Transfer Learning Model Research Articles

Performance Analysis of Various Deep Transfer Learning Models for Bacterial Blight Disease Detection and Classification in Pomegranate Fruits

Background: Detecting and classifying diseases in pomegranate fruit using computer vision remains a challenging task due to the presence of numerous diseases. Recent research findings suggest that models based on Convolutional Neural Networks (CNNs) have shown significant enhancements in accuracy when it comes to classifying images of fruits and leaves. The fungal infection responsible for the disease swiftly propagates through the soil, infiltrating the roots of pomegranate plants. Presently, the sole method to halt its spread entails farmers conducting thorough inspections and promptly eliminating infected plants, an uphill task. Methods: The present study introduces a classification PomeNetV1 model for identifying pomegranate fruit diseases employing CNN and transfer learning techniques. For executing this system, the dataset is created by taking the images directly from the farms in Ballari, Bengaluru, Bagalakote, etc. The proposed pomegranate fruit disease dataset contains 5099 images of five categories: Alternaria, Anthracnose, Bacterial Blight, Cercospora and Healthy. The system under consideration utilizes TensorFlow as its framework for developing deep learning models. Result: This paper evaluates ten different pre-trained models using a transfer learning approach and a proposed novel PomeNetV1 model for classification. All pre-trained models achieved over 99% classification accuracy. However, the PomeNetV1 model stands out with an impressive 99.80% accuracy, making it the most efficient for detecting healthy and bacterial blight diseases compared to others. Surpassing existing models like Vasumathi et al., 2021 deep CNN model with at least 1.63% accuracy, the PomeNetV1 model offers a feasible solution for detecting bacterial blight in pomegranate crops.

MetaPhenotype: A Transferable Meta-Learning Model for Single-Cell Mass Spectrometry-Based Cell Phenotype Prediction Using Limited Number of Cells.

Single-cell mass spectrometry (SCMS) is an emerging tool for studying cell heterogeneity according to variation of molecular species in single cells. Although it has become increasingly common to employ machine learning models in SCMS data analysis, such as the classification of cell phenotypes, the existing machine learning models often suffer from low adaptability and transferability. In addition, SCMS studies of rare cells can be restricted by limited number of cell samples. To overcome these limitations, we performed SCMS analyses of melanoma cancer cell lines with two phenotypes (i.e., primary and metastatic cells). We then developed a meta-learning-based model, MetaPhenotype, that can be trained using a small amount of SCMS data to accurately classify cells into primary or metastatic phenotypes. Our results show that compared with standard transfer learning models, MetaPhenotype can rapidly predict and achieve a high accuracy of over 90% with fewer new training samples. Overall, our work opens the possibility of accurate cell phenotype classification based on fewer SCMS samples, thus lowering the demand for sample acquisition.

Intelligent Crowd Density Classification Using Improved Metaheuristics with Transfer Learning Model on Smart Cities

Intelligent Crowd Density Classification Using Improved Metaheuristics with Transfer Learning Model on Smart Cities

Comparative Study: Flower Classification using Deep Learning, SMOTE and Fine-Tuning

Deep learning is a technology that can be used to classify flowers. In this research, flower type classification using the CNN method with several existing CNN architectures will be discussed. The data consists of 4317 images in .jpg format, covering 5 classes that is sunflower, dandelion, daisy, tulip and rose. The distribution of data for each class is daisy with 764 pictures, dandelion with 1052 pictures, rose with 784 pictures, sunflower with 733 pictures, and tulip with 984 pictures. With total dataset of 4317 pictures is further split to training data with ratio of 60%, validation with ratio of 10%, and testing with ratio of 30% to process with the CNN method and CNN framework. Due to the imbalance data distribution, the SMOTE method is applied to balancing number of samples in each class. This research compares CNN architectures, including CNN, GoogleNet, DenseNet, and MobileNet, where each transfer learning model undergoes fine-tuning to improve performance. At the classification stage, performance will be measured based on model testing accuracy. The accuracy obtained using CNN is 74.61%, using GoogleNet is 87.45%, DenseNet is 93.92%, and MobileNet is 88.34%.

