Optimal Transfer Learning Research Articles

A recommendation model for optimizing transfer learning hyper-parameter settings in building heat load prediction with limited data samples

The transfer learning method has gained increasing attention in the domain of building load prediction, particularly in scenarios with limited data samples. Its core principle involves leveraging knowledge obtained from abundant data in source buildings to aid the learning process of models for the target buildings. Existing research has predominantly concentrated on optimizing the selection of source building data to improve transfer learning effectiveness, while the optimization of transfer learning hyper-parameter settings is often neglected. This study proposes a recommendation model tailored for transfer learning hyper-parameter settings in the context of small sample prediction for building heat loads. The objective is to automatically suggest suitable transfer learning hyper-parameter combination based on the specific features of the building heat load data samples. In this study, 200 real building profiles were utilized to generate the input–output dataset required for the recommendation model. By employing data mining techniques such as clustering and classification, the correlation between the features of source building data and the most effective transfer learning hyper-parameter combination is investigated. The developed recommendation model for optimal transfer learning hyper-parameter settings achieves a classification accuracy of 90.5%，and the performance evaluation was conducted using an additional dataset of 30 source buildings. The results show that by employing this recommendation model, the prediction error of the target buildings can be reduced by 0.12% to 6.64% compared to the conventional method of empirically determining transfer learning hyper-parameter settings.

Chatter monitoring method of Ti-6Al-4V thin-walled parts based on MAML optimized transfer learning

Chatter monitoring method of Ti-6Al-4V thin-walled parts based on MAML optimized transfer learning

Optimized Transfer Learning Based Dementia Prediction System for Rehabilitation Therapy Planning

Abstract: Dementia is a progressive neurodegenerative disease leading to cognitive decline, and current medical interventions can only slow its progression. Early prediction of dementia onset holds potential for preventive measures. This study aims to develop a machine learning model using transfer learning from magnetic resonance imaging (MRI) data to predict dementia. The project employs k-fold cross-validation and various parameter optimization algorithms during model training to enhance prediction accuracy. The final voting classifier and optimized Transfer Learning algorithms achieved an impressive accuracy, surpassing the performance of competing methods on the same dataset. This suggests the efficacy of the proposed transferlearning approach in predicting dementia data. The developed model enables early diagnosis of dementia, a crucial factor in halting neurological deterioration associated with the disease. This is particularly relevant for regions with limited access to human physicians, providing a valuable tool for underserved populations. The proposed system holds promise for planning rehabilitation therapy programs for dementia patients. As a tool for early diagnosis and intervention, the model not only contributes to improved patient outcomes but also addresses the challenges faced by regions lacking access to sufficient healthcare resources. And also include ensemble methods (Voting Classifier, Stacking Classifier) and a hybrid CNN-LSTM approach are employed for improving accuracy, with the Voting Classifier achieving an impressive 100% accuracy. To facilitate user testing, a user-friendly front end using the Flask framework, incorporating user authentication, was proposed for seamless application of this advanced dementia prediction system.

Motor imagery decoding using source optimized transfer learning based on multi-loss fusion CNN

Motor imagery decoding using source optimized transfer learning based on multi-loss fusion CNN

MFO-TL: modified firefly optimal transfer learning based motion correction of fetal brain and placenta MRI for thyroid prediction

MFO-TL: modified firefly optimal transfer learning based motion correction of fetal brain and placenta MRI for thyroid prediction

Multimodal Feature Fusion Using Optimal Transfer Learning Approach for Lung Cancer Detection and Classification on CT Images

