Pasar al contenido principal

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)

WoodAD: A New Dataset and a Comparison of Deep Learning Approaches for Wood Anomaly Detection

  • del Tejo Catalá, Omar
  • Perez, Javier
  • García Sastre, Nicolás
  • Del Ser, Javier
  • Perez-Cortes, Juan-Carlos|||0000-0001-6506-090X
[EN] Anomaly detection is a crucial task in computer vision, with applications ranging from quality control to security monitoring, among many others. Recent technological advancements have enabled near-perfect solutions on benchmark datasets like MVTec, raising the need for novel datasets that pose new challenges for this modelling task. This work presents a novel Wood Anomaly Detection (WoodAD) dataset, which includes defects in wooden pieces that result in challenges for the most advanced techniques applied to other established datasets. This article evaluates such challenges posed by WoodAD with one-class and few-shot supervised learning approaches. Our experiments herein reveal that EfficientAD, a state-of-the-art method previously excelling on the MVTec dataset, outperforms all other one-class learning approaches. Nevertheless, there is room for improvement, as EfficientAD achieves a 0.535 pixel/segmentation average precision (AP) over the complete test set. UNet, a well-known pixel-level classification architecture, leveraged few-shot supervised learning to enhance the pixel AP score, achieving 0.862 pixel/segmentation AP over the entire test set. Our WoodAD dataset represents a valuable contribution to the field of anomaly detection, offering complex image textures and challenging defects. Researchers and practitioners are encouraged to leverage this dataset to push the boundaries of anomaly detection and develop more robust and effective solutions for more complex real-world applications. The WoodAD dataset has been made publicly available in Kaggle ()., This work has been developed under the framework of different funding projects. First, funding for open access charge: CRUE-Universitat Politècnica de València. Then, the Red de Excelencia Cervera en Tecnologias Habilitadoras basadas en el Dato (AI4ES) project, which has been co-funded by the Centro para el Desarrollo Tecnológico Industrial, E.P.E. (CDTI) and by the European Union (EU) through the Next Generation EU Fond, with grant number CER-20211030. Finally, J. Del Ser acknowledges funding support from the Basque Government through (ref. I1456-22, MATHMODE).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

A machine learning approach for type 2 diabetes diagnosis and prognosis using tailored heterogeneous feature subsets

  • Navarro Cerdan, José Ramón|||0000-0002-6692-5941
  • Arlandis, Joaquim|||0000-0003-2472-8883
  • Llobet Azpitarte, Rafael|||0000-0002-8278-9740
  • Perez-Cortes, Juan-Carlos|||0000-0001-6506-090X
  • Pons-Suñer, Pedro
  • Arnal, Laura
  • Lara-Hernández, Francisco
  • Moya-Valera, Celeste
  • Quiroz-Rodriguez, Maria Elena
  • Rojo-Martinez, Gemma
  • Valdés, Sergio
  • Montanya, Eduard
  • Calle-Pascual, Alfonso L.
  • Franch-Nadal, Josep
  • Delgado, Elias
  • Luis Castaño
  • Ana-Bárbara García-García
  • Chaves-Martinez, Felipe Javier
[EN] Type 2 diabetes (T2D) is becoming one of the leading health problems in Western societies, diminishing quality of life and consuming a significant share of healthcare resources. This study presents machine learning models for T2D diagnosis and prognosis, developed using heterogeneous data from a Spanish population dataset (Di@bet.es study). The models were trained exclusively on individuals classified as controls and undiagnosed diabetics, ensuring that the results are not influenced by treatment effects or behavioral changes due to disease awareness. Two data domains are considered: environmental (patient lifestyle questionnaires and measurements) and clinical (biochemical and anthropometric measurements). The preprocessing pipeline consists of four key steps: geospatial data extraction, feature engineering, missing data imputation, and quasiconstancy filtering. Two working scenarios (Environmental and Healthcare) are defined based on the features used, and applied to two targets (diagnosis and prognosis), resulting in four distinct models. The feature subsets that best predict the target have been identified based on permutation importance and sequential backward selection, reducing the number of features and, consequently, the cost of predictions. In the Environmental scenario, models achieved an AUROC of 0.86 for diagnosis and 0.82 for prognosis. The Healthcare scenario performed better, with an AUROC of 0.96 for diagnosis and 0.88 for prognosis. A partial dependence analysis of the most relevant features is also presented. An online demo page showcasing the Environmental and Healthcare T2D prognosis models is available upon request., Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature. The activities described in this document are part of the Cervera Network of Excellence in Data-Based Enabling Technologies (AI4ES) project, which is co-financed by the Center for Industrial Technological Development, E.P.E. (CDTI) and by the European Union through the Next Generation EU Fund, within the Cervera aid program for Technology Centers by the CDTI, with file number CER-20211030. This research was also funded by Generalitat Valenciana through IVACE and the European Union through FEDER funds, PANAKEIA project IMDEEA/2023/92. This work was also funded by Generalitat Valenciana through IVACE through a grant to support ITI s activity in independent research and development, research results dissemination and knowledge and technology transfer to business under project IMAMCA/2023/11. Research grants PI17/00544 and PI21/00506 from Ministerio de Ciencia e Innovación and Instituto de Salud Carlos III (ISCIII). CIBERDEM (CB07/08/0018) is part of CIBER (Consorcio Centro de Investigación Biomédica en Red) funded by ISCIII. Both research projects and CIBER are co-funded by the European Union (European Regional Development Fund (ERDF) A way to build Europe ). Funding for open access charge: CRUE-Universitat Politècnica de València.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Extended a Priori Probability (EAPP): A Data-Driven Approach for Machine Learning Binary Classification Tasks

