Transfer Learning Research Articles

A diesel engine valve leakage fault diagnostic method using modified VGG19 and transfer learning

A diesel engine valve leakage fault diagnostic method using modified VGG19 and transfer learning

Transfer Learning-Based Health Monitoring of Robotic Rotate Vector Reducer Under Variable Working Conditions

Due to their precision, compact size, and high torque transfer, Rotate vector (RV) reducers are becoming more popular in industrial robots. However, repetitive operations and varying speed conditions mean that these components are prone to mechanical failure. Therefore, it is important to develop effective health monitoring (HM) strategies. Traditional approaches for HM, including those using vibration and acoustic emission sensors, encounter such challenges as noise interference, data inconsistency, and high computational costs. Deep learning-based techniques, which use current electrical data embedded within industrial robots, address these issues, offering a more efficient solution. This research provides transfer learning (TL) models for the HM of RV reducers, which eliminate the need to train models from scratch. Fine-tuning pre-trained architectures on operational data for the three different reducers of health conditions, which are healthy, faulty, and faulty aged, improves fault classification across different motion profiles and variable speed conditions. Four TL models, EfficientNet, MobileNet, GoogleNet, and ResNET50v2, are considered. The classification accuracy and generalization capabilities of the suggested models were assessed across diverse circumstances, including low speed, high speed, and speed fluctuations. Compared to the other models, the proposed EfficientNet model showed the most promising results, achieving a testing accuracy and an F1-score of 98.33% each, which makes it best suited for the HM of robotic reducers.

Transfer learning based adaptive entropy loading for radio-over-fiber systems

Transfer learning based adaptive entropy loading for radio-over-fiber systems

Hierarchical Transfer Learning with Transformers to Improve Semantic Segmentation in Remote Sensing Land Use

Land use classification remains a significant challenge in remote sensing semantic segmentation. While convolutional neural networks (CNNs) are widely used, their inherent limitations, such as restricted receptive fields, hinder their widespread application in remote sensing. Additionally, the scarcity of labeled remote sensing data and domain shift issues adversely impact deep learning model performance. This study proposes a hierarchical transfer learning framework for fine-category semantic segmentation tasks, leveraging the powerful global relationship modeling capabilities of Transformer models to classify land use in Dongpo District, Meishan City, in mainland China. Our framework represents multilevel transfer learning, progressing from non-remote sensing classification to coarse classification, then to the refined classification of remote sensing. We compared the performance of Transformer models with representative baseline CNNs like U-Net and DeepLab V3+. Results show that the Swin-Unet model outperforms the other models used in this study. It achieved the highest test mean intersection over union (MIoU) of 0.837 and 0.810 for residential and transportation in level 1 (coarse) classification, respectively, and 0.545 for irrigated land in level 2 (fine-grained) classification. Transfer learning from pre-trained models significantly enhanced semantic segmentation accuracy compared to random parameter initialization (ranging from 0.4% to 17.7%), with up to a 17.7% improvement in test MIoU for the public land category. The hierarchical transfer learning framework further improved segmentation accuracy for corresponding level 2 categories, leveraging pre-trained level 1 models. Our study shows the applicability of Transformer-based transfer learning in remote sensing land use classification.

Abnormality detection in nailfold capillary images using deep learning with EfficientNet and cascade transfer learning

Nailfold Capillaroscopy (NFC) is a simple, non-invasive diagnostic tool used to detect microvascular changes in nailfold. Chronic pathological changes associated with a wide range of systemic diseases, such as diabetes, cardiovascular disorders, and rheumatological conditions like systemic sclerosis, can manifest as observable microvascular changes in the terminal capillaries of nailfolds. The current gold standard relies on experts performing manual evaluations, which is an exhaustive time-intensive, and subjective process. In this study, we demonstrate the viability of a deep learning approach as an automated clinical screening tool. Our dataset consists of NFC images from a total of 225 participants, with normal images accounting for 6% of the dataset. This study introduces a robust framework utilizing cascade transfer learning based on EfficientNet-B0 to differentiate between normal and abnormal cases within NFC images. The results demonstrate that pre-trained EfficientNet-B0 on the ImageNet dataset, followed by transfer learning from domain-specific classes, significantly enhances the classifier’s performance in distinguishing between Normal and Abnormal classes. Our proposed model achieved superior performance, with accuracy, precision, recall, F1 score, and ROC_AUC of 1.00, significantly outperforming both models of single transfer learning on the pre-trained EfficientNet-B0 and cascade transfer learning on a convolutional neural network, which each attained an accuracy, precision, recall, and F1 score of 0.67 and a ROC_AUC of 0.83. The framework demonstrates the potential to facilitate early preventive measures and timely interventions that aim to improve healthcare delivery and patients’ quality of life.