Lung cancer detection is the process of detecting the presence of lung tumor or abnormalities in the lungs. Early diagnosis is crucial for increasing the chances of patient survival and successful treatment. When compared to X-rays, Computed Tomography (CT) images are more sensitive and are increasingly being used for the diagnosis and screening of lung tumors. They provide complete cross-sectional images of the lungs and it will even detect small lesions. AI and Machine learning (ML) approaches are most commonly employed to analyse medical images (e.g. CT scans) and detect lung cancer. This algorithm can help radiologists identify patterns indicative or subtle abnormalities of cancer. Medical diagnosis, particularly in complex diseases such as lung cancer, frequently involves ambiguity. The diagnostic system can alleviate ambiguity via cross-verifying findings from various sources by fusing multimodal features. Multimodal feature fusion using deep learning (DL) algorithm is an advanced technology that leverages the abilities of deep neural networks to combine data from three different modalities or sources for better robustness in several applications, namely natural language processing, image, and data analysis, etc. This study introduces a Multimodal Feature Fusion using an Optimal Transfer Learning Method for Lung Cancer Detection and Classification (MFFOTL-LCDC) methodology on CT images. The chief objective of the MFFOTL-LCDC methodology is to exploit the feature fusion process for the identification and classification of lung tumor. To attain this, the MFFOTL-LCDC model undergoes a multimodal feature fusion approach to derive feature vectors using 3 DL approaches such as SqueezeNet, CapsNet, and Inception v3 models. Besides, the MFFOTL-LCDC technique applies the remora optimization algorithm (ROA) for the hyperparameter choice of 3 DL models. For lung cancer recognition, the MFFOTL-LCDC algorithm exploits the deep extreme learning machine (DELM) algorithm. A series of simulations were conducted to ensure the greater lung cancer recognition outcomes of the MFFOTL-LCDC methodology. The extensive outcomes determine the improved results of the MFFOTL-LCDC technique over recent DL approaches.

Modelling of Bayesian-Based Optimized Transfer Learning Model for Cyber-Attack Detection in Internet of Things Assisted Resource Constrained Systems

Modelling of Bayesian-Based Optimized Transfer Learning Model for Cyber-Attack Detection in Internet of Things Assisted Resource Constrained Systems

A Genetic Algorithm-Based Optimized Transfer Learning Approach for Breast Cancer Diagnosis

A Genetic Algorithm-Based Optimized Transfer Learning Approach for Breast Cancer Diagnosis

Optimized transfer learning approach for leaf disease classification in smart agriculture

Optimized transfer learning approach for leaf disease classification in smart agriculture

Optimized Transfer Learning Based Short-term Electrical Load Interval Prediction

Electrical load forecasting is an essential foundation for power reliable and economical operation of the power grid. Most forecasting models regard the prediction results as deterministic variables, which ignores the randomness and volatility of the power load. At the same time, insufficient historical load data often lead to undertrained models, which affects the accuracy of capturing uncertain information. Therefore, we proposed an optimized transfer learning-based method for short-term load-interval prediction. A deep learning quantile regression model would be constructed by source domain data in the method, and the weights of the source model would be optimized to avoid negative transfer. Then, the target model is constructed by parameter transfer based on key layers and is tuned with hyperparameters by target domain data. From the experimental discussion, it is known that the model with an optimized transfer learning strategy can accurately quantify the fluctuation range of future power load.

Emergency vehicles classification for traffic signal system using optimized transfer DenseNet201 model

<span>As a result of the rapid growth of the world population, traffic signaling systems for monitoring and controlling the roads have turned to be an important issue facing humanity. To effectively overcome this problem, an accurate method for congestion reduction on the roads should be used which has a direct relation between the population and the cars’ usage. Various approaches derived from deep and transfer learning have been investigated in this context. This research implemented an optimized transfer learning approach for densely connected convolutional neural network (DenseNet201) models for multiple classifications (non-emergency cars, ambulance, police, and firefighter). Due to the non-availability of public datasets, a customized dataset has been created. This paper aims to improve the performance accuracy of vehicle classification using certain preprocessing algorithms on the input images and testing various optimization methods. The performance accuracy of the proposed model is evaluated using k-folds cross-validation 20:80 for the test and training, respectively. The metrics which are used for comparison with other deep learning techniques are based on exactness, recall, accuracy and F1-score. Test outcomes specify that the proposed transfer model-based optimization outperforms alternative deep learning algorithms regarding vehicle accuracy in classifying and reaches 98.6%.</span>

Optimized transfer learning based multi-modal medical image retrieval

Optimized transfer learning based multi-modal medical image retrieval

Optimal transfer learning based nutrient deficiency classification model in ridge gourd (Luffa acutangula)

