Article
Multidisciplinary Sciences
Carmelo Militello, Leonardo Rundo, Salvatore Vitabile, Vincenzo Conti
Summary: Biometric classification is crucial in fingerprint characterization, and using Convolutional Neural Networks (CNNs) in deep learning for multi-class fingerprint classification has shown the best performance. The study aims to find the most efficient and precise model through comparisons of various CNN architectures on different fingerprint databases.
Review
Environmental Sciences
Leiyu Chen, Shaobo Li, Qiang Bai, Jing Yang, Sanlong Jiang, Yanming Miao
Summary: This article summarizes the application of deep learning in image classification, covering the development of CNNs from their predecessors to the latest network architectures, as well as a comprehensive comparison and analysis of various image classification methods.
Article
Computer Science, Software Engineering
Esther Mukoya, Richard Rimiru, Michael Kimwele, Consolata Gakii, Grace Mugambi
Summary: Biometric systems, particularly fingerprint biometrics, have been widely used for person identification and verification due to their permanence, uniqueness, ergonomics, throughput, low cost, and lifelong usability. Deep learning models have shown impressive performance in fingerprint classification tasks by reducing the number of comparisons. However, the high-level features and computational costs of these models can pose challenges in their deployment.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2023)
Article
Environmental Sciences
Tiago Marto, Alexandre Bernardino, Goncalo Cruz
Summary: This work proposes an active learning methodology for the segmentation of fire and smoke in video images. The model learns incrementally over several active learning rounds, selecting informative samples to update the training set. Using active learning in classification and segmentation tasks resulted in improved accuracy and mean intersection over union by 2%, while achieving similar results to non-active learning with fewer labeled data samples.
Article
Computer Science, Artificial Intelligence
Francesco Ponzio, Enrico Macii, Elisa Ficarra, Santa Di Cataldo
Summary: In real-world scenarios, training Convolutional Neural Networks (CNNs) with high quality images and correct labels is difficult. This affects the performance of CNNs during both training and inference. To tackle this issue, we propose a new two-module CNN called Wise2WipedNet (W2WNet), which uses Bayesian inference to identify and discard spurious images during training and provides prediction confidence during inference. Our experiments on various image classification tasks and histological image analysis demonstrate that W2WNet can effectively identify image degradation and mislabelling issues, resulting in improved classification accuracy.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Khondaker Tasrif Noor, Antonio Robles-Kelly
Summary: In this paper, the authors propose H-CapsNet, a capsule network designed for hierarchical image classification. The network effectively captures hierarchical relationships using dedicated capsules for each class hierarchy. A modified hinge loss is utilized to enforce consistency among the involved hierarchies. Additionally, a strategy for dynamically adjusting training parameters is presented to achieve better balance between the class hierarchies. Experimental results demonstrate that H-CapsNet outperforms competing hierarchical classification networks.
PATTERN RECOGNITION
(2024)
Article
Environmental Sciences
Mohammed Q. Q. Alkhatib, Mina Al-Saad, Nour Aburaed, Saeed Almansoori, Jaime Zabalza, Stephen Marshall, Hussain Al-Ahmad
Summary: A novel method called Tri-CNN and a three-branch feature fusion approach are proposed to address the issue of insufficient training samples in hyperspectral image (HSI) classification. Experimental results demonstrate that the proposed method exhibits remarkable performance in terms of overall accuracy (OA), average accuracy (AA), and Kappa metrics when compared to existing methods.
Article
Ecology
Ali Seydi Keceli, Aydin Kaya, Cagatay Catal, Bedir Tekinerdogan
Summary: The manual prediction of plant species and diseases is costly and time-consuming, and expertise may not always be available. Automated approaches, such as machine learning and deep learning, are being used to overcome these challenges. This study proposes a novel multi-task learning approach that combines plant species and disease prediction tasks using shared representations. The results show that this approach improves efficiency and learning speed.