A transfer learning model for rolling bearing fault diagnosis based on texture loss strategy and nuclear norm regularization

Abstract In recent years, transfer learning (TL) approaches have seen extensive application in diagnosing bearing faults due to their exceptional performance. However, mechanical noise, equipment aging, and wear lead to notable disparities and differences in the multi-level feature distributions across the source and target domain signals. The issue is addressed by proposing a TL model based on a texture loss strategy and nuclear norm regularization method. First, a feature-enhanced network is designed, which significantly improves the ability to capture local details and long-range dependencies by combining a multi-scale feature extraction module with a dilated residual module. Next, a texture loss strategy is proposed to align multi-scale features across domains by minimizing the Gram matrix of signal features. Finally, a nuclear norm regularization method is proposed to perform low-rank approximation on the signal matrix, facilitating the extraction of more robust feature data and mitigating the risk of overfitting. The experimental results demonstrate that the proposed method achieved an average accuracy of 98.58% on the University of Ottawa bearing fault dataset and 98.11% on the Jiangnan University bearing dataset, surpassing eight other algorithms in bearing fault diagnosis.

Fine-tuning pre-trained networks with emphasis on image segmentation: A multi-network approach for enhanced breast cancer detection

Accurate classification of mammography images into normal and cancerous categories is critical for the early detection of breast cancer. This study utilizes transfer learning and deep learning models to extract and diversify features from a combined dataset consisting of the Mini Digital Database for Screening Mammography (DDSM, containing 7808 images) and the Mammographic Image Analysis Society (MIAS) dataset (containing 322 images). The preprocessing steps involve image cropping, removal of artifacts, and enhancement of contrast using Contrast-Limited Adaptive Histogram Equalization (CLAHE). Data augmentation techniques, including the application of median and Gaussian blur, were used to improve the robustness of the models. Three pre-trained networks—Residual Networks with 50 layers (ResNet-50), Visual Geometry Group Network with 19 layers (VGG-19), and Residual Networks with 152 layers Version 2 (ResNet-152V2)—were fine-tuned specifically for mammography data. Image segmentation and the removal of the pectoral muscle significantly improved classification accuracy. The VGG-19 model achieved an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.80 for segmented images and 0.86 for non-segmented images. A stacked generalization model, which combined features from all three networks, further optimized performance. Artificial Neural Network (ANN) and Extreme Gradient Boosting (XGBoost) models achieved AUCs of 0.897 and 0.890, respectively, for segmented images. Data augmentation improved performance by 2.7%–4%.

Neonatal Jaundice Severity Detection from Skin Images using Deep Transfer Learning Techniques

Neonates in the initial weeks postpartum frequently experience jaundice, a prevalent medical condition characterized by the yellow discoloration of the sclera and integumentary surfaces. This phenomenon transpires as a result of the elevation of bilirubin concentrations within the circulatory system. When bilirubin levels reach critical thresholds, they present a considerable risk for severe complications, including neurological impairment, which represents one of the gravest outcomes that may ensue if the condition is not addressed with due diligence. This study investigates a non-invasive method for assessing jaundice severity in full-term infants from 1 to 29 days, focusing on infants in Mosul city. A dataset of 344 images was collected using an iPhone 12 Pro Max (9MP camera) at Ibn Al-Atheer Hospital, capturing various skin tones and lighting conditions to ensure accurate analysis. Advanced computer vision techniques were used to classify jaundice severity into three and four categories based on skin images. Pre-trained deep transfer learning models, namely VGG16 and ResNet50, were utilized for training, with the fully connected layer removed and a suitable classifier designed for each model. VGG16 achieved 91.71% accuracy for the three-category classification, while ResNet50 reached 95.98%. For the four-category classification, accuracies of 94.92% and 94.66% were achieved, respectively. These high accuracy levels suggest that non-invasive, image-based assessments can reduce the need for repeated blood tests. This research highlights the potential of using smartphone-based methods for jaundice screening in neonatal care, providing a reliable, accessible tool to reduce strain on medical facilities and improve early detection.