The efficient detection of nutrient deficiency and proper fertilizer for that deficiency becomes the critical challenges various farmers face. The family Cucurbitaceae includes members cultivated globally as a source of indigenous medicines, food, and fiber. Luffa acutangula (L.) Roxb, generally called Ridge gourd, belongs to the Cucurbitaceae family and is an annual herb originating in several areas of India, particularly in the coastal regions. Nutrient deficiency detection in ridge gourd is essential to improve crop productivity. In agricultural practises, the early identification and categorization of nutrient deficiencies in crops is essential for sustaining optimal growth and production. Addressing these nutrient deficiencies, we applied the Ring Toss Game Optimization with a Deep Transfer Learning-based Nutrient Deficiency Classification (RTGODTL-NDC) to Ridge Gourd (Luffa acutangula). This research proposes a new ring toss game optimization with a deep transfer learning-based nutrient deficiency classification (RTGODTL-NDC) method. The RTGODTL-NDC technique uses pre-processing, segmentation, feature extraction, hyperparameter tuning, and classification. The Gabor filter (GF) is mainly used for image pre-processing, and the Adam optimizer with SqueezeNet model is utilized for feature extraction. Finally, the RTGO algorithm with the deep hybrid learning (HDL) model is applied to classify nutrient deficiencies. The suggested framework has the potential to improve crop management practises by allowing for proactive and targeted interventions, which will result in improved agricultural health, production, and resource utilisation. The outcomes represented by the RTGODTL-NDC method have resulted in improved performance. For example, based on accuracy and specificity, the RTGODTL-NDC methodology rendered maximum acc{u}_{y} of 97.16% and specificity of 98.29%. The outcomes show how effective the transfer learning-based model is in identifying nutrient deficits in ridge gourd plants, as seen by its high level of accuracy.

Selecting the optimal transfer learning model for precise breast cancer diagnosis utilizing pre-trained deep learning models and histopathology images

Selecting the optimal transfer learning model for precise breast cancer diagnosis utilizing pre-trained deep learning models and histopathology images

Thermal displacement prediction model with a structural optimized transfer learning technique

Thermal deformation of the spindle accounts for a large proportion of existing errors. After gathering data on thermal deformation through an experiment with a machine tool, AI algorithms were used in this study to predict the displacement of a cutting tool caused by heat deformation. Thermal displacement and temperature data were entered into models constructed using several machine learning algorithms. These models were then quantitatively evaluated in terms of their accuracy and compared to each other. Subsequently, transfer learning and hyperparameter tuning were conducted to produce a model with optimal prediction capability. The experimental results revealed that after machine learning models were trained using data collected on the first day of the experiments, their predictions based on data collected on the second day of the experiments were rife with severe prediction errors. This outcome indicated that experimental data gathered at different times weakened the models’ predictive abilities. Thus, to increase the prediction accuracy and prevent time from being wasted on repeated training, transfer learning were incorporated with model optimization. Finally, this approach achieved excellent R2 scores of 0.99941, 0.99964, and 0.99902 for the prediction of displacement in the x-, y-, and z-directions.

Autoencoder and restricted Boltzmann machine for transfer learning in functional magnetic resonance imaging task classification