ECOLOGICAL INFORMATICS
(2022)
Article
Ecology
Emmanuel Dufourq, Carly Batist, Ruben Foquet, Ian Durbach
Summary: Progress in deep learning, specifically in using convolutional neural networks for classification models, has been significant. This study investigates the use of transfer learning in passive acoustic monitoring, showing that it can improve F1 score up to 82% while simplifying implementation and design decisions.
ECOLOGICAL INFORMATICS
(2022)
Review
Biochemical Research Methods
Murilo Horacio Pereira da Cruz, Douglas Silva Domingues, Priscila Tiemi Maeda Saito, Alexandre Rossi Paschoal, Pedro Henrique Bugatti
Summary: The study introduces a method called TERL, which preprocesses and transforms TE sequences into two-dimensional space for classification using deep convolutional neural networks. Through six experiments, TERL shows excellent performance with high accuracy and significantly faster speed compared to other methods.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Chemistry, Analytical
Roza Dzierzak, Zbigniew Omiotek
Summary: The aim of this study was to evaluate the possibility of using deep convolutional neural networks (DCNNs) to diagnose osteoporosis based on CT images of the spine. The study found that transfer learning technique, using pre-trained DCNN models and data augmentation, achieved satisfactory results even with small datasets.
Article
Medicine, General & Internal
Adam R. Chlopowiec, Konrad Karanowski, Tomasz Skrzypczak, Mateusz Grzesiuk, Adrian B. Chlopowiec, Martin Tabakov
Summary: Multiple studies have shown satisfactory performances in the treatment of various ocular diseases. However, there is a lack of multiclass models trained on large diverse datasets and no study has addressed the class imbalance problem in one giant dataset from multiple diverse eye fundus image collections. In this study, 22 publicly available datasets were merged to create a diverse dataset for medical validity, and the state-of-the-art models ConvNext, RegNet, and ResNet were utilized to achieve the best results.
Article
Computer Science, Information Systems
Keong-Hun Choi, Jin-Woo Kim, Yao Wang, Jong-Eun Ha
Summary: In this paper, a method of improving image classification performance using CNN and transformer fusion is proposed. CNN is good at extracting information about local areas on an image, while the transformer is better at global information extraction. By combining the advantages of CNN and transformer, the proposed method achieves the best classification performance in experiments using ImageNet-1K.
Article
Computer Science, Artificial Intelligence
Tien Huu Do, Duc Minh Nguyen, Giannis Bekoulis, Adrian Munteanu, Nikos Deligiannis
Summary: Graph convolutional neural networks (GCNNs) have gained attention for their ability to handle graph-structured data, but face issues related to node transition probabilities, over-fitting, and over-smoothing. This study introduces a novel method to improve message passing based on transition probabilities and proposes DropNode regularization to address these challenges, demonstrating effectiveness in experiments.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Chemistry, Multidisciplinary
Amila Akagic, Emir Buza
Summary: The devastation caused by wildfires has increased by nearly threefold, despite a decline in the number of wildfires. This study presents a lightweight wildfire image classification model based on convolutional neural networks, which outperforms existing methods and is suitable for real-time classification of wildfire images.
APPLIED SCIENCES-BASEL
(2022)
Correction
Computer Science, Artificial Intelligence
Chongsheng Zhang, Paolo Soda, Jingjun Bi, Gaojuan Fan, George Almpanidis, Salvador Garcia, Weiping Ding
APPLIED INTELLIGENCE
(2023)
Correction
Multidisciplinary Sciences
Panagiotis Karras, Ignacio Bordeu, Joanna Pozniak, Ada Nowosad, Cecilia Pazzi, Nina Van Raemdonck, Ewout Landeloos, Yannick Van Herck, Dennis Pedri, Greet Bervoets, Samira Makhzami, Jia Hui Khoo, Benjamin Pavie, Jochen Lamote, Oskar Marin-Bejar, Michael Dewaele, Han Liang, Xingju Zhang, Yichao Hua, Jasper Wouters, Robin Browaeys, Gabriele Bergers, Yvan Saeys, Francesca Bosisio, Joost van den Oord, Diether Lambrechts, Anil K. Rustgi, Oliver Bechter, Cedric Blanpain, Benjamin D. Simons, Florian Rambow, Jean-Christophe Marine
Article
Immunology
Jana Neirinck, Annelies Emmaneel, Malicorne Buysse, Jan Philippe, Sofie Van Gassen, Yvan Saeys, Xavier Bossuyt, Stefanie De Buyser, Mirjam van der Burg, Martin Perez-Andres, Alberto Orfao, Jacques J. M. van Dongen, Bart N. Lambrecht, Tessa Kerre, Mattias Hofmans, Filomeen Haerynck, Carolien Bonroy
Summary: The study tested PIDOT in 887 consecutive patients suspicious of PID and found that it has high sensitivity and specificity for diagnosing lymphoid-PID, particularly for discriminating patients with SCID, ID, and CVID. The combination of PIDOT and serum immunoglobulin levels can guide further immunophenotypic FCM analyses for accurate PID diagnostics.