Neural network–based transfer learning to improve stiffness modeling of industrial robots with small experimental data sets

Stiffness modeling is an essential subject for the composition of robot control. Accurate stiffness modeling is helpful for improving the control accuracy of industrial robots, particularly under dynamic load circumstances. The classic virtual joint modeling (VJM) method is challenging in predicting the deformation of the end-effector throughout the full workspace due to the nonlinear deformation of the robot joint and its serial articulated structure. This paper proposes a full-space stiffness modeling method for robots based on the integration of a multi-layer perceptual (MLP) model and VJM. To provide enough training data for the MLP model, VJM is used to build a stiffness model with a small set of experimental data to generate 106,400 training data. A model-based transfer learning approach is proposed to improve the model’s accuracy and generalization regarding the difference between generated training data and actual experimental data. The VJM stiffness model is compared with the MLP stiffness model and the existing CNN-based transfer learning model based on the same experimental data. Considering the deformation prediction in the three directions in Cartesian space, the mean absolute error, standard deviation, and maximum error of the MLP model are decreased by at least 24.90%, 14.20%, and 8.50%, respectively, than the VJM. These prediction results demonstrate that the proposed modeling technique can significantly increase the accuracy of robot stiffness modeling, which is essential for position compensation in precise motion control of robots under dynamic load.

Multi-condition hawser tension prediction of offshore offloading system based on long and short-term memory network and transfer learning

Real-time prediction of hawser tension of the side-by-side oil offloading system of Floating Production Storage and Offloading (FPSO) can provide early warnings for hawser breakages and ship collision risk, thus improving structural, property, and environmental security. Although offline numerical models and Long Short-Term Memory (LSTM) could offer substantial precision, they confront intensive time costs in recalibrating or retraining models. This paper proposes an integration method of LSTM networks and transfer learning for real-time tension prediction considering inputs of actual remote wave elevations. We use short-term environmental data and numerical simulation data of correlated hawser tensions during offloading operations to train a pre-trained benchmark model for transfer learning. Then, a highly generalized and efficient transferred model is constructed by using a small sample to realize short-term tension predictions in time-varying environments. The results show the occurrence time and value of extreme tensions predicted by transfer learning nearly match the reference data, and their maximum errors are 3 s and 0.11, respectively, superior to LSTM direct training. Therefore, it provides sufficient demand for real-time prediction and early collision risk prevention in dynamically changing ocean environment. The research results could provide an alternative framework for intelligent monitoring of large-scale marine structures.

Use of Data Augmentation and Fine-tuned Transfer Learning Models for Classification of Cervical Cancer

Innovative strategies for early and accurate diagnosis of cervical cancer continue to be a global health priority. In this research, we present a powerful framework for improving cervical cancer classification by combining data augmentation methods with tailored transfer learning models. The lack of properly labelled medical data is a major roadblock to developing reliable diagnosis models. The geometric transformations, contrast tweaks, and noise addition that we use to increase the dataset artificially help us overcome this difficulty. The model's robustness is improved as a result of the additional exposure to diverse real-world circumstances provided by this supplemented dataset. We use a transfer learning strategy based on pre-trained convolutional neural networks (CNNs) to harness the potential of deep learning. These networks are well-suited for medical image analysis due to their ability to quickly and accurately identify features in a wide variety of images. Our expanded cervical cancer dataset is used to fine-tune these algorithms for their diagnostic purpose. Our experiments show that utilising both data augmentation and fine-tuned transfer learning considerably raises the accuracy of cervical cancer categorization. In terms of sensitivity, specificity, and overall accuracy, the model performs at a state-of-the-art level. This suggests that it may be useful in the diagnosis and early identification of cervical cancer. Additionally, our method demonstrates the significance of efficient data utilisation and model adaptability in the field of medicine. Our method is scalable and affordable, and it can be easily implemented in healthcare settings since we handle the problem of data scarcity and take advantage of the features of pre-trained deep learning models.

Experimental Assessment between Dissimilar Techniques and Methodologies to Sports Knee Injury using Magnetic Resonance Imaging

The anterior cruciate ligaments, which are crucial for conserving the normal biomechanics of human being knees, are the majority commonly injured knee-ligaments. An anterior cruciate ligament injury is originated by a split or wrench of the anterior cruciate ligaments, which are imperative ligaments in the knee. ACL injure is mainly and frequently caused by sports like football, soccer, and the like that require quick pauses or direction changes, jumping, and landings. These days, the area of diagnostics heavily relies on magnetic resonance imaging. It is effective in determining the presence of meniscal tears and damage to the cruciate ligament. This study's primary objective is to use magnetic resonance imaging knee images to find anterior cruciate ligament tears, which can be useful in identifying issues with the knee. Inception-v3, an established deep transfer learning (DTL) model based on a DCNN was used in this study to classify anterior cruciate ligament tears in MRI scans. Classification, Preprocessing, and feature extraction are the major processes used in the current study executions. The dataset type utilized in this article of research study was built using the MRNet database. Seventy percent of the data set is used for preparation and testing, while the lingering thirty percent is utilized for performance analysis in this comparison model. The future augmented methodology can improve upon the present models' performance through the application of DL and ML techniques.