Attention-guided erasing for enhanced transfer learning in breast abnormality classification.

Breast cancer remains one of the most prevalent cancers globally, necessitating effective early screening and diagnosis. This study investigates the effectiveness and generalizability of our recently proposed data augmentation technique, attention-guided erasing (AGE), across various transfer learning classification tasks for breast abnormality classification in mammography. AGE utilizes attention head visualizations from DINO self-supervised pretraining to weakly localize regions of interest (ROI) in images. These localizations are then used to stochastically erase non-essential background information from training images during transfer learning. Our research evaluates AGE across two image-level and three patch-level classification tasks. The image-level tasks involve breast density categorization in digital mammography (DM) and malignancy classification in contrast-enhanced mammography (CEM). Patch-level tasks include classifying calcifications and masses in scanned film mammography (SFM), as well as malignancy classification of ROIs in CEM. AGE significantly boosts classification performance with statistically significant improvements in mean F1-scores across four tasks compared to baselines. Specifically, for image-level classification of breast density in DM and malignancy in CEM, we achieve gains of 2% and 1.5%, respectively. Additionally, for patch-level classification of calcifications in SFM and CEM ROIs, gains of 0.4% and 0.6% are observed, respectively. However, marginal improvement is noted in the mass classification task, indicating the necessity for further optimization in tasks where critical features may be obscured by erasing techniques. Our findings underscore the potential of AGE, a dataset- and task-specific augmentation strategy powered by self-supervised learning, to enhance the downstream classification performance of DL models, particularly involving ViTs, in medical imaging.

A Hot Rolling Full Process Rolling Force Prediction Method Based on Transfer Learning and Inception-LSTM Neural Network

A Hot Rolling Full Process Rolling Force Prediction Method Based on Transfer Learning and Inception-LSTM Neural Network

Adaptive Month Matching: A Phenological Alignment Method for Transfer Learning in Cropland Segmentation

The increasing demand for food and rapid population growth have made advanced crop monitoring essential for sustainable agriculture. Deep learning models leveraging multispectral satellite imagery, like Sentinel-2, provide valuable solutions. However, transferring these models to diverse regions is challenging due to phenological differences in crop growth stages between training and target areas. This study proposes the Adaptive Month Matching (AMM) method to align the phenological stages of crops between training and target areas for enhanced transfer learning in cropland segmentation. In the AMM method, an optimal Sentinel-2 monthly time series is identified in the training area based on deep learning model performance for major crops common to both areas. A month-matching process then selects the optimal Sentinel-2 time series for the target area by aligning the phenological stages between the training and target areas. In this study, the training area covered part of the Mississippi River Delta, while the target areas included diverse regions across the US and Canada. The evaluation focused on major crops, including corn, soybeans, rice, and double-cropped winter wheat/soybeans. The trained deep learning model was transferred to the target areas, and accuracy metrics were compared across different time series chosen by various phenological alignment methods. The AMM method consistently demonstrated strong performance, particularly in transferring to rice-growing regions, achieving an overall accuracy of 98%. It often matched or exceeded other phenological matching techniques in corn segmentation, with an average overall accuracy across all target areas exceeding 79% for cropland segmentation.

Assessment of using transfer learning with different classifiers in hypodontia diagnosis

