle this, we cast task inference as an optimization problem from a variational inference perspective and ground our approach in an expectation-maximization framework. We show that gradients backpropagated through a neural network to a task representation layer are an efficient heuristic to infer current task demands, a process we refer to as gradient-based inference (GBI). Further iterative optimization of the task representation layer allows for recomposing abstractions to adapt to novel situations. Using a toy example, a novel image classifier, and a language model, we demonstrate that GBI provides higher learning efficiency and generalization to novel tasks and limits forgetting. Moreover, we show that GBI has unique advantages such as preserving information for uncertainty estimation and detecting out-of-distribution samples.

As language models become more general purpose, increased attention needs to be paid to detecting out-of-distribution (OOD) instances, i.e., those not belonging to any of the distributions seen during training. Existing methods for detecting OOD data are computationally complex and storage-intensive. We propose a novel soft clustering approach for OOD detection based on non-negative kernel regression. Our approach greatly reduces computational and space complexities (up to 11x improvement in inference time and 87% reduction in storage requirements) and outperforms existing approaches by up to 4 AUROC points on four different benchmarks. We also introduce an entropy-constrained version of our algorithm, which leads to further reductions in storage requirements (up to 97% lower than comparable approaches) while retaining competitive performance. Our soft clustering approach for OOD detection highlights its potential for detecting tail-end phenomena in extreme-scale data settings.

Learning continually from a stream of non-i.i.d. data is an open challenge in deep learning, even more so when working in resource-constrained environments such as embedded devices. Visual models that are continually updated through supervised learning are often prone to overfitting, catastrophic forgetting, and biased representations. On the other hand, large language models contain knowledge about multiple concepts and their relations, which can foster a more robust, informed and coherent learning process. This work proposes Continual Visual Mapping (CVM), an approach that continually ground vision representations to a knowledge space extracted from a fixed Language model. Specifically, CVM continually trains a small and efficient visual model to map its representations into a conceptual space established by a fixed Large Language Model. Due to their smaller nature, CVM can be used when directly adapting large visual pre-trained models is unfeasible due to computational or data constraints. CVM overcome state-of-the-art continual learning methods on five benchmarks and offers a promising avenue for addressing generalization capabilities in continual learning, even in computationally constrained devices.

Photonic neural networks offer a promising alternative to traditional electronic systems for machine learning accelerators due to their low latency and energy efficiency. However, the challenge of implementing the backpropagation algorithm during training has limited their development. To address this, alternative machine learning schemes, such as extreme learning machines (ELMs), have been proposed. ELMs use a random hidden layer to increase the feature space dimensionality, requiring only the output layer to be trained through linear regression, thus reducing training complexity. Here, we experimentally demonstrate a programmable photonic extreme learning machine (PPELM) using a hexagonal waveguide mesh, and which enables to program directly on chip the input feature vector and the random hidden layer. Our system also permits to apply the nonlinearity directly on-chip by using the systems integrated photodetecting elements. Using the PPELM we solved successfully three different complex classification tasks. Additioanlly, we also propose and demonstrate two techniques to increase the accuracy of the models and reduce their variability using an evolutionary algorithm and a wavelength division multiplexing approach, obtaining excellent performance. Our results show that programmable photonic processors may become a feasible way to train competitive machine learning models on a versatile and compact platform.