Comparative analysis of vision transformers and fine-tuned transfer learning models for brain tumor classification

Brain tumors are a primary factor causing cancer-related deaths globally, and their classification remains a significant research challenge due to the variability in tumor intensity, size, and shape, as well as the similar appearances of different tumor types. Accurate differentiation is further complicated by these factors, making diagnosis difficult even with advanced imaging techniques such as magnetic resonance imaging (MRI). Recent techniques in artificial intelligence (AI), in particular deep learning (DL), have improved the speed and accuracy of medical image analysis, but they still face challenges like overfitting and the need for large annotated datasets. This study addresses these challenges by presenting two approaches for brain tumor classification using MRI images. The first approach involves fine-tuning transfer learning cutting-edge models, including SEResNet, ConvNeXtBase, and ResNet101V2, with global average pooling 2D and dropout layers to minimize overfitting and reduce the need for extensive preprocessing. The second approach leverages the Vision Transformer (ViT), optimized with the AdamW optimizer and extensive data augmentation. Experiments on the BT-Large-4C dataset demonstrate that SEResNet achieves the highest accuracy of 97.96%, surpassing ViT’s 95.4%. These results suggest that fine-tuning and transfer learning models are more effective at addressing the challenges of overfitting and dataset limitations, ultimately outperforming the Vision Transformer and existing state-of-the-art techniques in brain tumor classification.

Advancing mango leaf variant identification with a robust multi-layer perceptron model

Mango, often regarded as the “king of fruits,” holds a significant position in Bangladesh’s agricultural landscape due to its popularity among the general population. However, identifying different types of mangoes, especially from mango leaves, poses a challenge for most people. While some studies have focused on mango type identification using fruit images, limited work has been done on classifying mango types based on leaf images. Early identification of mango types through leaf analysis is crucial for taking proactive steps in the cultivation process. This research introduces a novel multi-layer perceptron model called WaveVisionNet, designed to address this challenge using mango leaf datasets collected from five regions in Bangladesh. The MangoFolioBD dataset, comprising 16,646 annotated high-resolution images of mango leaves, has been curated and augmented to enhance robustness in real-world conditions. To validate the model, WaveVisionNet is evaluated on both the publicly available dataset and the MangoFolioBD dataset, achieving accuracy rates of 96.11% and 95.21%, respectively, outperforming state-of-the-art models such as Vision Transformer and transfer learning models. The model effectively combines the strengths of lightweight Convolutional Neural Networks and noise-resistant techniques, allowing for accurate analysis of mango leaf images while minimizing the impact of noise and environmental factors. The application of the WaveVisionNet model for automated mango leaf identification offers significant benefits to farmers, agricultural experts, agri-tech companies, government agencies, and consumers by enabling precise diagnosis of plant health, enhancing agricultural practices, and ultimately improving crop yields and quality.

A combination between transfer learning models and UNet++ for COVID-19 diagnosis

A combination between transfer learning models and UNet++ for COVID-19 diagnosis

Introduction to the Special Issue on Realistic Synthetic Data: Generation, Learning, Evaluation

This is the foreword of our special issue volume on Realistic Synthetic Data: Generation, Learning, Evaluation organized with the ACM Transactions on Multimedia Computing, Communications, and Applications. It presents the target of the special issue that relates to synthetic data for various modalities, e.g., signals, images, volumes, audio, etc., controllable generation for learning from synthetic data, transfer learning and generalization of models, causality in data generation, addressing bias, limitations and trustworthiness in data generation, evaluation measures/protocols and benchmarks to assess quality of synthetic content, open synthetic datasets and software tools, and ethical aspects of synthetic data. The call for papers received a record number of 40 submissions out of which 15 were finally accepted for publication. This introduction provides an overview of the topics of each of the articles.

