Pasar al contenido principal

APRENDIZAJE PROFUNDO EN VOZ PARA APLICACIONES FORENSES Y DE SEGURIDAD

RTI2018-098091-B-I00

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal de I+D+i Orientado a los Retos de la Sociedad
Convocatoria Proyectos I+D+i Retos de Investigación
Año convocatoria 2018
Plan Estatal 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSIDAD AUTONOMA DE MADRID

Resultados relacionados

Resultados totales (Incluyendo duplicados): 4
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Calibration of deep probabilistic models with decoupled bayesian neural networks

  • Maroñas-Molano, Juan
  • Ramos, Daniel
  • Paredes Palacios, Roberto
[EN] Deep Neural Networks (DNNs) have achieved state-of-the-art accuracy performance in many tasks. However, recent works have pointed out that the outputs provided by these models are not well -calibrated, seriously limiting their use in critical decision scenarios. In this work, we propose to use a decoupled Bayesian stage, implemented with a Bayesian Neural Network (BNN), to map the uncalibrated probabilities provided by a DNN to calibrated ones, consistently improving calibration. Our results evi-dence that incorporating uncertainty provides more reliable probabilistic models, a critical condition for achieving good calibration. We report a generous collection of experimental results using high -accuracy DNNs in standardized image classification benchmarks, showing the good performance, flexibil-ity and robust behaviour of our approach with respect to several state-of-the-art calibration methods. Code for reproducibility is provided. (C) 2020 Elsevier B.V. All rights reserved., We gratefully acknowledge the feedback provided by Emilio Granell and Enrique Vidal on an earlier manuscript. The authors thank the EU-FEDER Comunitat Valenciana 2014-2020 grant IDIFE-DER/2018/025. We also acknowledge the support of NVIDIA by providing two GPU Titan XP from their grant program and Mario Parreno for providing the logits of the ADIENCE and VGGFACE2 models. Juan Maronas is supported by grant FPI-UPV. Daniel Ramos is supported by the Spanish Ministry of Science, Innovation and Universities via grant RTI2018-098091-B-I00.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

A deep analysis on high resolution dermoscopic image classification

  • Pollastri, Federico
  • Parreño Lara, Mario
  • Maroñas-Molano, Juan
  • Bolelli, Federico
  • Paredes Palacios, Roberto
  • Ramos, Daniel
  • Grana, Costantino
[EN] Convolutional neural networks (CNNs) have been broadly employed in dermoscopic image analysis, mainly as a result of the large amount of data gathered by the International Skin Imaging Collaboration (ISIC). As in many other medical imaging domains, state-of-the-art methods take advantage of architectures developed for other tasks, frequently assuming full transferability between enormous sets of natural images (e.g. ImageNet) and dermoscopic images, which is not always the case. A comprehensive analysis on the effectiveness of state-of-the-art deep learning techniques when applied to dermoscopic image analysis is provided. To achieve this goal, the authors consider several CNNs architectures and analyse how their performance is affected by the size of the network, image resolution, data augmentation process, amount of available data, and model calibration. Moreover, taking advantage of the analysis performed, a novel ensemble method to further increase the classification accuracy is designed. The proposed solution achieved the third best result in the 2019 official ISIC challenge, with an accuracy of 0.593., Juan Maroñas is supported by grant FPI-UPV, grant agreement No 825,111 DeepHealth Project, and by the Spanish National Ministry of Education through grant RTI2018-098091-B-I00. The research leading to these results has received funding from the European Union through Programa Operativo del Fondo Europeo de Desarrollo Regional (FEDER) from Comunitat Valencia (2014-2020) under project Sistemas de frabricación inteligentes para la indústria 4.0 (grant agreement IDIFEDER/2018/025).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Improving calibration of forensic glass comparisons by considering uncertainty in feature-based elemental data