  • ORTIZ, V.
  • Pérez-Benito, Francisco Javier
  • del Tejo Catalá, Omar
  • Salvador Igual, Ismael|||0000-0001-9269-3737
  • Llobet Azpitarte, Rafael|||0000-0002-8278-9740
  • Perez-Cortes, Juan-Carlos|||0000-0001-6506-090X
[EN] The a priori probability of a dataset is usually used as a baseline for comparing a particular algorithm's accuracy in a given binary classification task. ZeroR is the simplest algorithm for this, predicting the majority class for all examples. However, this is an extremely simple approach that has no predictive power and does not describe other dataset features that could lead to a more demanding baseline. In this paper, we present the Extended A Priori Probability (EAPP), a novel semi-supervised baseline metric for binary classification tasks that considers not only the a priori probability but also some possible bias present in the dataset as well as other features that could provide a relatively trivial separability of the target classes. The approach is based on the area under the ROC curve (AUC ROC), known to be quite insensitive to class imbalance. The procedure involves multiobjective feature extraction and a clustering stage in the input space with autoencoders and a subsequent combinatory weighted assignation from clusters to classes depending on the distance to nearest clusters for each class. Class labels are then assigned to establish the combination that maximizes AUC ROC for each number of clusters considered. To avoid overfit in the combined feature extraction and clustering method, a cross-validation scheme is performed in each case. EAPP is defined for different numbers of clusters, starting from the inverse of the minority class proportion, which is useful for a fair comparison among diversely imbalanced datasets. A high EAPP usually relates to an easy binary classification task, but it also may be due to a significant coarse-grained bias in the dataset, when the task is previously known to be difficult. This metric represents a baseline beyond the a priori probability to assess the actual capabilities of binary classification models., This work was supported in part by the Generalitat Valenciana through the Valencian Institute of Business Competitiveness (IVACE) Distributed Nominatively to Valencian Technological Innovation Centers under Project IMAMCN/2021/1, in part by the Cervera Network of Excellence Project in Data-Based Enabling Technologies (AI4ES) Co-Funded by the Centre for Industrial and Technological Development¿E. P. E. (CDTI), and in part by the European Union through the Next Generation EU Fund within the Cervera Aids Program for Technological Centers under Project CER-20211030.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

SEQENS: An ensemble method for relevant gene identification in microarray data

  • Signol, François
  • Arnal-Benedicto, Laura
  • Navarro Cerdan, José Ramón|||0000-0002-6692-5941
  • Llobet Azpitarte, Rafael|||0000-0002-8278-9740
  • Arlandis, Joaquim|||0000-0003-2472-8883
  • Perez-Cortes, Juan-Carlos|||0000-0001-6506-090X
[EN] This paper describes an ensemble feature identification algorithm called SEQENS, and measures its capability to identify the relevant variables in a case-control study using a genetic expression microarray dataset. SEQENS uses Sequential Feature Search on multiple sample splitting to select variables showing stronger relation with the target, and a variable relevance ranking is finally produced. Although designed for feature identification, SEQENS could also serve as a basis for feature selection (classifier optimisation). Cliff, a ranking evaluation metric is also presented and used to assess the feature identification algorithms when a groundtruth of relevant variables is available. To test performance, three types of synthetic groundtruths emulating fictitious diseases are generated from ten randomly chosen variables following different target pattern distributions using the E-MTAB-3732 dataset. Several sample-to-dimensionality ratios ranging from 300 to 3,000 observations and 854 to 54,675 variables are explored. SEQENS is compared with other feature selection or identification state-of-the-art methods. On average, the proposed algorithm identifies better the relevant genes and exhibits a stronger stability. The algorithm is available to the community., This work was partially funded by Generalitat Valenciana through IVACE (Valencian Institute of Business Competitiveness) distributed nominatively to Valencian technological innovation centres under project expedient IMAMCN/2021/1.It was also funded by the Cervera Network of Excellence Project in Data-based Enabling Technologies (AI4ES) , co-funded by the Centre for Industrial and Technological Development, E.P.E. (CDTI) and by the European Union through the NextGenerationEU Fund, within the Cervera Aids program for Technological Centres, with the expedient number CER-20211030.