1
|
Mohammed A, Kora R. A Comprehensive Review on Ensemble Deep Learning: Opportunities and Challenges. JOURNAL OF KING SAUD UNIVERSITY - COMPUTER AND INFORMATION SCIENCES 2023. [DOI: 10.1016/j.jksuci.2023.01.014] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 02/04/2023]
|
2
|
Elmi J, Eftekhari M, Mehrpooya A, Ravari MR. A novel framework based on the multi-label classification for dynamic selection of classifiers. INT J MACH LEARN CYB 2023; 14:2137-2154. [PMID: 36618577 PMCID: PMC9806828 DOI: 10.1007/s13042-022-01751-z] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/25/2021] [Accepted: 12/13/2022] [Indexed: 01/04/2023]
Abstract
Multi-classifier systems (MCSs) are some kind of predictive models that classify instances by combining the output of an ensemble of classifiers given in a pool. With the aim of enhancing the performance of MCSs, dynamic selection (DS) techniques have been introduced and applied to MCSs. Dealing with each test sample classification, DS methods seek to perform the task of classifier selection so that only the most competent classifiers are selected. The principal subject regarding DS techniques is how the competence of classifiers corresponding to every new test sample classification task can be estimated. In traditional dynamic selection methods, for classifying an unknown test sample x, first, a local region of data that is similar to x is detected. Then, those classifiers that efficiently classify the data in the local region are also selected so as to perform the classification task for x. Therefore, the main effort of these methods is focused on one of the two following tasks: (i) to provide a measure for identifying a local region, or (ii) to provide a criterion for measuring the efficiency of classifiers in the local region (competence measure). This paper proposes a new version of dynamic selection techniques that does not follow the aforementioned approach. Our proposed method uses a multi-label classifier in the training phase to determine the appropriate set of classifiers directly (without applying any criterion such as a competence measure). In the generalization phase, the suggested method is employed efficiently so as to predict the appropriate set of classifiers for classifying the test sample x. It is remarkable that the suggested multi-label-based framework is the first method that uses multi-label classification concepts for dynamic classifier selection. Unlike the existing meta-learning methods for dynamic ensemble selection in the literature, our proposed method is very simple to implement and does not need meta-features. As the experimental results indicate, the suggested technique produces a good performance in terms of both classification accuracy and simplicity which is fairly comparable with that of the benchmark DS techniques. The results of conducting the Quade non-parametric statistical test corroborate the clear dominance of the proposed method over the other benchmark methods.
Collapse
Affiliation(s)
- Javad Elmi
- Department of Computer Engineering, Shahid Bahonar University of Kerman, Kerman, Iran
| | - Mahdi Eftekhari
- Department of Computer Engineering, Shahid Bahonar University of Kerman, Kerman, Iran
| | - Adel Mehrpooya
- Department of Computer Engineering, Shahid Bahonar University of Kerman, Kerman, Iran
- School of Mathematical Sciences, Faculty of Science, Queensland university of Technology, Brisbane, Australia
| | | |
Collapse
|
3
|
Zamzmi G, Rajaraman S, Hsu LY, Sachdev V, Antani S. Real-time echocardiography image analysis and quantification of cardiac indices. Med Image Anal 2022; 80:102438. [PMID: 35868819 PMCID: PMC9310146 DOI: 10.1016/j.media.2022.102438] [Citation(s) in RCA: 6] [Impact Index Per Article: 3.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/18/2021] [Revised: 01/24/2022] [Accepted: 03/28/2022] [Indexed: 11/24/2022]
Abstract
Deep learning has a huge potential to transform echocardiography in clinical practice and point of care ultrasound testing by providing real-time analysis of cardiac structure and function. Automated echocardiography analysis is benefited through use of machine learning for tasks such as image quality assessment, view classification, cardiac region segmentation, and quantification of diagnostic indices. By taking advantage of high-performing deep neural networks, we propose a novel and eicient real-time system for echocardiography analysis and quantification. Our system uses a self-supervised modality-specific representation trained using a publicly available large-scale dataset. The trained representation is used to enhance the learning of target echo tasks with relatively small datasets. We also present a novel Trilateral Attention Network (TaNet) for real-time cardiac region segmentation. The proposed network uses a module for region localization and three lightweight pathways for encoding rich low-level, textural, and high-level features. Feature embeddings from these individual pathways are then aggregated for cardiac region segmentation. This network is fine-tuned using a joint loss function and training strategy. We extensively evaluate the proposed system and its components, which are echo view retrieval, cardiac segmentation, and quantification, using four echocardiography datasets. Our experimental results show a consistent improvement in the performance of echocardiography analysis tasks with enhanced computational eiciency that charts a path toward its adoption in clinical practice. Specifically, our results show superior real-time performance in retrieving good quality echo from individual cardiac view, segmenting cardiac chambers with complex overlaps, and extracting cardiac indices that highly agree with the experts’ values. The source code of our implementation can be found in the project ‘ s GitHub page.