We present an efficient implementation for running three-dimensional numerical simulations of fluid-structure interaction problems on single GPUs, based on Nvidia CUDA through Numba and Python. The incompressible flow around moving bodies is solved in this framework through an implementation of the Immersed Boundary Method tailored for the GPU, where different GPU grid architectures are exploited to optimize the overall performance. By targeting a single-GPU, we avoid GPU-CPU and GPU-GPU communication bottlenecks, since all the simulation data is always in the global memory of the GPU. We provide details about the numerical methodology, the implementation of the algorithm in the GPU and the memory management, critical in single-GPU implementations. Additionally, we verify the results comparing with our analogous CPU-based parallel solver and assess satisfactorily the efficiency of the code in terms of the relative computing time of the different operations and the scaling of the CPU code compared to a single GPU case. Overall, our tests show that the single-GPU code is between 34 to 54 times faster than the CPU solver in peak performance (96-128 CPU cores). This speedup mainly comes from the change in the method of solution of the linear systems of equations, while the speedup in sections of the algorithm that are equivalent in the CPU and GPU implementations is more modest (i.e., $\times 1.6-3$ speedup in the computation of the non-linear terms). Finally, we showcase the performance of this new GPU implementation in two applications of interest, one for external flows (i.e., bioinspired aerodynamics) and one for internal flows (i.e., cardiovascular flows), demonstrating the strong scaling of the code in two different GPU cards (hardware).%\red{for both single and double precision.}, Comment: 41 pages, 16 figures

In this work, we develop a comprehensive framework for F10.7, S10.7, M10.7, and Y10.7 solar driver forecasting with a time series Transformer (PatchTST). To ensure an equal representation of high and low levels of solar activity, we construct a custom loss function to weight samples based on the distance between the solar driver's historical distribution and the training set. The solar driver forecasting framework includes an 18-day lookback window and forecasts 6 days into the future. When benchmarked against the Space Environment Technologies (SET) dataset, our model consistently produces forecasts with a lower standard mean error in nearly all cases, with improved prediction accuracy during periods of high solar activity. All the code is available on Github https://github.com/ARCLab-MIT/sw-driver-forecaster., Comment: Short paper accepted for oral presentation at the SPAICE Conference 2024 (https://spaice.esa.int/)

Quantum valley Hall (QVH) domain wall states are a new class of one-dimensional (1D) one-way conductors that are topologically protected in the absence of valley mixing. Development beyond a single QVH channel raises important new questions as to how QVH channels in close spatial proximity interact with each other, and how that interaction may be controlled. Scalable epitaxial bilayer graphene synthesis produces layer stacking wall (LSW) bundles, where QVH channels are bound, providing an excellent platform to study QVH channel interactions. Here we show that distinct strain sources lead to the formation of both well-separated LSWs and close packed LSW bundles. Comparative studies of electronic transport in these two regimes reveal that close-packed LSW bundles support electrically tunable magnetoconductance. The coexistence of different strain sources offers a potential pathway to realize scalable quantum transport platform based on LSWs where electrically tunability enables programmable functionality.

Deep-learning (DL)-based image deconvolution (ID) has exhibited remarkable recovery performance, surpassing traditional linear methods. However, unlike traditional ID approaches that rely on analytical properties of the point spread function (PSF) to achieve high recovery performance - such as specific spectrum properties or small conditional numbers in the convolution matrix - DL techniques lack quantifiable metrics for evaluating PSF suitability for DL-assisted recovery. Aiming to enhance deconvolution quality, we propose a metric that employs a non-linear approach to learn the invertibility of an arbitrary PSF using a neural network by mapping it to a unit impulse. A lower discrepancy between the mapped PSF and a unit impulse indicates a higher likelihood of successful inversion by a DL network. Our findings reveal that this metric correlates with high recovery performance in DL and traditional methods, thereby serving as an effective regularizer in deconvolution tasks. This approach reduces the computational complexity over conventional condition number assessments and is a differentiable process. These useful properties allow its application in designing diffractive optical elements through end-to-end (E2E) optimization, achieving invertible PSFs, and outperforming the E2E baseline framework., Comment: Accepted at EUSIPCO 2024

We extend methods of Ding and Smart from their breakthrough paper in 2020 which showed Anderson localization for certain random Schr\"odinger operators on $\ell^2(\mathbb{Z}^2)$ via a quantitative unique continuation principle and Wegner estimate. We replace the requirement of identical distribution with the requirement of a uniform bound on the essential range of potential and a uniform positive lower bound on the variance of the variables giving the potential. Under those assumptions, we recover the unique continuation and Wegner lemma results, using Bernoulli decompositions and modifications of the arguments therein. This leads to a localization result at the bottom of the spectrum., Comment: 49 pages, comments welcome

