Publications

Using machine learning-based systems to help predict disengagement from the legal proceedings by women victims of intimate partner violence in Spain

Abstract Intimate partner violence (IPV) is an actual social issue which poses a challenge in terms of prevention, legal action, and reporting the abuse once it has occurred. In this last case, out of the total of female victims that fill a complaint against their abuser and initiate the legal proceedings, a significant number withdraw from it for different reasons. In this field, it is interesting to detect the victims that disengage from the legal process so that professionals can intervene before it occurs. Previous studies have applied statistical models to use input variables and make a prediction of withdrawal. However, it has not been found in the literature any study that uses machine learning models to predict disengagement from the legal proceedings in IPV cases, which can be a better option to detect these events with a higher precision. Therefore, in this work, a novel application of machine learning techniques to predict the decision of victims of IPV to withdraw from prosecution is studied. For this purpose, three different ML algorithms have been optimized and tested with the original dataset to prove the great performance of ML models against non-linear input data. Once the best models have been obtained, explainable artificial intelligence (xAI) techniques have been applied to search for the most informative input features and reduce the original dataset to the most important variables. Finally, these results have been compared to those obtained in the previous work that used statistical techniques, and the set of most informative parameters has been combined with the variables of the previous study, showing that ML-based models have a better predictive accuracy in all cases and that by adding one new variable to the previous work’ subset, the accuracy to detect withdrawal improves by 7.5%.

Sinergias en la investigación en STEM

Special Issue “Fighting COVID-19: Emerging Techniques and Aid Systems for Prevention, Forecasting and Diagnosis”

Since its emergence at the end of 2019, the pandemic caused by the COVID-19 virus has led to multiple changes in health protocols around the world [...]

Using machine learning-based systems to help predict disengagement from the legal proceedings by women victims of intimate partner violence in Spain

Intimate partner violence against women (IPVW) is a pressing social issue which poses a challenge in terms of prevention, legal action, and reporting the abuse once it has occurred. However, a significant number of female victims who file a complaint against their abuser and initiate legal proceedings, subsequently, withdraw charges for different reasons. Research in this field has been focusing on identifying the factors underlying women victims' decision to disengage from the legal process to enable intervention before this occurs. Previous studies have applied statistical models to use input variables and make a prediction of withdrawal. However, none have used machine learning models to predict disengagement from legal proceedings in IPVW cases. This could represent a more accurate way of detecting these events. This study applied machine learning (ML) techniques to predict the decision of IPVW victims to withdraw from prosecution. Three different ML algorithms were optimized and tested with the original dataset to assess the performance of ML models against non-linear input data. Once the best models had been obtained, explainable artificial intelligence (xAI) techniques were applied to search for the most informative input features and reduce the original dataset to the most important variables. Finally, these results were compared to those obtained in the previous work that used statistical techniques, and the set of most informative parameters was combined with the variables of the previous study, showing that ML-based models had a better predictive accuracy in all cases and that by adding one new variable to the previous work's predictive model, the accuracy to detect withdrawal improved by 7.5%.

On the AER Stereo-Vision Processing: A Spike Approach to Epipolar Matching

Real-time motor rotation frequency detection with event-based visual and spike-based auditory AER sensory integration for FPGA

Multisensory integration is commonly used in various robotic areas to collect more environmental information using different and complementary types of sensors. Neuromorphic engineers mimics biological systems behavior to improve systems performance in solving engineering problems with low power consumption. This work presents a neuromorphic sensory integration scenario for measuring the rotation frequency of a motor using an AER DVS128 retina chip (Dynamic Vision Sensor) and a stereo auditory system on a FPGA completely event-based. Both of them transmit information with Address-Event-Representation (AER). This integration system uses a new AER monitor hardware interface, based on a Spartan-6 FPGA that allows two operational modes: real-time (up to 5 Mevps through USB2.0) and data logger mode (up to 20Mevps for 33.5Mev stored in onboard DDR RAM). The sensory integration allows reducing prediction error of the rotation speed of the motor since audio processing offers a concrete range of rpm, while DVS can be much more accurate.