FRONTIERS IN IMMUNOLOGY
(2022)
Article
Biochemistry & Molecular Biology
Domien Vanneste, Jens Staal, Mira Haegman, Yasmine Driege, Marieke Carels, Elien Van Nuffel, Pieter De Bleser, Yvan Saeys, Rudi Beyaert, Inna S. Afonina
Summary: Prostate cancer is one of the most common cancer types in men, and its increasing prevalence worldwide is attributed to the influence of modern Western lifestyle. Through analysis of cancer databases, it has been found that overexpression of CARD14 is strongly associated with aggressive prostate cancer in human patients. Furthermore, it has been demonstrated that CARD14-induced signaling plays a role in regulating prostate cancer cell survival and gene expression.
Article
Automation & Control Systems
Pourya Shamsolmoali, Masoumeh Zareapoor, Swagatam Das, Salvador Garcia, Eric Granger, Jie Yang
Summary: Image-to-image translation is crucial in generative adversarial networks. Convolutional neural networks have limitations in capturing spatial relationships, making them unsuitable for image translation tasks. Capsule networks are proposed as a remedy, capturing hierarchical spatial relationships. In this paper, a new framework for capsule networks is presented, which can be applied to generator-discriminator architectures without computational overhead. A Gromov-Wasserstein distance is used as a loss function to guide the learned distribution. The proposed method, called generative equivariant network, is evaluated on I2I translation and image generation tasks and shows a principled connection between generative and capsule models.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Biochemistry & Molecular Biology
Dena Panovska, Pouya Nazari, Basiel Cole, Pieter-Jan Creemers, Marleen Derweduwe, Lien Solie, Sofie Van Gassen, Annelies Claeys, Tatjana Verbeke, Elizabeth F. Cohen, Michael Y. Tolstorukov, Yvan Saeys, David van der Planken, Francesca M. Bosisio, Eric Put, Sven Bamps, Paul M. Clement, Michiel Verfaillie, Raf Sciot, Keith L. Ligon, Steven De Vleeschouwer, Asier Antoranz, Frederik De Smet
Summary: The PROSPERO assay is a precise method to evaluate therapy efficacy by measuring molecular drug responses in freshly resected brain tumor samples. It provides key functional insights in cellular behavior and can complement standard clinical biomarker evaluations.
CELLULAR AND MOLECULAR LIFE SCIENCES
(2023)
Article
Engineering, Electrical & Electronic
Timo De Waele, Adnan Shahid, Daniel Peralta, Anniek Eerdekens, Margot Deruyck, Frank A. M. Tuyttens, Eli De Poorter
Summary: To track the activities and performance of horses, inertial measurement units (IMUs) combined with machine learning algorithms are commonly used. A data-efficient algorithm is proposed that only requires 3 minutes of labeled calibration data. This algorithm achieved a 95% accuracy on datasets captured with leg-mounted IMUs and neck-mounted IMU. However, when the algorithm was calibrated on multiple horses and evaluated on unfamiliar horses, there was a 15% drop in classification accuracy.
