The effect of choosing optimizer algorithms to improve computer vision tasks: a comparative study

Multimedia Tools and Applications, Journal Year: 2022, Volume and Issue: 82(11), P. 16591 - 16633

Published: Sept. 28, 2022

Optimization algorithms are used to improve model accuracy. The optimization process undergoes multiple cycles until convergence. A variety of strategies have been developed overcome the obstacles involved in learning process. Some these considered this study learn more about their complexities. It is crucial analyse and summarise techniques methodically from a machine standpoint since can provide direction for future work both optimization. approaches under consideration include Stochastic Gradient Descent (SGD), with Momentum, Rung Kutta, Adaptive Learning Rate, Root Mean Square Propagation, Moment Estimation, Deep Ensembles, Feedback Alignment, Direct Adfactor, AMSGrad, Gravity. prove ability each optimizer applied models. Firstly, tests on skin cancer using ISIC standard dataset detection were three common optimizers (Adaptive Moment, SGD, Propagation) explore effect images. optimal training results analysis indicate that performance values enhanced Adam optimizer, which achieved 97.30% second COVIDx CT images, 99.07% accuracy based optimizer. result indicated utilisation such as SGD improved training, testing, validation stages.

Language: Английский

---

Optimizing classification of diseases through language model analysis of symptoms

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Jan. 17, 2024

Abstract This paper investigated the use of language models and deep learning techniques for automating disease prediction from symptoms. Specifically, we explored two Medical Concept Normalization—Bidirectional Encoder Representations Transformers (MCN-BERT) a Bidirectional Long Short-Term Memory (BiLSTM) model, each optimized with different hyperparameter optimization method, to predict diseases symptom descriptions. In this paper, utilized distinct dataset called Dataset-1, Dataset-2. Dataset-1 consists 1,200 data points, point representing unique combination labels While, Dataset-2 is designed identify Adverse Drug Reactions (ADRs) Twitter data, comprising 23,516 rows categorized as ADR (1) or Non-ADR (0) tweets. The results indicate that MCN-BERT model AdamP achieved 99.58% accuracy 96.15% AdamW performed well 98.33% 95.15% Dataset-2, while BiLSTM Hyperopt 97.08% 94.15% Our findings suggest have promise supporting earlier detection more prompt treatment diseases, expanding remote diagnostic capabilities. demonstrated robust performance in accurately predicting symptoms, indicating potential further related research.

Language: Английский

---

An optimized capsule neural networks for tomato leaf disease classification

EURASIP Journal on Image and Video Processing, Journal Year: 2024, Volume and Issue: 2024(1)

Published: Jan. 8, 2024

Abstract Plant diseases have a significant impact on leaves, with each disease exhibiting specific spots characterized by unique colors and locations. Therefore, it is crucial to develop method for detecting these based spot shape, color, location within the leaves. While Convolutional Neural Networks (CNNs) been widely used in deep learning applications, they suffer from limitations capturing relative spatial orientation relationships. This paper presents computer vision methodology that utilizes an optimized capsule neural network (CapsNet) detect classify ten tomato leaf using standard dataset images. To mitigate overfitting, data augmentation, preprocessing techniques were employed during training phase. CapsNet was chosen over CNNs due its superior ability capture positioning image. The proposed approach achieved accuracy of 96.39% minimal loss, relying 0.00001 Adam optimizer. By comparing results existing state-of-the-art approaches, study demonstrates effectiveness accurately identifying classifying location. findings highlight potential as alternative improving detection classification plant pathology research.

Language: Английский

---

Chest X-ray pneumothorax segmentation using U-Net with EfficientNet and ResNet architectures

PeerJ Computer Science, Journal Year: 2021, Volume and Issue: 7, P. e607 - e607

Published: June 29, 2021

Medical imaging refers to visualization techniques provide valuable information about the internal structures of human body for clinical applications, diagnosis, treatment, and scientific research. Segmentation is one primary methods analyzing processing medical images, which helps doctors diagnose accurately by providing detailed on body’s required part. However, segmenting images faces several challenges, such as requiring trained experts being time-consuming error-prone. Thus, it appears necessary an automatic image segmentation system. Deep learning algorithms have recently shown outstanding performance tasks, especially semantic networks that pixel-level understanding. By introducing first fully convolutional network (FCN) segmentation, been proposed its basis. One state-of-the-art in field U-Net. This paper presents a novel end-to-end model, named Ens4B-UNet, ensembles four U-Net architectures with pre-trained backbone networks. Ens4B-UNet utilizes U-Net’s success significant improvements adapting powerful robust neural (CNNs) backbones U-Nets encoders using nearest-neighbor up-sampling decoders. designed based weighted average ensemble encoder-decoder models. The all ensembled models are ImageNet dataset exploit benefit transfer learning. For improving our models, we apply training predicting, including stochastic weight averaging (SWA), data augmentation, test-time augmentation (TTA), different types optimal thresholds. We evaluate test 2019 Pneumothorax Challenge dataset, contains 12,047 12,954 masks 3,205 images. Our achieves 0.8608 mean Dice similarity coefficient (DSC) set, among top one-percent systems Kaggle competition.