We show that real semi-simple Lie groups of higher rank contain (infinitely generated) discrete subgroups with full limit sets in the corresponding Furstenberg boundaries. Additionally, we provide criteria under which discrete subgroups of $G = \operatorname{SL}(3,\mathbb{R})$ must have a full limit set in the Furstenberg boundary of $G$. In the appendix, we show the the existence of Zariski-dense discrete subgroups $\Gamma$ of $\operatorname{SL}(n,\mathbb{R})$, where $n\ge 3$, such that the Jordan projection of some loxodromic element $\gamma \in\Gamma$ lies on the boundary of the limit cone of $\Gamma$., Comment: Comments are welcome!

Soft materials play an integral part in many aspects of modern life including autonomy, sustainability, and human health, and their accurate modeling is critical to understand their unique properties and functions. Today's finite element analysis packages come with a set of pre-programmed material models, which may exhibit restricted validity in capturing the intricate mechanical behavior of these materials. Regrettably, incorporating a modified or novel material model in a finite element analysis package requires non-trivial in-depth knowledge of tensor algebra, continuum mechanics, and computer programming, making it a complex task that is prone to human error. Here we design a universal material subroutine, which automates the integration of novel constitutive models of varying complexity in non-linear finite element packages, with no additional analytical derivations and algorithmic implementations. We demonstrate the versatility of our approach to seamlessly integrate innovative constituent models from the material point to the structural level through a variety of soft matter case studies: a frontal impact to the brain; reconstructive surgery of the scalp; diastolic loading of arteries and the human heart; and the dynamic closing of the tricuspid valve. Our universal material subroutine empowers all users, not solely experts, to conduct reliable engineering analysis of soft matter systems. We envision that this framework will become an indispensable instrument for continued innovation and discovery within the soft matter community at large.

In a knowledge society, the term knowledge must be considered a core resource for organizations. So, beyond being a medium to progress and to innovate, knowledge is one of our most important resources: something necessary to decide.Organizations that are embracing knowledge retention activities are gaining a competitive advantage. Organizational rearrangements from companies, notably outsourcing, increase a possible loss of knowledge, making knowledge retention an essential need for them. When Knowledge is less shared, collaborative decision-making seems harder to obtain insofar as a ``communication breakdown'' characterizes participants' discourse. At best, stakeholders have to finda consensus according to their knowledge. Sharing knowledge ensures its retention and catalyzes the construction of this consensus. Our vision of collaborative decision-making aims not only at increasing the quality of the first parts of the decision-making process: intelligence and design, but also at increasing the acceptance of the choice. Intelligence and design will be done by more than one individual and constructed together; the decision is more easily accepted. The decided choice will then be shared. Thereby where decision-making could be seen as a constructed model, collaborative decision-making, for us,is seen as the use of socio-technical media to improve decision-making performance and acceptability. The shared decision making is a core activity in a lot of human activities. For example, the sustainable decision-making is the job of not only governments and institutions but also broader society. Recognizing the urgent need for sustainability, we can argue that to realize sustainable development, it must be considered as a decision-making strategy. The location of knowledge in the realization of collaborative decision-making has to be regarded insofar as knowledge sharing leads to improve collaborative decision-making: a ``static view'' has to be structured and constitutes the ``collaborative knowledge.'' Knowledge has an important role in individual decision-making, and we consider that for collaborative decision-making, knowledge has to be shared. What is required is a better understanding of the nature of group work''. Knowledge has to be shared, but how do we share knowledge?, Comment: Decisioning 2023 is the second workshop on Collaboration in knowledge discovery and decision-making, organized by six research teams from France, Argentina, and Colombia to explore the current frontier of knowledge and applications in different areas related to knowledge discovery and decision-making