Efficient Brain Cancer Classification Based on Advanced CNN and Transformer Technologies

Abstract. This study addresses the challenge faced by primary medical staff in accurately distinguishing different types of brain cancer by comparing the training accuracy and iteration speed of various classification models. Specifically, thesis evaluated a basic Convolutional Neural Network (CNN) alongside two transfer learning models: EfficientNetV2 and Vision Transformer. The analysis is based on publicly available data from Kaggle, which provided a robust dataset for evaluation. The results indicate that both EfficientNetV2 and Vision Transformer models achieve significantly higher iteration convergence speeds and classification accuracy compared to the basic CNN model trained solely on the database. These advanced models demonstrate a capacity for efficient and rapid brain cancer differentiation, making them suitable for settings with limited computing resources. The study highlights that the transfer learning models can effectively support outpatient doctors by providing fast and accurate etiological screening without requiring extensive hardware. Overall, the findings suggest that EfficientNetV2 and Vision Transformer models offer a practical solution for improving brain cancer classification in primary healthcare environments, where rapid and precise diagnosis is crucial. This research underscores the potential of transfer learning techniques to enhance diagnostic capabilities while accommodating resource constraints.

Transforming Skin Cancer Diagnosis: A Deep Learning Approach with the Ham10000 Dataset

Skin cancer (SC) is one of the three most common cancers worldwide. Melanoma has the deadliest potential to spread to other parts of the body among all SCs. For SC treatments to be effective, early detection is essential. The high degree of similarity between tumor and non-tumors makes SC diagnosis difficult even for experienced doctors. To address this issue, authors have developed a novel Deep Learning (DL) system capable of automatically classifying skin lesions into seven groups: actinic keratosis (AKIEC), melanoma (MEL), benign keratosis (BKL), melanocytic Nevi (NV), basal cell carcinoma (BCC), dermatofibroma (DF), and vascular (VASC) skin lesions. Authors introduced the Multi-Grained Enhanced Deep Cascaded Forest (Mg-EDCF) as a novel DL model. In this model, first, researchers utilized subsampled multigrained scanning (Mg-sc) to acquire micro features. Second, authors employed two types of Random Forest (RF) to create input features. Finally, the Enhanced Deep Cascaded Forest (EDCF) was utilized for classification. The HAM10000 dataset was used for implementing, training, and evaluating the proposed and Transfer Learning (TL) models such as ResNet, AlexNet, and VGG16. During the validation and training stages, the performance of the four networks was evaluated by comparing their accuracy and loss. The proposed method outperformed the competing models with an average accuracy score of 98.19%. Our proposed methodology was validated against existing state-of-the-art algorithms from recent publications, resulting in consistently greater accuracies than those of the classifiers.

Automatic delineation of cervical cancer target volumes in small samples based on multi-decoder and semi-supervised learning and clinical application

