Pasar al contenido principal

COMPUTACION DE ALTAS PRESTACIONES, HETEROGENEA Y EN LA NUBE PARA APLICACIONES DE ALTA DEMANDA

PID2022-141623NB-I00

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprograma Subprograma Estatal de Generación de Conocimiento
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2022
Plan Estatal 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSIDADE DE SANTIAGO DE COMPOSTELA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 24
Encontrada(s) 1 página(s)

Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Deep learning with simulated laser scanning data for 3D point cloud classification

  • Esmorís Pena, Alberto Manuel
  • Weiser, Hannah
  • Winiwarter, Lukas
  • Cabaleiro Domínguez, José Carlos
  • Höfle, Bernhard
Laser scanning is an active remote sensing technique applied in many disciplines to acquire state-of-the-art spatial measurements. Semantic labeling is often necessary to extract information from the raw point cloud. Deep learning methods constitute a data-hungry solution for the semantic segmentation of point clouds. In this work, we investigate the use of simulated laser scanning for training deep learning models, which are applied to real data subsequently. We show that training a deep learning model purely on virtual laser scanning data can produce results comparable to models trained on real data when evaluated on real data. For leaf-wood segmentation of trees, using the KPConv model trained with virtual data achieves 93.7% overall accuracy, while the model trained on real data reaches 94.7% overall accuracy. In urban contexts, a KPConv model trained on virtual data achieves 74.1% overall accuracy on real validation data, while the model trained on real data achieves 82.4%. Our models outperform the state-of-the-art model FSCT in terms of generalization to unseen real data as well as a baseline model trained on points randomly sampled from the tree mesh surface. From our results, we conclude that the combination of laser scanning simulation and deep learning is a cost-effective alternative to real data acquisition and manual labeling in the domain of geospatial point cloud analysis. The strengths of this approach are that (a) a large amount of diverse laser scanning training data can be generated quickly and without the need for expensive equipment, (b) the simulation configurations can be adapted so that the virtual training data have similar characteristics to the targeted real data, and (c) the whole workflow can be automated through procedural scene generation.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Deep Learning for Ultra-Large-Scale Semantic Segmentation of Geographic 3D Point Clouds With Missing Labels

  • Esmorís Pena, Alberto Manuel
  • Yermo, Miguel
  • Rodríguez Alcaraz, Silvia
  • Soutullo, Samuel
  • Fernández Rivera, Francisco
Semantic segmentation of 3D point clouds is a critical task essential for research and industry in a wide variety of domains. Most works until now use datasets where the concept of large-scale varies between 1.17 and 9,261 million points and 0.31 and 250 square kilometers. In this research, we push the limits of 3D deep learning by considering approximately 36,369 million points and 29,557 square kilometers in our case study. To the best of our knowledge, it is the largest geographic region ever classified with neural networks for 3D point clouds in scientific works. The main contributions of our research are: 1) The first published results on semantic segmentation of ultra-large-scale and low-resolution airborne laser scanning point clouds, 2) adapted neural networks to work under missing labels and high measurement error conditions, 3) the introduction of class ambiguity as a more robust uncertainty measurement compared to entropy, and 4) an open-source deep learning framework for 3D point clouds that enables the reproducibility of our experiments and further studies.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

NetQIR: An extension of QIR for distributed quantum computing

  • Cardama Santiago, Francisco Javier
  • Vázquez Pérez, Jorge
  • Piñeiro Pomar, César Alfredo
  • Fernández Pena, Anselmo Tomás
  • Pichel Campos, Juan Carlos
  • Gómez, Andrés
The rapid advancement of quantum computing has highlighted the need for scalable and efficient software infrastructures to fully exploit its potential. Current quantum processors face significant scalability constraints due to the limited number of qubits per chip. In response, distributed quantum computing (DQC) — achieved by networking multiple quantum processor units (QPUs)— is emerging as a promising solution. To support this paradigm, robust intermediate representations (IRs) are needed to translate high-level quantum algorithms into executable instructions suitable for distributed systems. This paper presents NetQIR, an extension of Microsoft’s Quantum Intermediate Representation (QIR), specifically designed to facilitate DQC by incorporating new instruction specifications. NetQIR was developed in response to the lack of abstraction at the network and hardware layers identified in the existing literature as a significant obstacle to effectively implementing distributed quantum algorithms. Based on this analysis, NetQIR introduces new essential abstraction features to support compilers in DQC contexts. It defines network communication instructions independent of specific hardware, abstracting the complexities of inter-QPU communication. Although the proposed work allows abstraction of the underlying network, it is important to note that it is intended for the development of high-performance code on future modular quantum architectures. Leveraging the QIR framework, NetQIR aims to bridge the gap between high-level quantum algorithm design and low-level hardware execution, thus promoting modular and scalable approaches to quantum software infrastructures for distributed applications. Furthermore, its design may serve as a foundational component for future implementations of distributed quantum standards such as the Quantum Message Passing Interface (QMPI).
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Capítulo o parte de un libro (BookPart)

