Examinando por Autor "Concha, David"

Mostrando 1 - 7 de 7

Logic Neural Networks for Efficient FPGA Implementation
(Institute of Electrical and Electronics Engineers, 2024-11-07) Ramírez, Iván; Garcia-Espinosa, Francisco J.; Concha, David; Aranda, Luis Alberto
Logic Neural Networks (LNNs) represent a new paradigm for implementing neural networks in hardware devices such as Field-Programmable Gate Arrays (FPGAs). These network architectures exhibit unique attributes that can leverage the inherent parallelism of FPGAs, enabling the development of networks characterized by low power consumption and fast inference capabilities. Despite their potential advantages, the relative novelty of LNNs poses a challenge, as there are currently no established guidelines for defining their architectures. In this paper, we present a comprehensive study of LNNs, aiming to address the existing gap in understanding and guide decision-making during the design phase. Through systematic experimentation and analysis, we explore various aspects of logic networks, including their impact on inference time, power consumption, and overall simplicity. The findings derived from these experiments provide valuable insights for the creation of improved networks, thereby paving the way for further advancements in this field
Monitoring Volcanic and Tectonic Sandbox Analogue Models Using the Kinect v2 Sensor
(Wiley, 2022-05-06) Rincón, Marta; Márquez, Álvaro; Herrera, Raquel; Galland, Olivier; Sánchez Oro, Jesús; Concha, David; Montemayor, Antonio S
The measurement of surface deformation in analogue models of volcanic and tectonic processes is an area in continuous development. Properly quantifying topography change in analogue models is key for a useful comparison between experiment results and nature. The aim of this work is to evaluate the capabilities of the simple and cheap MicrosoftR Kinect v2 sensor for monitoring analogue models made of granular materials. MicrosoftR Kinect v2 is a video-gaming RedGreenBlue-Depth device combining an optical camera and an infrared distance measurement sensor. The precision of the device for model topography measurements has been quantified using 64 experiments, with variable granular materials materials and distance to the model. Additionally, we tested the capabilities of averaging several distance images to increase the precision. We have developed a specific software to facilitate the acquisition and processing of the Kinect v2 data in experiment monitoring. Our results show that measurement precision is material dependent: with clear-colored and finegrained materials, a precision ∼1.0 mm for digital elevation models with a 1.6 mm pixel size can be obtained. We show that by averaging ≥5 consecutive images the distance precision can reach values as low as 0.5 mm. To show the Kinect v2 capabilities, we present monitoring results from case study experiments modelling tectonics and volcano deformation. The Kinect v2 achieves lower spatial resolutions and precision than moresophisticated techniques such as photogrammetry. However, Kinect v2 provides a cheap, straightforward and powerful tool for monitoring the topography changes in sandbox analogue models.
Monitoring Volcanic and Tectonic Sandbox Analogue Models Using the Kinect v2 sensor
(Earth and Space Science, 2022-06-01) Rincón, Marta; Marquez, A; Herrera, R; Galland, O; Sanchez-Oro, Jesús; Concha, David; Sanz, Antonio
The measurement of surface deformation in analogue models of volcanic and tectonic processes is an area in continuous development. Properly quantifying topography change in analogue models is key for a useful comparison between experiment results and nature. The aim of this work is to evaluate the capabilities of the simple and cheap Microsoft® Kinect v2 sensor for monitoring analogue models made of granular materials. Microsoft® Kinect v2 is a video-gaming RedGreenBlue-Depth device combining an optical camera and an infrared distance measurement sensor. The precision of the device for model topography measurements has been quantified using 64 experiments, with variable granular materials materials and distance to the model. Additionally, we tested the capabilities of averaging several distance images to increase the precision. We have developed a specific software to facilitate the acquisition and processing of the Kinect v2 data in experiment monitoring. Our results show that measurement precision is material dependent: with clear-colored and fine-grained materials, a precision ∼1.0 mm for digital elevation models with a 1.6 mm pixel size can be obtained. We show that by averaging ≥5 consecutive images the distance precision can reach values as low as 0.5 mm. To show the Kinect v2 capabilities, we present monitoring results from case study experiments modeling tectonics and volcano deformation. The Kinect v2 achieves lower spatial resolutions and precision than more sophisticated techniques such as photogrammetry. However, Kinect v2 provides a cheap, straightforward and powerful tool for monitoring the topography changes in sandbox analogue models.