BackgroundHypodontia is the absence of one or more teeth in the primary or permanent dentition during development, and radiographic imaging is the most common method of diagnosis. However, in recent years, artificial intelligence-based decision support systems have been employed to make highly accurate diagnoses. The aim of this study was to classify single premolar agenesis, multiple premolar agenesis, and without tooth agenesis using various artificial intelligence approaches.MethodsOne thousand sixty-eight panoramic radiographs from pediatric patients aged between 6 and 12 years without systemic disease were sorted into three separate classes: single premolar agenesis (n = 336), multiple premolar agenesis (n = 324), and without tooth agenesis (n = 408). Pretrained convolutional neural network models (AlexNet, DarkNet-19, DarkNet-53, DenseNet-201, EfficientNet, GoogLeNet, InceptionV3, IncResV2, MobileNetV2, NasNet-Mobile, Places365, ResNet-18, ResNet-50, ResNet-101, ShuffleNet, SqueezeNet, VGG-16, VGG-19, and Xception) were used for training with the fine-tuning method and different machine learning classifiers (decision trees, discriminant analysis, logistic regression, naive Bayes, support vector machines, nearest neighbor, ensemble method, and artificial neural network). The dataset was divided into 80% for training and 20% for testing. Performance was evaluated via accuracy, recall, precision, F1-score, specificity and area under the curve (AUC) parameters.ResultsAll of the data were classified via a VGG-19 model with a bilayered neural network classifier, which achieved 95.63% accuracy, 93.26% precision, 93.34% recall, 96.73% specificity, 93.25% F1-score and 95.03% AUC and was identified as the most successful model. The accuracy values for this model were distributed as follows: 96.72% for without tooth agenesis, 95.79% for multiple premolar agenesis, and 94.39% for single premolar agenesis.ConclusionsSuccessful results of pretrained models have been demonstrated for the radiographic diagnosis of hypodontia in pediatric patients. It is expected that artificial intelligence approaches will facilitate the diagnosis of hypodontia.

Potato Leaf Insight: An AI-Driven System for Detection of Potato Leaf Disease

Agricultural productivity is crucial for global economic development and growth. Crop diseases can severely risk to Nation’s food security, economic resources, environmental sustainability and agricultural output. Crop disease detection at early stage not only reduce losses for farmers but also improve crop yields. As agricultural technology and artificial intelligence advance, research into sustainable agricultural development becomes increasingly crucial. Plant diseases significantly impact the yield and quality of potatoes, and manually diagnosing these diseases can be both time-consuming and complex due to the required expertise. This paper introduces PotatoLeaf Insight System that employs U-Net technique for image segmentation followed by VGG19, for feature extraction through transfer learning. The model is trained and evaluated using a dataset of potato leaf images, covering early blight disease, late blight disease, leaf roll disease, Verticillium wilt disease, and healthy leaves. This AI driven system provide overall accuracy of 94% and useful for farmers to take necessary care of their crops in case detected with disease.

Thermal Image Classification of Autistic Children Using Res-Net Architecture

The thermal Image Classification Method has been widely used for significant applications in many fields, including thermal images of the face. This study presents a method for thermal facial classification in children with autism spectrum disorder (ASD). Children with ASD have a neurological disorder that affects communication skills essential in daily life and often causes difficulties in social situations. As we know, the diagnosis of ASD currently still relies on human methods and does not yet have definite biological markers. Early diagnosis of ASD has a significant positive impact, especially in children. Deep learning techniques, especially in facial medical image analysis, have become a new research focus in ASD detection. Initial screening using a Convolutional Neural Network (CNN) model with a transfer learning approach offers great potential for early diagnosis of ASD. The use of thermal imaging as a passive method to analyze ASD-related physiological signals has been proposed. In previous research, a deep learning model was developed to classify the faces of autistic children using thermal images. Therefore, this study aims to create a new Thermal Image Classification model for Autistic Children Using Res-Net Architecture. The architectures applied are ResNet-18, ResNet-34, and ResNet-50. As a comparison system, several of the same parameter values are used: epoch 100, batch size 2, SGD, Cross-entropy, learning rate 0.001, and momentum 0.9. The study test results show that the results of ResNet-18 are 97.22%, ResNet-34 99.22%, and ResNet-50 99.41%. Based on these results, ResNet-50 has the highest value.

Efficient Pest Detection Through Advanced Machine Learning Technique

Crop Protection is the key element to achieve food security. Many studies have been conducted over the decades to avoid crop losses during pre- harvest and post-harvest stages. Crop losses due to pest attack and plant disease spread, reduces the agriculture production and possess a direct impact on the economy of a country. Deployment of Artificial Intelligence (AI) based pest control strategies to detect pest species is under research. In this manuscript, deep learning based EfficientNetB7 architecture and transfer learning methodology is used to develop a real-time resource-efficient pest detection system. EfficientNetB7's innovative compound scaling technique has managed to balance efficiency in terms of computations and accuracy to effectively classify pests through images with minimal resources. The proposed system uses appropriate fine-tuning of training parameters and regularization mechanisms such as optimizers, data augmentation, to effectively develop a pest detection system. The trained model is ported on to STM32 microcontroller using CubeAI in STM32CubeIDE.The model is assessed by using a publicly available dataset, and a classification accuracy of 93.5% is achieved. This system not only automates pest detection but also provides a match percentage for the identified pests, thus supporting precision agriculture. Future work includes increasing the dataset and exploring edge AI techniques for decentralized decision-making. This manuscript showcases the transformative potential of EfficientNetB7 in precision agriculture, offering a scalable, cost-effective, and sustainable solution for the detection of pests with relevant broader impacts on agricultural automation practices.