Estudio preliminar del registro de alta precisión de imágenes multitemporales de observación terrestre

  • Fuentes, Daniel
  • Ordóñez Iglesias, Álvaro
  • Blanco Heras, Dora
  • Argüello Pedreira, Francisco
El creciente uso de varias imágenes multidimensionales de teledetección para una misma zona geográfica con el objetivo de realizar procesos de observación terrestre hace imprescindible realizar un alineamiento previo de dichas imágenes. El alineamiento, conocido también como registro, consiste en, dadas dos imágenes, explotar la información espacial y espectral de la zona capturada para obtener la transformación geométrica que permita alinear una imagen respecto de la otra. En este artículo se propone un análisis previo del registro de alta precisión de imágenes multitemporales basado en la combinación de métodos ya existentes en la literatura. Se realizará sobre imágenes multiespectrales de alta resolución espacial capturadas en diferentes fechas. Las imágenes presentan diferentes condiciones de captura tanto lumínicas como de inclinación que suponen un reto en cuanto a su registro. Mediante una evaluación sistemática de los métodos disponibles en la literatura seleccionamos los más adecuados para ser aplicados de manera consecutiva. Se mide la calidad del registro mediante la evaluación de su precisión, la robustez ante solapamientos de imágenes y el tiempo de ejecución. Finalmente se selecciona la combinación de métodos que proporciona la más alta precisión de registro.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

A novel machine learning workflow to optimize cooling devices grounded in solid-state physics

  • García Fernández, Julián
  • Etesse, Guéric
  • Seoane Iglesias, Natalia
  • Comesaña Figueroa, Enrique
  • Hirakawa, Kazuhiko
  • García Loureiro, Antonio Jesús
  • Bescond, Marc
Cooling devices grounded in solid-state physics are promising candidates for integrated-chip nanocooling applications. These devices are modeled by coupling the quantum non-equilibirum Green’s function for electrons with the heat equation (NEGF+H), which allows to accurately describe the energetic and thermal properties. We propose a novel machine learning (ML) workflow to accelerate the design optimization process of these cooling devices, alleviating the high computational demands of NEGF+H. This methodology, trained with NEGF+H data, obtains the optimum heterostructure designs that provide the best trade-off between the cooling power of the lattice (CP) and the electron temperature (Te). Using a vast search space of 1.18 × 10−5 different device configurations, we obtained a set of optimum devices with prediction relative errors lower than 4 % for CP and 1 % for Te. The ML workflow reduces the computational resources needed, from two days for a single NEGF+H simulation to 10 s to find the optimum designs.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

EMViT-DDPM: An Equilibrium-Based ViT Diffusion Framework for Data Augmentation in Multispectral Land Cover Classification