This paper discusses the stability assessment of low-inertia power systems through a real-world large-scale low-inertia system, namely, the All-Island power system (AIPS) of Ireland and Northern Ireland. This system currently accommodates world-record levels of system non-synchronous penetration namely 75% (planning to increase to 80% next year). The paper discusses one-month results obtained with the state-of-the-art stability tool called look-ahead security assessment (LSAT). This tool carries out rotor-angle, frequency and voltage stability analyses and is implemented in the control centres of the transmission system operators (TSOs). The paper shows that, at the time of writing, the main binding stability constraint of the AIPS is related to the limits on the rate of change of frequency (RoCoF).

This paper discusses emerging operational challenges associated with the integration of solar photovoltaic (PV) in the All-Island power system (AIPS) of Ireland and Northern Ireland. These include the impact of solar PV on: (i) dispatch down levels; (ii) long-term frequency deviations; (iii) voltage magnitude variations; and (iv) operational demand variations. A case study based on actual data from the AIPS is used to analyze the above challenges. It is shown that despite its (still) relatively low penetration compared to wind power penetration, solar PV is challenging the real-time operation of the AIPS, e.g., maintaining frequency within operational limits. EirGrid and SONI, the transmission system operators (TSOs) of the AIPS, are working toward addressing all the above challenges.

In this PhD thesis, I investigate the properties of symmetry-breaking dynamical phase transitions that manifest in the fluctuations of time-integrated observables within classical systems. In particular, I analyze how these phase transitions impose stringent constraints on the structure of the eigenvectors of the system dynamical generator of the dynamics. Additionally, I identify a dynamical phase transition to a time-crystal phase in a model of driven-diffusive lattice gas. The study of this transition then allows the identification of the "packing-field" mechanism responsible for its emergence. This mechanism is then exploited to propose new transport models displaying time-crystal behavior., Comment: PhD thesis

This paper introduces a novel Functional Graph Convolutional Network (funGCN) framework that combines Functional Data Analysis and Graph Convolutional Networks to address the complexities of multi-task and multi-modal learning in digital health and longitudinal studies. With the growing importance of health solutions to improve health care and social support, ensure healthy lives, and promote well-being at all ages, funGCN offers a unified approach to handle multivariate longitudinal data for multiple entities and ensures interpretability even with small sample sizes. Key innovations include task-specific embedding components that manage different data types, the ability to perform classification, regression, and forecasting, and the creation of a knowledge graph for insightful data interpretation. The efficacy of funGCN is validated through simulation experiments and a real-data application.

Hyperparameter selection in continual learning scenarios is a challenging and underexplored aspect, especially in practical non-stationary environments. Traditional approaches, such as grid searches with held-out validation data from all tasks, are unrealistic for building accurate lifelong learning systems. This paper aims to explore the role of hyperparameter selection in continual learning and the necessity of continually and automatically tuning them according to the complexity of the task at hand. Hence, we propose leveraging the nature of sequence task learning to improve Hyperparameter Optimization efficiency. By using the functional analysis of variance-based techniques, we identify the most crucial hyperparameters that have an impact on performance. We demonstrate empirically that this approach, agnostic to continual scenarios and strategies, allows us to speed up hyperparameters optimization continually across tasks and exhibit robustness even in the face of varying sequential task orders. We believe that our findings can contribute to the advancement of continual learning methodologies towards more efficient, robust and adaptable models for real-world applications.