Radiotherapy has been demonstrated to be one of the most significant treatments for cervical cancer, during which accurate and efficient delineation of target volumes is critical. To alleviate the data demand of deep learning and promote the establishment and promotion of auto-segmentation models in small and medium-sized oncology departments and single centres, we proposed an auto-segmentation algorithm to determine the cervical cancer target volume in small samples based on multi-decoder and semi-supervised learning (MDSSL), and we evaluated the accuracy via an independent test cohort. In this study, we retrospectively collected computed tomography (CT) datasets from 71 pelvic cervical cancer patients, and a 3:4 ratio was used for the training and testing sets. The clinical target volumes (CTVs) of the primary tumour area (CTV1) and pelvic lymph drainage area (CTV2) were delineated. For definitive radiotherapy (dRT), the primary gross target volume (GTVp) was simultaneously delineated. According to the data characteristics for small samples, the MDSSL network structure based on 3D U-Net was established to train the model by combining clinical anatomical information, which was compared with other segmentation methods, including supervised learning (SL) and transfer learning (TL). The dice similarity coefficient (DSC), 95% Hausdorff distance (HD95) and average surface distance (ASD) were used to evaluate the segmentation performance. The ability of the segmentation algorithm to improve the efficiency of online adaptive radiation therapy (ART) was assessed via geometric indicators and a subjective evaluation of radiation oncologists (ROs) in prospective clinical applications. Compared with the SL model and TL model, the proposed MDSSL model displayed the best DSC, HD95 and ASD overall, especially for the GTVp of dRT. We calculated the above geometric indicators in the range of the ground truth (head-foot direction). In the test set, the DSC, HD95 and ASD of the MDSSL model were 0.80/5.85 mm/0.95 mm for CTV1 of post-operative radiotherapy (pRT), 0.84/ 4.88 mm/0.73 mm for CTV2 of pRT, 0.84/6.58 mm/0.89 mm for GTVp of dRT, 0.85/5.36 mm/1.35 mm for CTV1 of dRT, and 0.84/4.09 mm/0.73 mm for CTV2 of dRT, respectively. In a prospective clinical study of online ART, the target volume modification time (MTime) was 3–5 min for dRT and 2–4 min for pRT, and the main duration of CTV1 modification was approximately 2 min. The introduction of the MDSSL method successfully improved the accuracy of auto-segmentation for the cervical cancer target volume in small samples, showed good consistency with RO delineation and satisfied clinical requirements. In this prospective online ART study, the application of the segmentation model was demonstrated to be useful for reducing the target volume delineation time and improving the efficiency of the online ART workflow, which can contribute to the development and promotion of cervical cancer online ART.

CAD-PsorNet: deep transfer learning for computer-assisted diagnosis of skin psoriasis

Psoriasis, being a chronic, inflammatory, lifelong skin disorder, has become a major threat to the human population. The precise and effective diagnosis of psoriasis continues to be difficult for clinicians due to its varied nature. In northern India, the prevalence of psoriasis among adult population ranges from 0.44 to 2.8%. Chronic plaque psoriasis accounts for over 90% of cases. This study utilized a dataset of 325 raw images collected from a reputable local hospital using a digital camera under uniform lighting conditions. These images were processed to generate 496 image patches (both diseased and normal), which were then normalized and resized for model training. An automated psoriasis image recognition framework was developed using four state-of-the-art deep transfer learning models: VGG16, VGG19, MobileNetV1, and ResNet-50. The convolutional layers adopted various edge, shape, and color filters to generate the feature map for psoriasis detection. Each pre-trained model was adapted with two dense layers, one dropout layer, and one output layer to classify input images. Among these models, MobileNetV1 achieved the best performance, with 94.84% sensitivity, 89.37% specificity, and 97.24% overall accuracy. Hyper-parameter tuning was performed using grid search to optimize learning rates, batch sizes, and dropout rates. The AdaGrad (Adaptive gradient)) optimizer was chosen for its adaptive learning rate capabilities, facilitating quicker convergence in model performance. Consequently, the methodology’s performance improved to 94.25% sensitivity, 96.42% specificity, and 99.13% overall accuracy. The model’s performance was also compared with non-machine learning-based diagnostic methods, yielding a Dice coefficient of 0.98. However, the model’s effectiveness is dependent upon high-quality input images, as poor image conditions may affect accuracy, and it may not generalize well across diverse demographics or psoriasis variations, highlighting the need for varied training datasets for robustness.

A secure fault detection for digital microfluidic biochips

Abstract Among all the modern technological advances, digital microfluidic biochip has been extending a salient solution to healthcare and bio-laboratories with the pledge of high sensitivity and reconfigurability. Such biochip devices fulfill the requirement of a faster testing kit for the detection of different novel diseases, which is indispensable in the market due to the tremendously disrupted scenario of healthcare systems. To eliminate erroneous testing, the current scope of digital microfluidic biochips is widened as a viable testing method using various bioprotocols with a reduced cost in developing countries. This paper addresses the existing security challenges and operational faults in the identification mechanism of proteins such as severe acute respiratory syndrome coronavirus 2 spike protein in state-of-the-art digital microfluidic biochips. We are the first to propose a safety detection solution along with a fault identification algorithm using an inductive transfer learning model. Experimental results of the proposed model register a threshold accuracy of 98% while applying the own dataset. This work will provide a better security-enabled fault-free safety assurance framework against attack and fault identification with better accuracy in digital microfluidic biochips for the detection of different diseases and many other healthcare diagnoses, without any overhead of completion time for bioprotocols.