  • Barreiro Domínguez, Víctor Xesús
  • Blanco Heras, Dora
  • Argüello Pedreira, Francisco
The scarcity and imbalance of labeled samples, common in remote sensing datasets, pose significant challenges for accurate analysis and classification, often leading to substantial bias against minority classes. To address these issues, we propose EMViT-DDPM, an equilibrium-based data augmentation framework leveraging vision transformer (ViT)-based denoising diffusion probabilistic models (DDPMs). Unlike generative adversarial network (GAN)-based augmentation techniques commonly found in the literature, our framework is not tied to any specific classifier. It leverages a ViT architecture enhanced with the AdaLN block, which is designed to minimize computational costs while effectively capturing data complexity. By adopting diffusion models (DDPMs), the framework achieves greater training robustness, improved generalization, and better quality control over generated samples compared to GANs. To address class imbalance, we introduce equilibrium-based data augmentation (), which assigns different augmentation proportions to each class based on their respective sizes. In addition, a superpixel-based segmentation preprocessing step is proposed for patch generation, tailoring the data augmentation method specifically to high-spatial-resolution multispectral remote sensing imagery. Finally, we propose a novel strategy for the evaluation of data-augmentation quality, based on a new judge model trained over balanced classes. This allows a more precise evaluation of Fréchet inception distance, precision (fidelity), and recall (diversity)
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

BigOPERA: An OPportunistic and Elastic Resource Allocation for big data frameworks

  • VázquezCaderno, Pablo
  • Awaysheh, Feras
  • Cabaleiro Domínguez, José Carlos
  • Fernández Pena, Anselmo Tomás
Efficient asset management is essential for optimizing the performance and scalability of modern Big Data (BD) frameworks. However, traditional resource allocation methods often suffer from static partitioning, inefficient resource utilization, and high operational costs, limiting their ability to adapt to fluctuating workloads dynamically. This paper introduces BigOPERA, an opportunistic and elastic resource allocation framework designed to enhance BD processing environments by integrating dedicated and non-dedicated computing assets. Leveraging containerization and a two-tiered scheduling mechanism, BigOPERA dynamically manages available resources to improve workload execution efficiency. Experimental results demonstrate that BigOPERA achieves up to 35% performance improvement over native Apache Spark configurations, significantly enhancing computational throughput while optimizing resource consumption. Our findings highlight the potential of BigOPERA in scalable, cost-effective, and sustainable BD processing.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Distributed multi-GPU algorithm for accurate registration of UAV-based multispectral and multitemporal orthomosaics

  • Fuentes Rodríguez, Daniel
  • Ordóñez Iglesias, Álvaro
  • Blanco Heras, Dora
  • Argüello Pedreira, Francisco
Accurate registration of high-resolution multispectral UAV orthomosaics acquired on different dates and sensors is essential for a wide range of remote sensing applications. However, this task remains challenging due to variations in acquisition conditions, including seasonal changes, differences in illumination, weather, and sensor characteristics. This article presents a parallel multilevel registration method that combines and improves two existing algorithms: HSI-KAZE, a feature-based approach, and HYFM, an area-based method. The three-level approach first applies an optimized HSI-KAZE (OHSI-KAZE) for coarse estimation of scale, rotation, and translation, followed by HYFM for fine correction. The multi-node multi-GPU proposed implementation, leveraging MPI, OpenMP, and CUDA, enables efficient processing on GPU-accelerated HPC clusters. Experiments on six real multispectral orthomosaic pairs from river environments, compared against state-of-the-art classical and deep learning methods, achieve high registration accuracy with RMSE values below 1.57 pixels and a 30 speedup, confirming both the accuracy and scalability of the proposed method
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Capítulo o parte de un libro (BookPart)

Coste computacional de una red de aprendizaje profundo basada en EfficientNet sobre dispositivos basados en GPU para computación en el borde

  • Vilela Pérez, Nicolás
  • Blanco Heras, Dora
  • Argüello Pedreira, Francisco
El despliegue de redes de aprendizaje profundo para la teledetección en entornos de computación en el borde sigue siendo un gran reto, ya que estas plataformas operan bajo restricciones de energía y de recursos de cómputo. Este desafío exige aplicar técnicas de optimización de redes y realizar un trabajo especialmente cuidadoso en el caso de redes diseñadas para operar con un coste computacional reducido. El problema se agrava en el caso de redes de tamaño medio, ya que la mayoría de las técnicas de optimización están orientadas a redes de gran escala. En este trabajo se analizan el coste computacional y la eficiencia energética generados por técnicas de optimización aplicadas a la red EffBaGAN, una red de tamaño medio basada en EfficientNet, orientada a la clasificación de imágenes multiespectrales de alta resolución espacial para la cobertura terrestre en escenarios de escasez y desequilibrio de datos. Se evalúa el impacto de la precisión mixta automática (AMP), la cuantización durante y después del entrenamiento (QAT y PTQ, respectivamente) y la poda, tanto de forma aislada como combinada, en una plataforma NVIDIA Jetson AGX Orin. Además, se analiza el rendimiento computacional en función de los límites de potencia del dispositivo. Los resultados revelan que la combinación del entrenamiento sensible a la cuantización (QAT) con la poda representa el enfoque más efectivo para lograr reducciones conjuntas en el tamaño de la red y en el número de parámetros y FLOPs, al mismo tiempo que se mantienen métricas de precisión competitivas.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Efficient Semantic Segmentation of Multispectral Land Cover Images Using Mask2Former