Collapse
Affiliation(s)
- Ghada Zamzmi
- Computational Health Research Branch, National Library of Medicine, National Institutes of Health, Bethesda, MD, USA.
| | - Sivaramakrishnan Rajaraman
- Computational Health Research Branch, National Library of Medicine, National Institutes of Health, Bethesda, MD, USA
| | - Li-Yueh Hsu
- Radiology and Imaging Sciences, Clinical Center, National Institutes of Health, Bethesda, MD, USA
| | - Vandana Sachdev
- Echocardiography Laboratory, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, USA
| | - Sameer Antani
- Computational Health Research Branch, National Library of Medicine, National Institutes of Health, Bethesda, MD, USA
| |
Collapse
|
4
|
Granule-Based-Classifier (GbC): A Lattice Computing Scheme Applied on Tree Data Structures. MATHEMATICS 2021. [DOI: 10.3390/math9222889] [Citation(s) in RCA: 5] [Impact Index Per Article: 1.7] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/16/2022]
Abstract
Social robots keep proliferating. A critical challenge remains their sensible interaction with humans, especially in real world applications. Hence, computing with real world semantics is instrumental. Recently, the Lattice Computing (LC) paradigm has been proposed with a capacity to compute with semantics represented by partial order in a mathematical lattice data domain. In the aforementioned context, this work proposes a parametric LC classifier, namely a Granule-based-Classifier (GbC), applicable in a mathematical lattice (T,⊑) of tree data structures, each of which represents a human face. A tree data structure here emerges from 68 facial landmarks (points) computed in a data preprocessing step by the OpenFace software. The proposed (tree) representation retains human anonymity during data processing. Extensive computational experiments regarding three different pattern recognition problems, namely (1) head orientation, (2) facial expressions, and (3) human face recognition, demonstrate GbC capacities, including good classification results, and a common human face representation in different pattern recognition problems, as well as data induced granular rules in (T,⊑) that allow for (a) explainable decision-making, (b) tunable generalization enabled also by formal logic/reasoning techniques, and (c) an inherent capacity for modular data fusion extensions. The potential of the proposed techniques is discussed.
Collapse
|
5
|
Wei R, Mahmood A. Optimizing Few-Shot Learning Based on Variational Autoencoders. ENTROPY (BASEL, SWITZERLAND) 2021; 23:1390. [PMID: 34828088 PMCID: PMC8618453 DOI: 10.3390/e23111390] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.7] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Subscribe] [Scholar Register] [Received: 09/21/2021] [Revised: 10/20/2021] [Accepted: 10/21/2021] [Indexed: 11/16/2022]
Abstract
Despite the importance of few-shot learning, the lack of labeled training data in the real world makes it extremely challenging for existing machine learning methods because this limited dataset does not well represent the data variance. In this research, we suggest employing a generative approach using variational autoencoders (VAEs), which can be used specifically to optimize few-shot learning tasks by generating new samples with more intra-class variations on the Labeled Faces in the Wild (LFW) dataset. The purpose of our research is to increase the size of the training dataset using various methods to improve the accuracy and robustness of the few-shot face recognition. Specifically, we employ the VAE generator to increase the size of the training dataset, including the basic and the novel sets while utilizing transfer learning as the backend. Based on extensive experimental research, we analyze various data augmentation methods to observe how each method affects the accuracy of face recognition. The face generation method based on VAEs with perceptual loss can effectively improve the recognition accuracy rate to 96.47% using both the base and the novel sets.