IEEE SENSORS JOURNAL
(2023)
Letter
Oncology
Margot F. van Spronsen, Sophie Horrevorts, Claudia Cali, Theresia M. Westers, Sofie Van Gassen, Yvan Saeys, Sandra J. van Vliet, Yvette van Kooyk, Arjan A. van de Loosdrecht
Article
Rheumatology
Celine Mortier, Katrien Quintelier, Ann-Sophie De Craemer, Thomas Renson, Liselotte Deroo, Emilie Dumas, Eveline Verheugen, Julie Coudenys, Tine Decruy, Zuzanna Lukasik, Sofie Van Gassen, Yvan Saeys, Anne Hoorens, Triana Lobaton, Filip van den Bosch, Tom van de Wiele, Koen Venken, Dirk Elewaut
Summary: The study found marked type 17 skewing in the inflamed gut mucosa of patients with nonradiographic axial SpA, linked to gamma delta-hi T cells and associated with intestinal inflammation and disease activity.
ARTHRITIS & RHEUMATOLOGY
(2023)
Review
Biochemistry & Molecular Biology
Corinne Urwyler-Rosselet, Giel Tanghe, Michael Devos, Paco Hulpiau, Yvan Saeys, Wim Declercq
Summary: Receptor interacting protein kinases (RIPKs) are a family of enzymes that play important roles in integrating stress signals in the skin homeostasis. By activating various signaling pathways, RIPKs regulate the activation of NF-& kappa;B and mitogen-activated protein kinases, influencing processes such as cell death, inflammation, differentiation, and Wnt signaling.
CELLULAR AND MOLECULAR LIFE SCIENCES
(2023)
Article
Automation & Control Systems
Alejandro Rosales-Perez, Salvador Garcia, Francisco Herrera
Summary: This article introduces EBCS-SVM, an evolutionary bilevel cost-sensitive SVM, for handling imbalanced classification problems. It simultaneously learns the support vectors and optimizes the SVM hyperparameters, utilizing evolutionary algorithm and sequential minimal optimization.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Letter
Biotechnology & Applied Microbiology
Anthony Sonrel, Almut Luetge, Charlotte Soneson, Izaskun Mallona, Pierre-Luc Germain, Sergey Knyazev, Jeroen Gilis, Reto Gerber, Ruth Seurinck, Dominique Paul, Emanuel Sonder, Helena L. L. Crowell, Imran Fanaswala, Ahmad Al-Ajami, Elyas Heidari, Stephan Schmeing, Stefan Milosavljevic, Yvan Saeys, Serghei Mangul, Mark D. Robinson
Summary: Computational methods are crucial in modern molecular biology. Benchmarking is critical in dissecting analysis pipelines, evaluating performance, and guiding users on tool selection. It also plays a key role in community building and advancing methods.
Review
Biochemistry & Molecular Biology
Abhishek Subramanian, Pooya Zakeri, Mira Mousa, Halima Alnaqbi, Fatima Yousif Alshamsi, Leo Bettoni, Ernesto Damiani, Habiba Alsafar, Yvan Saeys, Peter Carmeliet
Summary: This review aims to introduce the potential of computational methods for angiogenic target discovery to vascular biologists lacking expertise in these methods. It provides a comprehensive survey of computational approaches that can be used for prioritizing candidate genes, predicting associated mechanisms, and identifying their specificity to endothelial cell subtypes.
COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Arne Gevaert, Jonathan Peck, Yvan Saeys
Summary: Deep Reinforcement Learning with deep neural networks is highly effective but lacks interpretability. Neuro-fuzzy controllers offer interpretable alternatives but often require numerous rules. This work presents an algorithm to distill policies from deep Q-networks into compact neuro-fuzzy controllers, combining the flexibility of DRL and interpretability of compact rule bases.
2022 IEEE INTERNATIONAL CONFERENCE ON FUZZY SYSTEMS (FUZZ-IEEE)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Robin Vandaele, Bo Kang, Tijl De Bie, Yvan Saeys
Summary: This paper investigates the effects of noise on distances and neighborhood relations in high-dimensional data. By deriving and empirically verifying under certain conditions, the researchers find that neighborhood relations affected by noise can still be truthful even when distance concentration occurs.
INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND STATISTICS, VOL 151
(2022)