  • Canosa García, Pablo
  • Ordóñez Iglesias, Álvaro
  • Blanco Heras, Dora
  • Argüello Pedreira, Francisco
Semantic segmentation for EO is a process that involves assigning a specific label or category to each pixel in an image, enabling precise analysis for land cover applications such as environmental conservation, urban planning or disaster management. Deep learning-based segmentation models have proliferated in recent years, but they often are not well adapted to the unique properties of multi and hyperspectral images, frequently used in remote sensing. Mask2Former is a universal segmentation model based on the concept of masked attention and employs a pretrained classification model as backbone to create intermediate representations. This article presents a preliminary adaptation of Mask2Former for the segmentation of multispectral remote sensing images. This adaptation includes modifying the backbone to accept multispectral inputs and adapting the data processing pipelines to leverage all available spectral bands effectively. The computational cost of the method has also been analyzed as an initial assessment of potential scalability and efficiency for large-scale applications. Experimental results using the FiveBillionPixels dataset reveal a notable improvement in segmentation accuracy when incorporating multispectral bands, outperforming RGB-only performance without a relevant increase in computational cost.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Introducing the Taubin-Weingarten algorithm to compute second-order geometric descriptors in 3D point clouds

  • Esmorís Pena, Alberto Manuel
  • García-Martínez, Xabier
  • Ladra González, Manuel
  • Cabaleiro Domínguez, José Carlos
  • Fernández Rivera, Francisco
We propose the Taubin-Weingarten algorithm to compute second-order geometric features in 3D point clouds. This method is well-suited for working with large-scale 3D datasets due to its embarrassingly parallel nature. The speedup of our open-source C++ implementation is systematically above 30 for a high-performance CPU with 32 cores. It provides reliable estimates for standard quantifications in differential geometry, such as the Gaussian and mean curvatures of a surface, when compared to other approaches. Additionally, it achieves improvements of between 1.6% and 10% in F1-score compared to the standard first-order approach in point-wise classification tasks, such as object part segmentation and large-scale semantic scene segmentation. The results demonstrate that the Taubin-Weingarten algorithm is both efficient and robust, enabling consistent improvements in machine learning performance across various tasks.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

QPU integration in OpenCL for heterogeneous programming

  • Vázquez Pérez, Jorge
  • Piñeiro Pomar, César Alfredo
  • Pichel Campos, Juan Carlos
  • Fernández Pena, Anselmo Tomás
  • Gómez, Andrés
The integration of quantum processing units (QPUs) in a heterogeneous high-performance computing environment requires solutions that facilitate hybrid classical–quantum programming. Standards such as OpenCL facilitate the programming of heterogeneous environments, consisting of CPUs and hardware accelerators. This study presents an innovative method that incorporates QPU functionality into OpenCL, standardizing quantum processes within classical environments. By leveraging QPUs within OpenCL, hybrid quantum–classical computations can be sped up, impacting domains like cryptography, optimization problems, and quantum chemistry simulations. Using Portable Computing Language (Jääskeläinen et al. in Int J Parallel Program 43(5):752–785, 2014. https://doi.org/10.1007/s10766-014-0320-y) and the Qulacs library (Suzuki et al. in Quantum 5:559, 2021. https://doi.org/10.22331/q-2021-10-06-559), results demonstrate, for instance, the successful execution of Shor’s algorithm (Nielsen and Chuang in Quantum computation and quantum information, 10th anniversary edn. Cambridge University Press, Cambridge, 2010), serving as a proof of concept for extending the approach to larger qubit systems and other hybrid quantum–classical algorithms. This integration approach bridges the gap between quantum and classical computing paradigms, paving the way for further optimization and application to a wide range of computational problems.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Reconstruction of phylogenetic trees via graph-splitting using quantum computing