Collapse
Affiliation(s)
| | - Ausif Mahmood
- Department of Computer Science & Engineering, University of Bridgeport, Bridgeport, CT 06604, USA;
| |
Collapse
|
6
|
Kavitha N, Ruba Soundar K, Sathis Kumar T. An improved DFA based kernel ensemble learning machine using local feature representations for face recognition. JOURNAL OF INTELLIGENT & FUZZY SYSTEMS 2021. [DOI: 10.3233/jifs-210130] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/15/2022]
Abstract
In recent years, the Face recognition task has been an active research area in computer vision and biometrics. Many feature extraction and classification algorithms are proposed to perform face recognition. However, the former usually suffer from the wide variations in face images, while the latter usually discard the local facial features, which are proven to be important for face recognition. In this paper, a novel framework based on merging the advantages of the Key points Local Binary/Tetra Pattern (KP-LTrP) and Improved Hough Transform (IHT) with the Improved DragonFly Algorithm-Kernel Ensemble Learning Machine (IDFA-KELM) is proposed to address the face recognition problem in unconstrained conditions. Initially, the face images are collected from the publicly available dataset. Then noises in the input image are removed by performing preprocessing using Adaptive Kuwahara filter (AKF). After preprocessing, the face from the preprocessed image is detected using the Tree-Structured Part Model (TSPM) structure. Then, features, such as KP-LTrP, and IHT are extracted from the detected face and the extracted feature is reduced using the Information gain based Kernel Principal Component Analysis (IG-KPCA) algorithm. Then, finally, these reduced features are inputted to IDFA-KELM for performing FR. The outcomes of the proposed method are examined and contrasted with the other existing techniques to confirm that the proposed IDFA-KELM detects human faces efficiently from the input images.
Collapse
Affiliation(s)
- N Kavitha
- AP/CSE Department, Indra Ganesan College of Engineering, Trichy, Tamil Nadu, India
| | - K Ruba Soundar
- CSE Department, P.S.R Engineering College, Sivakasi, Tamil Nadu, India
| | - T Sathis Kumar
- AP/CSE Department, Saranathan College of Engineering, Trichy, Tamil Nadu, India
| |
Collapse
|
7
|
Elgendy M, Sik-Lanyi C, Kelemen A. A Novel Marker Detection System for People with Visual Impairment Using the Improved Tiny-YOLOv3 Model. COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE 2021; 205:106112. [PMID: 33915507 DOI: 10.1016/j.cmpb.2021.106112] [Citation(s) in RCA: 3] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Subscribe] [Scholar Register] [Received: 10/12/2020] [Accepted: 04/07/2021] [Indexed: 06/12/2023]
Abstract
BACKGROUND AND OBJECTIVE Daily activities such as shopping and navigating indoors are challenging problems for people with visual impairment. Researchers tried to find different solutions to help people with visual impairment navigate indoors and outdoors. METHODS We applied deep learning to help visually impaired people navigate indoors using markers. We propose a system to help them detect markers and navigate indoors using an improved Tiny-YOLOv3 model. A dataset was created by collecting marker images from recorded videos and augmenting them using image processing techniques such as rotation transformation, brightness, and blur processing. After training and validating this model, the performance was tested on a testing dataset and on real videos. RESULTS The contributions of this paper are: (1) We developed a navigation system to help people with visual impairment navigate indoors using markers; (2) We implemented and tested a deep learning model to detect Aruco markers in different challenging situations using Tiny-YOLOv3; (3) We implemented and compared several modified versions of the original model to improve detection accuracy. The modified Tiny-YOLOv3 model achieved an accuracy of 99.31% in challenging conditions and the original model achieved an accuracy of 96.11 %. CONCLUSION The training and testing results show that the improved Tiny-YOLOv3 models are superior to the original model.
Collapse
Affiliation(s)
- Mostafa Elgendy
- Department of Electrical Engineering and Information Systems, University of Pannonia, 8200 Veszprém, Hungary; Department of Computer Science, Faculty of Computers and Artificial Intelligence, Benha University, Benha 13511, Egypt.
| | - Cecilia Sik-Lanyi
- Department of Electrical Engineering and Information Systems, University of Pannonia, 8200 Veszprém, Hungary.
| | - Arpad Kelemen
- Department of Organizational Systems and Adult Health, University of Maryland, Baltimore, MD, 21201, USA.
| |
Collapse
|