Multi-Modal Fusion of Routine Care Electronic Health Records (EHR): A Scoping Review

Background: Electronic health records (EHR) are now widely available in healthcare institutions to document the medical history of patients as they interact with healthcare services. In particular, routine care EHR data are collected for a large number of patients.These data span multiple heterogeneous elements (i.e., demographics, diagnosis, medications, clinical notes, vital signs, and laboratory results) which contain semantic, concept, and temporal information. Recent advances in generative learning techniques were able to leverage the fusion of multiple routine care EHR data elements to enhance clinical decision support. Objective: A scoping review of the proposed techniques including fusion architectures, input data elements, and application areas is needed to synthesize variances and identify research gaps that can promote re-use of these techniques for new clinical outcomes. Design: A comprehensive literature search was conducted using Google Scholar to identify high impact fusion architectures over multi-modal routine care EHR data during the period 2018 to 2023. The guidelines from the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) extension for scoping review were followed. The findings were derived from the selected studies using a thematic and comparative analysis. Results: The scoping review revealed the lack of standard definition for EHR data elements as they are transformed into input modalities. These definitions ignore one or more key characteristics of the data including source, encoding scheme, and concept level. Moreover, in order to adapt to emergent generative learning techniques, the classification of fusion architectures should distinguish fusion from learning and take into consideration that learning can concurrently happen in all three layers of new fusion architectures (i.e., encoding, representation, and decision). These aspects constitute the first step towards a streamlined approach to the design of multi-modal fusion architectures for routine care EHR data. In addition, current pretrained encoding models are inconsistent in their handling of temporal and semantic information thereby hindering their re-use for different applications and clinical settings. Conclusions: Current routine care EHR fusion architectures mostly follow a design-by-example methodology. Guidelines are needed for the design of efficient multi-modal models for a broad range of healthcare applications. In addition to promoting re-use, these guidelines need to outline best practices for combining multiple modalities while leveraging transfer learning and co-learning as well as semantic and temporal encoding.

Transfer Learning for Thickener Control

Thickener control is a key area of focus in the minerals processing industry, particularly due to its crucial role in water recovery, which is essential for sustainable resource management. The highly nonlinear nature of thickener dynamics presents significant challenges in modeling and optimization, making it a strong candidate for advanced surrogate modeling techniques. However, traditional data-driven approaches often require extensive datasets, which are frequently unavailable, especially in new plants or unexplored operational domains. Developing data-driven models without enough data representative of the dynamics of the system could result in incorrect predictions and consequently, unstable response of the controller. This paper proposes the application of a methodology that leverages transfer learning to address these data limitations to enhance surrogate modeling and model predictive control (MPC) of thickeners. The performance of three approaches—a base model, a transfer learning model, and a physics-informed neural network (PINN)—are compared to demonstrate the effectiveness of transfer learning in improving control strategies under limited data conditions.

Skin cancer classification using machine learning

Abstract Skin cancer is a major public health problem, especially in the western world, having direct negative impact on life expectancy, quality of life, and on the economy in general. However, early detection of skin cancer could significantly reduce the mortality rate. For this purpose, tremendous efforts have been deployed in recent years for developing machine learning algorithms that can help in the early detection of skin cancer. In this paper, we focus on the classification of the three most common types of skin cancer lesions: Basal cell carcinoma (BCC), squamous cell carcinoma (SCC), and melanoma (MEL). Two different convolution neural network (CNN) architectures were implemented for this aim: YOLO, version 7 (v7) using transfer learning and an in-house developed CNN algorithm with optimum number of layers and hyper-parameters. The results obtained by implementing the two algorithms with a total number of 2,792 training samples (after performing data augmentation) show better performance compared to some of the recently published works in the literature. Using YOLO, v7, the average accuracy, sensitivity, and specificity are 89.65%, 85%, and 91.90%, respectively. The aforementioned average values using the proposed CNN algorithm are 90.12%, 85.55%, and 92.57%, respectively.