  • Fernández Otero, Nicolás
  • Fernández Pena, Anselmo Tomás
  • Pichel Campos, Juan Carlos
Quantum computing applies principles of quantum mechanics, such as superposition and entanglement, to process information with exponential parallelism. This paradigm offers significant computational advantages over classical methods, particularly for NP-hard problems like phylogenetic tree reconstruction in evolutionary biology. Phylogenetic trees model the evolutionary relationships among species or genes, and their reconstruction is computationally challenging as the number of possible topologies grows exponentially with the number of taxa. To address this, biologists often rely on heuristic methods; however, recent work has shown that recursive graph-cut techniques can achieve high accuracy in phylogenetic inference, though at high computational cost. In this study, we present a quantum algorithm based on the normalized cut ( ) criterion, enabling efficient recursive graph partitioning. Implemented using Quantum Annealing (QA) and the Quantum Approximate Optimization Algorithm (QAOA), demonstrating promising results on real quantum hardware for complex bioinformatics tasks.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Gallium nitride: a strong candidate to replace GaAs as base material for optical photovoltaic converters in space exploration

  • Fernández Lozano, Javier
  • Seoane Iglesias, Natalia
  • Moares Guedes, Jose Augusto
  • Comesaña Figueroa, Enrique
  • García Fernández, Julián
  • Almonacid Cruz, Florencia Marin
  • Fernández Fernández, Eduardo
  • García Loureiro, Antonio Jesús
High power laser transmission technology is expected to play an important role in spatial exploration. To increase the amount of power delivered, some issues must be addressed. Currently, optical photovoltaic converters are based on GaAs, a material with a bandgap energy of 1.42 eV. In this work we propose gallium nitride (GaN) as base material for optical photovoltaic converters due to its high bandgap (3.39 eV), which reduces both ohmic and intrinsic entropic losses, and its high thermal conductivity and resistance to radiation damage, making it suitable for space applications. We have optimized several GaN optical photovoltaic converter devices under a wide range of laser power densities and temperatures. The resilience to variations in the design parameters is also tested. Results show that, due to their large bandgap energy, GaN devices could suffer from fewer performance losses with the temperature when compared to other materials with lower bandgaps. The devices show great tolerance to variations in the P layer (bottom layer), while the N layer thickness and doping concentration must be carefully manufactured. When compared to GaAs-based devices, GaN shows higher efficiency across the entire laser power density range, achieving efficiencies near 80 % and surpassing the current state-of-the-art power converter by 10 % at 10 . The proposed GaN devices show a peak of performance at a laser power density as high as 100 . Although manufacturing issues could degrade the efficiency of GaN power converters, this results position GaN as a promising material for a new generation of ultra-high efficient optical photovoltaic converters.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Review of Distributed Quantum Computing: From single QPU to High Performance Quantum Computing

  • Barral, David
  • Cardama Santiago, Francisco Javier
  • Díaz-Camacho, Guillermo
  • Faílde, Daniel
  • Llovo, Iago F.
  • Mussa Juane, Mariamo
  • Vázquez Pérez, Jorge
  • Villasuso, Juan
  • Piñeiro Pomar, César Alfredo
  • Costas, Natalia
  • Pichel Campos, Juan Carlos
  • Fernández Pena, Anselmo Tomás
  • Gómez, Andrés
The emerging field of quantum computing has shown it might change how we process information by using the unique principles of quantum mechanics. As researchers continue to push the boundaries of quantum technologies to unprecedented levels, distributed quantum computing raises as an obvious path to explore with the aim of boosting the computational power of current quantum systems. This paper presents a comprehensive survey of the current state of the art in the distributed quantum computing field, exploring its foundational principles, landscape of achievements, challenges, and promising directions for further research. From quantum communication protocols to entanglement-based distributed algorithms, each aspect contributes to the mosaic of distributed quantum computing, making it an attractive approach to address the limitations of classical computing. Our objective is to offer a comprehensive review that serves both experts in the field and researchers or enthusiasts in quantum computing looking for a starting point to explore the area of distributed quantum computing.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