We report a multi-modal spiking neuron that allows optical and electronic input and control, and wavelength-multiplexing operation, for use in novel high-speed neuromorphic sensing and computing functionalities. The photonic-electronic neuron is built with a micro-scale, nanostructure resonant tunnelling diode (RTD) with photodetection (PD) capability. Leveraging the advantageous intrinsic properties of this RTD-PD system, namely highly nonlinear characteristics, photo-sensitivity, light-induced I-V curve shift, and the ability to deliver excitable responses under electrical and optical inputs, we successfully achieve flexible neuromorphic spike activation and inhibition regimes through photonic-electrical control. We also demonstrate the ability of this RTD-PD spiking sensing-processing neuron to operate under the simultaneous arrival of multiple wavelength-multiplexed optical signals, due to its large photodetection spectral window (covering the 1310 and 1550 nm telecom wavelength bands). Our results highlight the potential of RTD photonic-electronic neurons to reproduce multiple key excitatory and inhibitory spiking regimes, at high speed (ns-rate spiking responses, with faster sub-ns regimes theoretically predicted) and low energy (requiring only ~10 mV and ~150 microW, electrical and optical input amplitudes, respectively), similar in nature to those commonly found in the biological neurons of the visual system and the brain. This work offers a highly promising approach for the realisation of high-speed, energy-efficient photonic-electronic spiking neurons and spiking neural networks, enabling multi-modal and multi-wavelength operation for sensing and information processing tasks. This work therefore paves the way for innovative high-speed, photonic-electronic, and spike-based neuromorphic sensing and computing systems and artificial intelligence hardware., Comment: 12 pages, 9 figures

Earth's magnetosphere, beyond protecting the ozone layer, is a natural phenomena which allows to study the interaction between charged particles from solar activity and electromagnetic fields. In this paper we studied trajectories of charged particles interacting with a constant dipole magnetic field as first approach of the Earth's magnetosphere using different initial conditions. As a result of this interaction there is a formation of well defined radiation regions by a confinement of charged particles around the lines of the magnetic field. These regions, called Van Allen radiation belts, are described by classical electrodynamics and appear naturally in the numerical modeling done in this work., Comment: Matches the version published in the Revista Mexicana de F\'isica E

Teaching magnetism is one of the most challenging topics at undergraduate level in programmes with scientific background. A basic course includes the description of the magnetic interaction along with empirical results such as the Biot-Savart law's. However, evaluating the magnetic field due to certain current carrying system at any point in space is not an easy task, especially for points in space where symmetry arguments cannot be applied. In this paper we study the magnetic field produced by both Helmholtz and Maxwell coils at all points in space by using a hybrid methodology that combines the superposition principle and an analytical result. We implement a computational approach, that is based on iterating $n$ times the magnetic field produced by a finite current-carrying wire, to evaluate the magnetic field at any point in space for coils arrangements without using advanced calculus. This methodology helps teachers and students to explore the field due to systems with different levels of complexity, combining analytical and computational skills to visualize and analyse the magnetic field. After our analysis, we show that this is an useful approach to emphasize fundamental concepts and mitigate some of the issues that arise when evaluating the magnetic field for systems proposed in introductory physics textbooks., Comment: Matches the version published in Revista Brasileira de Ensino de F\'isica

Targeted therapeutic interventions utilizing low-inten\-sity ultrasound (LIUS) exhibit substantial potential for hindering the proliferation of cancer stem cells. This investigation introduces a multiscale model and computational framework to comprehensively explore the therapeutic LIUS on poroelastic tumor dynamics, thereby unraveling the intricacies of mechanotransduction mechanisms at play. Our model includes both macroscopic timescales encompassing days and rapid timescales spanning from microseconds to seconds, facilitating an in-depth comprehension of tumor behavior. We unveil the discerning suppression or reorientation of cancer cell proliferation and migration, enhancing a notable redistribution of cellular phases and stresses within the tumor microenvironment. Our findings defy existing paradigms by elucidating the impact of LIUS on cancer stem cell behavior. This endeavor advances our fundamental understanding of mechanotransduction phenomena in the context of LIUS therapy, thus underscoring its promising as a targeted therapeutic modality for cancer treatment. Furthermore, our results make a substantial contribution to the broader scientific community by shedding light on the intricate interplay between mechanical forces, cellular responses, and the spatiotemporal evolution of tumors. These insights hold the promising to promote a new perspective for the future development of pioneering and highly efficacious therapeutic strategies for combating cancer in a personalized manner.