NATURAL LANGUAGE PROCESSING APPLIED FOR TEACHING EVALUATION
(ICERI2022 Proceedings, 2022-11-09) Montalvo, Soto; Rodríguez, Miguel Ángel; Cabido, Raúl; Concha, David
In the context of higher education, the improvement of teaching quality is a constant challenge. Student ratings of teaching are often considered fundamental for measuring the quality of teaching, educational development, and the enhancement of student learning. In the university context, the institution develops methods for measuring teaching quality and teacher effectiveness defining the process of collecting data and making judgments. However, for decades, there has been a debate about whether student ratings of teaching be trusted. Different studies have shown that several factors influence student ratings as teacher body language, timing, or student fatigue, among others. On the other hand, student ratings and student learning are unrelated. The surveys must become the starting point for critical discourse on effective teaching practices. Teaching thousands of students without engaging them in discussions about their learning experiences is not rational. In this sense, we propose an easy way to collect the students' opinions for teaching evaluation. We developed an anonymous online form, with several questions about different general aspects, shared by all subjects, such as theory, practice, evaluation, etc., and a final question to get the general evaluation of the subject. Each question must be answered with a numerical value between 1 and 5; the student can introduce text to explain or justify the numerical value. The proposed form is sufficiently general to be applied to any subject. Still, at the same time, with the part of text answers that accompanies each question, plus the general open question, the form allows the singularities of each subject to be considered. The objective is to know the student's opinion about some aspects of subjects through a general question. Concretely, the system aims at understanding all that a student wants to outline about the subject, good or bad, and the possible correlation or not between numerical and text responses. The proposed system automatically identifies the sentiment of text answers through Natural Language Processing techniques. We collected student feedback for four different subjects from 2-degree programs. First, we could see that approximately 40% of the enrolled students responded to the form. In the case of the university's institutional surveys, they are usually done by more students because they are mandatory, but this means that the answers they give could be not sincere. On the other hand, we have taken as a gold standard the numerical response of the students to the open-ended question about the general assessment, and we have contrasted it with the results of the automatic sentiment analysis. A significant correlation was found for negative evaluations, whereas the less was for positive ones. The students indicated that they liked the subject according to the numerical evaluation but did not express the same in the text response. Consequently, it can indicate two things: either that the sentiment analyzer needs to be adjusted or that the students are inconsistent in their answers. Further work will be done on the sentiment analysis system, reviewing the vocabulary used by the students, extending the study to all the questions on the form, and analyzing whether the students' opinions on the overall rating are different from those stated on the form itself.
Optimization and parallelization of the discrete ordinate method for radiation transport simulation in OpenFOAM: Hierarchical combination of shared and distributed memory approaches
(Open Research Europe, 2021-01-01) Marugán, J; Moreno-Sansegundo, José Ángel; Casado, Cintia; Concha, David; Sanz, Antonio
This paper describes the reduction in memory and computational time for the simulation of complex radiation transport problems with the discrete ordinate method (DOM) model in the open-source computational fluid dynamics platform OpenFOAM. Finite volume models require storage of vector variables in each spatial cell; DOM introduces two additional discretizations, in direction and wavelength, making memory a limiting factor. Using specific classes for radiation sources data, changing the store of fluxes and other minor changes allowed a reduction of 75% in memory requirements. Besides, a hierarchical parallelization was developed, where each node of the standard parallelization uses several computing threads, allowing higher speed and scalability of the problem. This architecture, combined with optimization of some parts of the code, allowed a global speedup of x15. This relevant reduction in time and memory of radiation transport opens a new horizon of applications previously unaffordable.
Performance evaluation of a 3D multi-view-based particle filter for visual object tracking using GPUs and multicore CPUs
(Journal of Real-Time Image Processing, 2018-08-01) Concha, David; Cabido, Raúl; Pantrigo, Juan José; Sanz, Antonio
This paper presents a deep and extensive performance analysis of the particle filter (PF) algorithm for a very compute intensive 3D multi-view visual tracking problem. We compare different implementations and parameter settings of the PF algorithm in a CPU platform taking advantage of the multithreading capabilities of the modern processors and a graphics processing unit (GPU) platform using NVIDIA CUDA computing environment as developing framework. We extend our experimental study to each individual stage of the PF algorithm, and evaluate the quality versus performance trade-off among different ways to design these stages. We have observed that the GPU platform performs better than the multithreaded CPU platform when handling a large number of particles, but we also demonstrate that hybrid CPU/GPU implementations can run almost as fast as only GPU solutions.
Visual classification of dumpsters with capsule networks
(ACS, 2022) Garcia-Espinosa, Francisco J.; Concha, David; Pantrigo, Juan J.; Cuesta-Infante, Alfredo
Garbage management is an essential task in the everyday life of a city. In many countries, dumpsters are owned and deployed by the public administration. An updated what-and-where list is in the core of the decision making process when it comes to remove or renew them. Moreover, it may give extra information to other analytics in a smart city context. In this paper, we present a capsule network-based architecture to automate the visual classification of dumpsters. We propose different network hyperparameter settings, such as reducing convolutional kernel size and increasing convolution layers. We also try several data augmentation strategies, as crop and flip image transformations. We succeed in reducing the number of network parameters by 85% with respect to the best previous method, thus decreasing the required training time and making the whole process suitable for low cost and embedded software architectures. In addition, the paper provides an extensive experimental analysis including an ablation study that illustrates the contribution of each component in the proposed method. Our proposal is compared with the state-of-the-art method, which is based on a Google Inception V3 architecture pretrained with Imagenet. Experimental results show that our proposal achieves a 95.35% accuracy, 2.35% over the previous best method.