MLFoMpy: A post-processing tool for semiconductor TCAD data with machine-learning capabilities

  • Comesaña Figueroa, Enrique
  • García Fernández, Julián
  • Seoane Iglesias, Natalia
  • García Loureiro, Antonio Jesús
We present MLFoMpy, a Python package for post-processing data from semiconductor device simulations. The software automatically extracts key figures of merit from current–voltage curves of field effect transistor and calculates statistical analyses for these curves. MLFoMpy also includes machine learning tools to predict figures of merit and current–voltage curves for devices with intrinsic variability. Additionally, it offers data visualization tools to plot current–voltage curves and statistical graphs.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Attention-Based Convolutional Neural Network for Anomaly Detection in Multispectral Images of Semi-Natural Ecosystems

  • López Fandiño, Javier
  • Ordóñez Iglesias, Álvaro
  • Quesada Barriuso, Pablo
  • Suárez Garea, Jorge Alberto
  • Argüello Pedreira, Francisco
  • Blanco Heras, Dora
The monitoring of semi-natural ecosystems has become increasingly critical due to the rising impact of ecological disturbances, including natural disasters and unauthorized human-made constructions. Anomaly detection (AD) in multispectral imagery serves as a fundamental tool in this context. Deep-learning (DL)-based techniques are particularly effective at capturing the intricate spectral and spatial patterns of anomalies. This letter proposes a new AD technique called attention-based convolutional neural network (ACNN), designed to enhance AD performance in multispectral images of high spatial resolution for the detection of human-made constructions. The model integrates attention mechanisms to prioritize informative features while suppressing irrelevant background information, thereby improving sensitivity to subtle and rare anomalies. Experimental results on multispectral datasets from semi-natural ecosystems show that the proposed approach outperforms existing DL techniques in terms of detection accuracy. These findings highlight the potential of attention-based models as a robust framework for environmental monitoring and AD in complex remote sensing scenarios.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

A new thread-level speculative automatic parallelization model and library based on duplicate code execution

  • Martínez, Millan A.
  • Fraguela, Basilio B.
  • Cabaleiro Domínguez, José Carlos
  • Fernández Rivera, Francisco
Loop-efficient automatic parallelization has become increasingly relevant due to the growing number of cores in current processors and the programming effort needed to parallelize codes in these systems efficiently. However, automatic tools fail to extract all the available parallelism in irregular loops with indirections, race conditions or potential data dependency violations, among many other possible causes. One of the successful ways to automatically parallelize these loops is the use of speculative parallelization techniques. This paper presents a new model and the corresponding C++ library that supports the speculative automatic parallelization of loops in shared memory systems, seeking competitive performance and scalability while keeping user effort to a minimum. The primary speculative strategy consists of redundantly executing chunks of loop iterations in a duplicate fashion. Namely, each chunk is executed speculatively in parallel to obtain results as soon as possible and sequentially in a different thread to validate the speculative results. The implementation uses C++11 threads and it makes intensive use of templates and advanced multithreading techniques. An evaluation based on various benchmarks confirms that our proposal provides a competitive level of performance and scalability.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Capítulo o parte de un libro (BookPart)

Region-based multispectral image registration on heterogeneous computing platforms

  • Castillo de la Rosa, Daniel del
  • Ordóñez Iglesias, Álvaro
  • Blanco Heras, Dora
  • Argüello Pedreira, Francisco
Feature-based methods are widely used for the registration of remote sensing images because of their robustness to viewpoint, scale, and light changes. However, they are computationally demanding, especially when dealing with multi or hyperspectral images. Hyperspectral Maximally Stable Extremal Regions (HSI-MSER) is a hyperspectral remote sensing image registration method based on MSER for feature detection and Scale Invariant Feature Transform (SIFT) for feature description. This article presents a first approach to a parallel implementation of the HSI-MSER algorithm for the registration of multispectral images on a heterogeneous computing platform. The results of the registration capabilities under extreme scaling and rotating conditions show that the proposed parallel implementation obtains a speedup of 3.33 times compared to the sequential implementation making it suitable for applications with execution time constraints.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Capítulo o parte de un libro (BookPart)