Remaining Useful Life Prediction of Lithium-ion Battery based on AConvST-LSTM-Net-TL

Abstract The accurate prediction of the Remaining Useful Life (RUL) of lithium-ion batteries can significantly enhance the safety and reliability of these batteries, thereby reducing operational risks. However, numerous existing methodologies operate under the assumption that both training and testing data adhere to the same distribution pattern, hindering the application of successful laboratory-based models to different target batteries. Hence, this study introduces a transfer learning model for predicting lithium-ion battery life, utilizing an attention mechanism-driven convolutional neural network (ACNN) in conjunction with a spatiotemporal Long Short-Term Memory network (ST-LSTM). Initially, the deep charging process data is leveraged for battery capacity estimation, where a Generalized Additive Model (GAM) is employed to delineate the nonlinear trend of battery capacity decay and characterize the capacity degradation. Following feature extraction, the Maximum Mean Discrepancy (MMD) value is computed to assess the distribution disparity between the two domains. Subsequently, the AConvST-LSTM-Net computes capacity estimations, loss functions for both source and target domains, merging these with the MMD value to formulate a comprehensive loss function. In conclusion, by employing transfer learning, the refined model is adapted to the target domain dataset, enabling the accurate prediction of the RUL of lithium-ion batteries. This approach is validated using the NASA lithium-ion battery dataset and the National Big Data Alliance for New Energy Vehicles (NDANEV), demonstrating the superior accuracy and robustness of the AConvST-LSTM-Net-TL model in comparisn to other prediction methods.

Artificial intelligence in retinopathy of prematurity: transfer learning and federated learning

Artificial intelligence in retinopathy of prematurity: transfer learning and federated learning

Progress in estimating the state of health using transfer learning–based electrochemical impedance spectroscopy of lithium-ion batteries

Progress in estimating the state of health using transfer learning–based electrochemical impedance spectroscopy of lithium-ion batteries

A Target Domain-Specific Classifier Weight Partial Transfer Adversarial Network for Bearing Fault Diagnosis

In actual industry applications, the failure categories of practical equipment are usually a subset of laboratory conditions failure categories. Due to the strict constraints, partial transfer learning can address a more practical diagnostic scenario. In view of this, this paper proposes a target domain-specific classifier weight partial transfer adversarial network. Initially, the 1-D convolutional neural network is employed as the basic architecture. By training the domain discriminator and feature generator with an adversarial strategy, the recognition ability of the domain discriminant network and the feature extraction ability of the feature generation network can be enhanced. After that, a weighted learning strategy is introduced to guide the model to learn the cross-domain invariant feature. Also, a specific target domain classifier is utilized to redivide the target domain decision boundary to accurately classify the unlabeled target domain samples. Finally, five mainstream deep neural network methods are taken for comparison using the data from Western Reserve University and the motor-magnetic brake test designed by us. The results show that the proposed method reaches 90.18% and 96.53% classification accuracy on two datasets, respectively, which demonstrates superior performance compared with the state-of-the-art methods.

MMFuncPhos: A Multi-Modal Learning Framework for Identifying Functional Phosphorylation Sites and Their Regulatory Types.

Protein phosphorylation plays a crucial role in regulating a wide range of biological processes, and its dysregulation is strongly linked to various diseases. While many phosphorylation sites have been identified so far, their functionality and regulatory effects are largely unknown. Here, a deep learning model MMFuncPhos, based on a multi-modal deep learning framework, is developed to predict functional phosphorylation sites. MMFuncPhos outperforms existing functional phosphorylation site prediction approaches. EFuncType is further developed based on transfer learning to predict whether phosphorylation of a residue upregulates or downregulates enzyme activity for the first time. The functional phosphorylation sites predicted by MMFuncPhos and the regulatory types predicted by EFuncType align with experimental findings from several newly reported protein phosphorylation studies. The study contributes to the understanding of the functional regulatory mechanism of phosphorylation and provides valuable tools for precision medicine, enzyme engineering, and drug discovery. For user convenience, these two prediction models are integrated into a web server which can be accessed at http://pkumdl.cn:8000/mmfuncphos.