Internet of Things - New Trends, Challenges and Hurdles

The Internet of Things (IoT) has emerged as a popular area of research and has piqued the interest of academics and scholars worldwide. As such, many works have been done on IoT in a variety of application areas. Written by leading experts in the field, this book serves as a showcase of the breadth of IoT research conducted in recent years for people who, while not experts in the field, do have prior knowledge of the IoT. The book also serves curious, non-technical readers, enabling them to understand necessary concepts and terminologies associated with the IoT.

A study on the use of Edge TPUs for eye fundus image segmentation

Classification of skin blemishes with cell phone images using deep learning techniques

Skin blemishes can be caused by multiple events or diseases and, in some cases, it is difficult to distinguish where they come from. Therefore, there may be cases with a dangerous origin that go unnoticed or the opposite case (which can lead to overcrowding of health services). To avoid this, the use of artificial intelligence-based classifiers using images taken with mobile devices is proposed; this would help in the initial screening process and provide some information to the patient prior to their final diagnosis. To this end, this work proposes an optimization mechanism based on two phases in which a global search for the best classifiers (from among more than 150 combinations) is carried out, and, in the second phase, the best candidates are subjected to a phase of evaluation of the robustness of the system by applying the cross-validation technique. The results obtained reach 99.95% accuracy for the best case and 99.75% AUC. Comparing the developed classifier with previous works, an improvement in terms of classification rate is appreciated, as well as in the reduction of the classifier complexity, which allows our classifier to be integrated in a specific purpose system with few computational resources.

ECGVisualizer: A 12-Lead ECG Pre-Processing, Feature Extraction and Report Generation Tool

The tools provided by electrocardiograph manufacturers are intended for the general population and only show basic information. In a more specific context, like elite sports, other signal’s features need to be analysed and therefore different processing is required. This tool allows importing the XML files generated by the General Electric CardioSoft 12SL electrocardiograph, perform a manual or automatic signal filtering process and PQRST peak detection, and finally generate a customisable report as a CSV file containing the features obtained after signal analysis. This pre-processed information can be used in artificial intelligence systems to develop diagnostic support systems.

Performance evaluation over HW/SW co-design SoC memory transfers for a CNN accelerator

Many FPGAs vendors have recently included embedded processors in their devices, like Xilinx with ARM-Cortex A cores, together with programmable logic cells. These devices are known as Programmable System on Chip (PSoC). Their ARM cores (embedded in the processing system or PS) communicates with the programmable logic cells (PL) using ARM-standard AXI buses. In this paper we analyses the performance of exhaustive data transfers between PS and PL for a Xilinx Zynq FPGA in a co-design real scenario for Convolutional Neural Networks (CNN) accelerator, which processes, in dedicated hardware, a stream of visual information from a neuromorphic visual sensor for classification. In the PS side, a Linux operating system is running, which recollects visual events from the neuromorphic sensor into a normalized frame, and then it transfers these frames to the accelerator of multi-layered CNNs, and read results, using an AXI-DMA bus in a per-layer way. As these kind of accelerators try to process information as quick as possible, data bandwidth becomes critical and maintaining a good balanced data throughput rate requires some considerations. We present and evaluate several data partitioning techniques to improve the balance between RX and TX transfer and two different ways of transfers management: through a polling routine at the userlevel of the OS, and through a dedicated interrupt-based kernellevel driver. We demonstrate that for longer enough packets, the kernel-level driver solution gets better timing in computing a CNN classification example. Main advantage of using kernel-level driver is to have safer solutions and to have tasks scheduling in the OS to manage other important processes for our application, like frames collection from sensors and their normalization.