Redes convolucionales basadas en escalado uniforme de dimensiones para la clasificación de imágenes multiespectrales

  • Vilela Pérez, Nicolás
  • Goldar Dieste, Álvaro
  • Blanco Heras, Dora
  • Argüello Pedreira, Francisco
En el campo de la clasificación de imágenes multi e hiperespectrales se ha popularizado en la última década el uso de técnicas basadas en aprendizaje profundo, en particular las basadas en redes convolucionales. Dado que su coste computacional es más alto que el de las técnicas de aprendizaje automático más clásicas, recientemente se han propuesto nuevos esquemas de clasificación que tienen como objetivo reducir el tiempo de ejecución requerido por ellas. Un modelo de red que ha sido aplicado con éxito a problemas de clasificación de imágenes en la literatura es la llamada red EfficientNet. Este tipo de redes de aprendizaje profundo se caracterizan por realizar un escalado uniforme de las tres dimensiones que componen la red: profundidad, anchura y resolución. De esta forma, se obtiene un modelo con buenas precisiones de clasificación a la vez que se cumplen las restricciones computacionales impuestas por el problema. En este trabajo se propone la adaptación de las redes tipo EfficientNet para resolver la clasificación de imágenes multiespectrales de alta resolución. Se analizan diferentes configuraciones de este tipo de redes variando el escalado de la arquitectura, tanto en términos de precisión de la clasificación del esquema resultante como de tiempo de ejecución. Los experimentos realizados sobre imágenes multiespectrales de alta resolución espacial obtenidas mediante vehículos aéreos no tripulados han mostrado que algunas variantes específicas de EfficientNet permiten una reducción importante del coste computacional sin degradar la precisión de clasificación.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Cheesemap: A high-performance point-indexing data structure for neighbor search in LiDAR data

  • Laso Rodríguez, Rubén
  • Yermo, Miguel
Point-cloud data, as the representation of three-dimensional spatial information, is a fundamental piece of information in various domains where indexing and querying these point clouds efficiently is crucial for tasks such as object recognition, autonomous navigation, and environmental modeling. In this paper, we present a novel data structure, cheesemap, designed for fast neighbor search in 3D LiDAR point clouds. Points are indexed using a grid of voxels, which can be organized in three different ways, originating three flavors of the cheesemap: dense, sparse, and mixed. The lookup of the voxels is theoretically ensured to be performed in constant or amortized constant time, speeding up the search for neighboring points. Experimental results show that cheesemap can outperform, in terms of performance and memory footprint, other state-of-the-art data structures both in region-based and k-NN queries throughout different types of point clouds, particularly for Airborne Laser Scanning (ALS) point clouds
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Review of intermediate representations for quantum computing

  • Cardama Santiago, Francisco Javier
  • Vázquez Pérez, Jorge
  • Piñeiro Pomar, César Alfredo
  • Pichel Campos, Juan Carlos
  • Fernández Pena, Anselmo Tomás
  • Gómez, Andrés
Intermediate representations (IRs) are fundamental to classical and quantum computing, bridging high-level quantum programming languages and the hardware-specific instructions required for execution. This paper reviews the development of quantum IRs, focusing on their evolution and the need for abstraction layers that facilitate portability and optimization. Monolithic quantum IRs, such as QIR (Lubinski et al. in Front Phys 10:940293, 2022. https://doi.org/10.3389/fphy.2022.940293), QSSA (Peduri et al. in Proceedings of the 31st ACM SIGPLAN international conference on compiler construction. CC 2022. Association for Computing Machinery, New York, 2022), or Q-MLIR (McCaskey and Nguyen in Proceedings-2021 IEEE International Conference on Quantum Computing and Engineering, QCE, 2021), their effectiveness in handling abstractions, and their hybrid support between quantum-classical operations are evaluated. However, a key limitation is their inability to address qubit locality, an essential feature for distributed quantum computing (DQC). To overcome this, InQuIR (Nishio and Wakizaka in InQuIR: Intermediate Representation for Interconnected Quantum Computers, 2023. https://arxiv.org/abs/2302.00267) was introduced as an IR specifically designed for distributed systems, providing explicit control over qubit locality and inter-node communication. While effective in managing qubit distribution, InQuIR’s dependence on manual manipulation of communication protocols increases complexity for developers. NetQIR (Vázquez-Pérez et al. in NetQIR: An Extension of QIR for Distributed Quantum Computing, 2024. https://arxiv.org/abs/2408.03712), an extension of QIR for DQC, emerges as a solution to achieve the abstraction of quantum communications protocols. This review emphasizes the need for further advancements in IRs for distributed quantum systems, which will play a crucial role in the scalability and usability of future quantum networks.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Inqasm: InQuIR compiler to NetQASM