  • Ramos, Daniel
  • Maroñas-Molano, Juan
  • Almirall, Jose
[EN] The computation of likelihood ratios (LR) to measure the weight of forensic glass evidence with LA-ICP-MS data directly in the feature space without computing any kind of score as an intermediate step is a complex problem. A probabilistic two-level modeling of the within-source and between-source variability of the glass samples is needed in order to compare the elemental profiles measured from glass recovered from a suspect or a crime scene and compared to glass samples of a known source of origin. Calibration of the likelihood ratios generated using previously reported models is essential to the realistic reporting of the value of the glass evidence comparisons. We propose models that outperform previously proposed feature-based LR models, in particular by improving the calibration of the computed LRs. We assume that the within-source variability is heavy-tailed, in order to incorporate uncertainty when the available data is scarce, as it typically happens in forensic glass comparison. Moreover, we address the complexity of the between-source variability by the use of probabilistic machine learning algorithms, namely a variational autoencoder and a warped Gaussian mixture. Our results show that the overall performance of the likelihood ratios generated by our model is superior to classical approaches, and that this improvement is due to a dramatic improvement in the calibration despite some loss in discriminating power. Moreover, the robustness of the calibration of our proposal is remarkable, This work has been partially developed during a research stay of D. R. at the Machine Learning Group of the Department of Engineering, University of Cambridge, UK; funded by the Spanish Ministerio de Educacion, Cultura y Deporte under the program for mobility of professors and researchers in higher education and research centers. We strongly thank Dr. Jose Miguel Hernandez-Lobato for all the ideas and discussions about the algorithmic developments in this article, and for hosting the aforementioned research stay. We also thank Peter Weis and Sonja Menges from the German Federal Police (Bundeskriminalamt) for their support, their revision of the manuscript, and for providing the BKA database. This development of the Florida International University (FIU) background database of vehicle glass samples was supported by Award No. 2018-DU-BX-0194 from the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice to Florida International University. The opinions, findings, and conclusions or recommendations expressed in this manuscript are those of the authors and do not necessarily reflect those of the National Institute of Justice or the U.S. Department of Justice. D. R and J. M. were supported by the Spanish Ministerio de Educaci ~on, Cultura y Deporte through grant RTI2018-098091-B-I00. J.M is also supported by grant FPI-UPV associated to the DeepHealthProject, grant agreement No 825111
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Search on speech from spoken queries: the multi-domain International ALBAYZIN 2018 query-by-example spoken term detection evaluation

  • Tejedor Noguerales, Javier
  • Toledano, Doroteo T.
  • López Otero, Paula
  • Docío Fernández, Laura
  • Peñagarikano, Mikel
  • Rodríguez Fuentes, Luis Javier
  • Moreno Sandoval, Antonio
The huge amount of information stored in audio and video repositories makes search on speech (SoS) a priority areanowadays. Within SoS, Query-by-Example Spoken Term Detection (QbE STD) aims to retrieve data from a speechrepository given a spoken query. Research on this area is continuously fostered with the organization of QbE STDevaluations. This paper presents a multi-domain internationally open evaluation for QbE STD in Spanish. Theevaluation aims at retrieving the speech files that contain the queries, providing their start and end times, and a scorethat reflects the confidence given to the detection. Three different Spanish speech databases that encompassdifferent domains have been employed in the evaluation: MAVIR database, which comprises a set of talks fromworkshops; RTVE database, which includes broadcast television (TV) shows; and COREMAH database, which contains2-people spontaneous speech conversations about different topics. The evaluation has been designed carefully sothat several analyses of the main results can be carried out. We present the evaluation itself, the three databases, theevaluation metrics, the systems submitted to the evaluation, the results, and the detailed post-evaluation analysesbased on some query properties (within-vocabulary/out-of-vocabulary queries, single-word/multi-word queries, andnative/foreign queries). Fusion results of the primary systems submitted to the evaluation are also presented. Threedifferent teams took part in the evaluation, and ten different systems were submitted. The results suggest that theQbE STD task is still in progress, and the performance of these systems is highly sensitive to changes in the datadomain. Nevertheless, QbE STD strategies are able to outperform text-based STD in unseen data domains., Xunta de Galicia | Ref. ED431G/01, Xunta de Galicia | Ref. ED431G/04, Ministerio de Economía y Competitividad | Ref. TEC2015-68172-C2-1-P, Agencia Estatal de Investigación | Ref. RTI2018-098091-B-I00