Deep neural networks (DNNs) have been adopted widely as classifiers for functional magnetic resonance imaging (fMRI) data, advancing beyond traditional machine learning models. Consequently, transfer learning of the pre-trained DNN becomes crucial to enhance DNN classification performance, specifically by alleviating an overfitting issue that occurs when a substantial number of DNN parameters are fitted to a relatively small number of fMRI samples. In this study, we first systematically compared the two most popularly used, unsupervised pretraining models for resting-state fMRI (rfMRI) volume data to pre-train the DNNs, namely autoencoder (AE) and restricted Boltzmann machine (RBM). The group in-brain mask used when training AE and RBM displayed a sizable overlap ratio with Yeo's seven functional brain networks (FNs). The parcellated FNs obtained from the RBM were fine-grained compared to those from the AE. The pre-trained AE and RBM served as the weight parameters of the first of the two hidden DNN layers, and the DNN fulfilled the task classifier role for fMRI (tfMRI) data in the Human Connectome Project (HCP). We tested two transfer learning schemes: (1) fixing and (2) fine-tuning the DNN's pre-trained AE or RBM weights. The DNN with transfer learning was compared to a baseline DNN, trained using random initial weights. Overall, DNN classification performance from the transfer learning proved superior when the pre-trained RBM weights were fixed and when the pre-trained AE weights were fine-tuned (average error rates: 14.8% for fixed RBM, 15.1% fine-tuned AE, and 15.5% for the baseline model) compared to the alternative scenarios of DNN transfer learning schemes. Moreover, the optimal transfer learning scheme between the fixed RBM and fine-tuned AE varied according to seven task conditions in the HCP. Nonetheless, the computational load reduced substantially for the fixed-weight-based transfer learning compared to the fine-tuning-based transfer learning (e.g., the number of weight parameters for the fixed-weight-based DNN model reduced to 1.9% compared with a baseline/fine-tuned DNN model). Our findings suggest that weight initialization at the DNN's first layer using RBM-based pre-trained weights provides the most promising approach when the whole-brain fMRI volume supports associated task classification. We believe that our proposed scheme could be applied to a variety of task conditions to improve their classification performance and to utilize computational resources efficiently using our AE/RBM-based pre-trained weights compared to random initial weights for DNN training.

Ensemble classification based optimized transfer learning feature method for early stage diagnosis of diabetic retinopathy

Ensemble classification based optimized transfer learning feature method for early stage diagnosis of diabetic retinopathy

A Bionic Dynamic Path Planning Algorithm of the Micro UAV Based on the Fusion of Deep Neural Network Optimization/Filtering and Hawk-Eye Vision

A micro unmanned aerial vehicle (UAV) only equipped with a monocular camera is hard to accomplish a flying task with obstacles avoidance and target tracking simultaneously. In this article, a bionic dynamic path planning algorithm was developed for cooperation of obstacles avoidance and target tracking. An improved bat algorithm (BA) optimized transfer learning convolutional neural network (CNN) and bio-inspired optical flow balance algorithm was combined for obstacles avoidance. The Hawk-eye algorithm with line of sight (LOS) tracking rules is aimed at UAV dynamic tracking with obstacles avoidance. All of perception information, including avoidance and tracking were fused in UAV motion decision phase. The experiments include “obstacles avoidance” and “obstacles avoidance <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$+$</tex-math> </inline-formula> target tracking” parts. Comparing with manual control and other algorithms, the bionic dynamic path planning algorithm in this article showed certain advantages in success rate, less obstacles collisions, and less major accidents.

Optimized transfer learning for polyp detection

Early diagnosis of colorectal cancer focuses on detecting polyps in the colon as early as possible so that patients can have the best chances for success- ful treatment. This research presents the optimized parameters for polyp detection using a deep learning technique. Polyp and non-polyp images are trained on the InceptionResnetV2 model by the Faster Region Con- volutional Neural Networks (Faster R-CNN) framework to identify polyps within the colon images. The proposed method revealed more remarkable results than previous works, precision: 92.9 %, recall: 82.3%, F1-Measure: 87.3%, and F2-Measure: 54.6% on public ETIS-LARIB data set. This detection technique can reduce the chances of missing polyps during a pro- longed clinical inspection and can improve the chances of detecting multiple polyps in colon images.

An Optimized Transfer Learning Model Based Kidney Stone Classification

The kidney is an important organ of humans to purify the blood. The healthy function of the kidney is always essential to balance the salt, potassium and pH levels in the blood. Recently, the failure of kidneys happens easily to human beings due to their lifestyle, eating habits and diabetes diseases. Early prediction of kidney stones is compulsory for timely treatment. Image processing-based diagnosis approaches provide a greater success rate than other detection approaches. In this work, proposed a kidney stone classification method based on optimized Transfer Learning(TL). The Deep Convolutional Neural Network (DCNN) models of DenseNet169, MobileNetv2 and GoogleNet applied for classification. The combined classification results are processed by ensemble learning to increase classification performance. The hyperparameters of the DCNN model are adjusted by the metaheuristic algorithm of Gorilla Troops Optimizer (GTO). The proposed TL model outperforms in terms of all the parameters compared to other DCNN models.