  • Vázquez Pérez, Jorge
  • Cardama Santiago, Francisco Javier
  • Piñeiro Pomar, César Alfredo
  • Pichel Campos, Juan Carlos
  • Fernández Pena, Anselmo Tomás
  • Gómez, Andrés
Quantum computing is a rapidly evolving field, with almost every aspect open to change or improvement. This includes moving from using a single quantum processing unit to interconnecting multiple quantum processing units (or several of them), establishing a new paradigm called distributed quantum computing and increasing the overall computing capability. Some research is already underway in this area to prepare the ground for an eventual architecture with these characteristics. This is the case of InQuIR (Nishio and Wakizaka in arXiv:2302.00267 2023) and NetQASM (Dahlberg et al in QST 7:035023 2022), two languages developed for distributed quantum computing. This paper presents the development of the InQASM compiler with the aim of translating code from the InQuIR language to NetQASM, establishing a compilation stack for the new distributed paradigm. An example of this compilation and a simulation of the compiled code are shown to showcase it.
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

A new thread-level speculative automatic parallelization model and library based on duplicate code execution

  • Álvarez Martínez, Millán
  • Fraguela, Basilio B.
  • Cabaleiro, José Carlos
  • Rivera, Francisco F.
Funding for open access charge: Universidade da Coruña/CISUG, Loop-efficient automatic parallelization has become increasingly relevant due to the growing number of cores in current processors and the programming effort needed to parallelize codes in these systems efficiently. However, automatic tools fail to extract all the available parallelism in irregular loops with indirections, race conditions or potential data dependency violations, among many other possible causes. One of the successful ways to automatically parallelize these loops is the use of speculative parallelization techniques. This paper presents a new model and the corresponding C++ library that supports the speculative automatic parallelization of loops in shared memory systems, seeking competitive performance and scalability while keeping user effort to a minimum. The primary speculative strategy consists of redundantly executing chunks of loop iterations in a duplicate fashion. Namely, each chunk is executed speculatively in parallel to obtain results as soon as possible and sequentially in a different thread to validate the speculative results. The implementation uses C++11 threads and it makes intensive use of templates and advanced multithreading techniques. An evaluation based on various benchmarks confirms that our proposal provides a competitive level of performance and scalability., Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature. This research was supported by Grants PID2019-104184RB-I00, PID2019-104834GB-I00, PID2022-141623NB-I00, and PID2022-136435NB-I00, funded by MCIN/AEI/ 10.13039/501100011033, PID2022 also funded by "ERDF A way of making Europe", EU, and the predoctoral Grant of Millán Álvarez Ref. BES-2017-081320, and by the Xunta de Galicia co-founded by the European Regional Development Fund (ERDF) under the Consolidation Programme of Competitive Reference Groups (ED431C 2021/30 and ED431C 2022/16). Funding for open access charge: Universidade da Coruña/CISUG. We also acknowledge the support from the Centro Singular de Investigación de Galicia "CITIC" and the Centro Singular de Investigación en Tecnoloxías Intelixentes "CiTIUS", funded by Xunta de Galicia and the European Union (European Regional Development Fund- Galicia 2014-2020 Program), by grants ED431G 2019/01 and ED431G 2019/04. We also acknowledge the Centro de Supercomputación de Galicia (CESGA)., Xunta de Galicia; ED431C 2021/30, Xunta de Galicia; ED431C 2022/16, Xunta de Galicia; ED431G 2019/01, Xunta de Galicia; ED431G 2019/04