Language: Английский

---

The COVID-19 epidemic analysis and diagnosis using deep learning: A systematic literature review and future directions

Computers in Biology and Medicine, Journal Year: 2021, Volume and Issue: 141, P. 105141 - 105141

Published: Dec. 14, 2021

Language: Английский

---

Detecting COVID-19 in chest CT images based on several pre-trained models

Multimedia Tools and Applications, Journal Year: 2024, Volume and Issue: 83(24), P. 65267 - 65287

Published: Jan. 15, 2024

Abstract This paper explores the use of chest CT scans for early detection COVID-19 and improved patient outcomes. The proposed method employs advanced techniques, including binary cross-entropy, transfer learning, deep convolutional neural networks, to achieve accurate results. COVIDx dataset, which contains 104,009 images from 1,489 patients, is used a comprehensive analysis virus. A sample 13,413 this dataset categorised into two groups: 7,395 individuals with confirmed 6,018 normal cases. study presents pre-trained learning models such as ResNet (50), VGG (19), (16), Inception V3 enhance DCNN classifying input images. cross-entropy metric compare cases based on predicted probabilities each class. Stochastic Gradient Descent Adam optimizers are employed address overfitting issues. shows that accuracies 99.07%, 98.70%, 98.55%, 96.23%, respectively, in validation set using optimizer. Therefore, work demonstrates effectiveness enhancing accuracy DCNNs image classification. Furthermore, provides valuable insights development more efficient diagnostic tools COVID-19.

Language: Английский

---

Applications of artificial neural network to solve the nonlinear COVID-19 mathematical model based on the dynamics of SIQ

Journal of Taibah University for Science, Journal Year: 2022, Volume and Issue: 16(1), P. 874 - 884

Published: Sept. 19, 2022

The purpose of this research work is to present a numerical study through artificial neural networks (ANNs) solve SIQ-based COVID-19 mathematical model using the effects lockdown. lockdown are considered three-dimensional model, "susceptible", "infective" and "quarantined", i.e. SIQ system. ANNs Levenberg–Marquardt backpropagation (LMB) used analyses system-based COVID-19. Three different types authentications, testing training as sample data applied Statistical ratios for selected – 80% 10% both authentication. obtained results system have been compared with reference dataset, which constructed Adams solutions. performances nonlinear dynamical testified reduction error in mean square sense range 10−9 10−12.

Language: Английский

---

An augmentation aided concise CNN based architecture for COVID-19 diagnosis in real time

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Jan. 11, 2024

Abstract Over 6.5 million people around the world have lost their lives due to highly contagious COVID 19 virus. The virus increases danger of fatal health effects by damaging lungs severely. only method reduce mortality and contain spread this disease is promptly detecting it. Recently, deep learning has become one most prominent approaches CAD, helping surgeons make more informed decisions. But models are computation hungry devices with TPUs GPUs needed run these models. current focus machine research on developing that can be deployed mobile edge devices. To end, aims develop a concise convolutional neural network-based computer-aided diagnostic system for in X-ray images, which may limited processing resources, such as phones tablets. proposed architecture aspires use image enhancement first phase data augmentation second pre-processing, additionally hyperparameters also optimized obtain optimal parameter settings third provide best results. experimental analysis provided empirical evidence impact enhancement, augmentation, hyperparameter tuning network model, increased accuracy from 94 98%. Results evaluation show suggested gives an 98%, better than popular transfer like Xception, Resnet50, Inception.

Language: Английский

---

HANA: A Healthy Artificial Nutrition Analysis model during COVID-19 pandemic

Computers in Biology and Medicine, Journal Year: 2021, Volume and Issue: 135, P. 104606 - 104606

Published: June 30, 2021

Language: Английский

---

Modeling COVID-19 Cases Statistically and Evaluating Their Effect on the Economy of Countries

Mathematics, Journal Year: 2021, Volume and Issue: 9(13), P. 1558 - 1558

Published: July 2, 2021

COVID-19 infections have plagued the world and led to deaths with a heavy pneumonia manifestation. The main objective of this investigation is evaluate performance certain economies during crisis derived from pandemic. gross domestic product (GDP) global health security index (GHSI) countries belonging–or not–to Organization for Economic Cooperation Development (OECD) are considered. In paper, statistical models formulated study performance. models’ specifications include, as response variable, GDP variation/growth percentage in 2020, covariates: disease rate its start March 2020 until 31 December 2020; GHSI 2019; countries’ risk by default spreads July 2019 May belongingness or not OECD; per capita 2020. We test heteroscedasticity phenomenon present modeling. variable “COVID-19 cases million inhabitants” statistically significant, showing impact on each country’s economy through variation. Therefore, we report that affect economies, but OECD membership other factors also relevant.

Language: Английский

---