Introductory Chapter: Understanding Human Gait

Práctica de desarrollo de interfaces hardware/software para la monitorización del estado de un PC

Este articulo presenta una practica laboratorio impartida mediante una metodologia de aprendizaje basado en proyectos (ABP) [1] para dotar de la capacidad de disenar y desarrollar un monitor del estado de un ordenador, integrado en un sistema empotrado que se comunica con una aplicacion de escritorio, a nuestros alumnos de la asignatura de Diseno de Microcontroladores (DM) en el contexto del Master en Ingenieria de Computadores y Redes. Esta practica abarca la comunicacion Hardware/ Software entre un microcontrolador con un nucleo Cortex-M4 y una aplicacion software escrita en lenguaje C# usando el entorno Visual Studio Community 2015 a traves de puertos series virtuales (VCP). Esta practica esta enfocada como un proyecto que los alumnos han de ir realizando desde cero, avanzando mediante la consecucion de hitos, hasta conseguir obtener un sistema final. El sistema a desarrollar se divide en dos partes, por un lado tenemos un PC con un sistema operativo de la familia Windows, en el que se construye una aplicacion visual mediante Windows Forms, la cual obtiene informacion del sistema de forma periodica y la envia al microcontrolador mediante comandos usando el puerto serie (USB o comunicacion Bluetooth). Por otro lado tenemos un microcontrolador de la familia STM32 que dispone de un display LCD ejecutando una plataforma completamente libre, .NET Micro Framework, la cual recibe a traves del puerto serie la informacion obtenida gracias a la aplicacion software del PC y la muestra en la pantalla, obteniendo asi una herramienta de monitorizacion del PC sin tener que estar conectado fisicamente a este. El desarrollo de este tipo de proyectos se anade la dificultad de la necesidad del uso de diferentes herramientas para el desarrollo del firmware y del software en paralelo, de manera incremental, y enfocadas para ambitos de uso muy distintos. Esta practica ha tenido una gran acogida por parte de los alumnos, ya que les ha servido de ejemplo del desarrollo de firmware para un microcontrolador usando la plataforma .NET MF y de su comunicacion con el PC por medio de una aplicacion visual.

pyNAVIS: An open-source cross-platform software for spike-based neuromorphic audio information processing

Bio-Inspired Stereo Vision Calibration for Dynamic Vision Sensors

Many advances have been made in the field of computer vision. Several recent research trends have focused on mimicking human vision by using a stereo vision system. In multi-camera systems, a calibration process is usually implemented to improve the results accuracy. However, these systems generate a large amount of data to be processed; therefore, a powerful computer is required and, in many cases, this cannot be done in real time. Neuromorphic Engineering attempts to create bio-inspired systems that mimic the information processing that takes place in the human brain. This information is encoded using pulses (or spikes) and the generated systems are much simpler (in computational operations and resources), which allows them to perform similar tasks with much lower power consumption, thus these processes can be developed over specialized hardware with real-time processing. In this work, a bio-inspired stereo-vision system is presented, where a calibration mechanism for this system is implemented and evaluated using several tests. The result is a novel calibration technique for a neuromorphic stereo vision system, implemented over specialized hardware (FPGA - Field-Programmable Gate Array), which allows obtaining reduced latencies on hardware implementation for stand-alone systems, and working in real time.

NAVIS: Neuromorphic Auditory VISualizer Tool

IoT garment for remote elderly care network

A Deep-Learning Based Posture Detection System for Preventing Telework-Related Musculoskeletal Disorders

The change from face-to-face work to teleworking caused by the pandemic has induced multiple workers to spend more time than usual in front of a computer; in addition, the sudden installation of workstations in homes means that not all of them meet the necessary characteristics for the worker to be able to position himself/herself comfortably with the correct posture in front of their computer. Furthermore, from the point of view of the medical personnel in charge of occupational risk prevention, an automated tool able to quantify the degree of incorrectness of a postural habit in a worker is needed. For this purpose, in this work, a system based on the postural detection of the worker is designed, implemented and tested, using a specialized hardware system that processes video in real time through convolutional neural networks. This system is capable of detecting the posture of the neck, shoulders and arms, providing recommendations to the worker in order to prevent possible health problems, due to poor posture. The results of the proposed system show that this video processing can be carried out in real time (up to 25 processed frames/sec) with a low power consumption (less than 10 watts) using specialized hardware, obtaining an accuracy of over 80% in terms of the pattern detected.

Spikes Monitors for FPGAs, an Experimental Comparative Study