Enhancing continuous speech recognition with CapsNet and WaveRNN: a transfer learning approach

In this study, an automated panicle blast grading methodology using support vector machine based on deep features of small convolutional neural network (CNN) is suggested. MobileNetV2 and ShuffleNet are the two small and powerful architecture among CNN. The transfer learning and deep learning approach are adapted for MobileNetV2 and ShuffleNet for classifying the panicle into three levels; healthy, less blast and high blast. The experimental results reveal the support vector machine (SVM) using deep features perform better compared to its counterpart i.e., transfer learning approach. In addition, the comparative analysis based on the accuracy score of MobileNetV2 & ShuffleNet in transfer learning & deep learning approach with the traditional image classification methods for panicle blast grading is performed. In over all, the ShuffleNet plus SVM is the best classification model for grading of panicle blast with accuracy of 89.37%, sensitivity 89.37%, specificity 94.68% and computational time 14.936 second.

PurposeThis paper aims to realize an in-situ quality inspection system rapidly for new injection molding (IM) tasks via transfer learning (TL) approach and automation technology.Design/methodology/approachThe proposed in-situ quality inspection system consists of an injection machine, USB camera, programmable logic controller and personal computer, interconnected via OPC or USB communication interfaces. This configuration enables seamless automation of the IM process, real-time quality inspection and automated decision-making. In addition, a MobileNet-based deep learning (DL) model is proposed for quality inspection of injection parts, fine-tuned using the TL approach.FindingsUsing the TL approach, the MobileNet-based DL model demonstrates exceptional performance, achieving validation accuracy of 99.1% with the utilization of merely 50 images per category. Its detection speed and accuracy surpass those of DenseNet121-based, VGG16-based, ResNet50-based and Xception-based convolutional neural networks. Further evaluation using a random data set of 120 images, as assessed through the confusion matrix, attests to an accuracy rate of 96.67%.Originality/valueThe proposed MobileNet-based DL model achieves higher accuracy with less resource consumption using the TL approach. It is integrated with automation technologies to build the in-situ quality inspection system of injection parts, which improves the cost-efficiency by facilitating the acquisition and labeling of task-specific images, enabling automatic defect detection and decision-making online, thus holding profound significance for the IM industry and its pursuit of enhanced quality inspection measures.

Crack inspections of automotive engine components are usually conducted manually; this is often tedious, with a high degree of subjectivity and cost. Therefore, establishing a robust and efficient method will improve the accuracy and minimize the subjectivity of the inspection. This paper presents a robust approach towards crack classification, using transfer learning and fine-tuning to train a pre-trained ConvNet model. Two deep convolutional neural network (DCNN) approaches to training a crack classifier—namely, via (1) a Light ConvNet architecture from scratch, and (2) fined-tuned and transfer learning top layers of the ConvNet architectures of AlexNet, InceptionV3, and MobileNet—are investigated. Data augmentation was utilized to minimize over-fitting caused by an imbalanced and inadequate training sample. Data augmentation improved the accuracy index by 4%, 5%, 7%, and 4%, respectively, for the proposed four approaches. The transfer learning and fine-tuning approach achieved better recall and precision scores. The transfer learning approach using the fine-tuned features of MobileNet attained better classification accuracy and is thus proposed for the training of crack classifiers. Moreover, we employed an up-to-date YOLOv5s object detector with transfer learning to detect the crack region. We obtained a mean average precision (mAP) of 91.20% on the validation set, indicating that the model effectively distinguished diverse engine part cracks.

Various computer systems have been proposed to classify vehicles according to several criteria (category, brand, model). Unfortunately, there is not much research on the classification of views, especially front and rear views. Several factors make this classification very difficult including similarity in shape, size, and color. This work aims to classify front and rear views of vehicles using the Transfer Learning (TL) approach. Here, we used a pre-trained CNN (AlexNet) that has been trained on more than a million images and can classify images into 1000 object categories. Thus, we transferred its learned knowledge and applied it to our new task (Classifying vehicle views). We conducted then two experiments. The first experiment has two scenarios: the first scenario is devoted to Transfer Learning using the AlexNet model, and the second scenario aims to build a network from scratch inspired from AlexNet. Experimental results reveal that the Transfer Learning approach gives high results. On the other hand, in the second experiment, we decided to use TL-AlexNet to extract features and train them with an SVM classifier instead of fully connected layers. And also, we combined the SVM with the fully connected layers. The accuracy rates have been improved after this experiment.

One of the serious challenges in machine learning and pattern recognition is to transfer knowledge from related but different domains to a new unlabeled domain. Feature selection with maximum mean discrepancy (f-MMD) is a novel and effective approach to transfer knowledge from source domain (training set) into target domain (test set) where training and test sets are drawn from different distributions. However, f-MMD has serious challenges in facing datasets with large number of samples and features. Moreover, f-MMD ignores the feature-label relation in finding the reduced representation of dataset. In this paper, we exploit jointly transfer learning and class discrimination to cope with domain shift problem on which the distribution difference is considerably large. We therefore put forward a novel transfer learning and class discrimination approach, referred to as RandOm k-samplesets feature Weighting Approach (ROWA). Specifically, ROWA reduces the distribution difference across domains in an unsupervised manner where no label is available in the test set. Moreover, ROWA exploits feature-label relation to separate various classes alongside the domain transfer, and augments the relation of selected features and source domain labels. In this work, we employ disjoint/overlapping small-sized samplesets to iteratively converge to final solution. Employment of local sets along with a novel optimization problem constructs a robust and effective reduced representation for adaptation across domains. Extensive experiments on real and synthetic datasets verify that ROWA can significantly outperform state-of-the-art transfer learning approaches.

Recognition of emotional states using frequency effective connectivity maps through transfer learning approach from electroencephalogram signals

Food security completely relies on agriculture at the global level. Diseases of plants are a threat to crop yields which ultimately leads to economic loss and food insufficiency. With the boom of the technological revolution, traditional methods to detect diseases were left behind. Techniques from AI, computer vision, image processing, machine learning (ML), and deep learning (DL) were applied to automate processes and provide solutions quickly and labor-free. Transfer learning comes under the domain of machine learning, and can leverage pre-trained models on huge datasets, and acclimate them to certain tasks (detect plant disease). This research investigates the implications of the transfer learning (TL) approach in detecting plant diseases and getting better accuracy on larger datasets. The verdicts of this study have the potential to decrease the time and computational cost associated with model training and get more efficiency. This research used a dataset available on Kaggle named “plant disease recognition dataset”.

A long clinical testing period is one of the key elements for the COVID-19 pandemic’s fast spread. Controlling the spread of COVID-19 requires early detection and diagnosis. Chest X-ray (CXR), for example, is an imaging technology that helps to speed up the identifying procedure of COVID-19 in patients. As a result, our goal is to create an automatic CAD system that can recognize COVID-19 samples from healthy people and COVID patients using CT scans. We used transfer learning (TL) approach, i.e., modified Visual Geometry Group (VGG19) and compared our proposed system results with other machine learning (ML) and deep learning (DL) approaches in order to discover the best one for this job. The proposed technique and various DL and ML models are tested using the COVID-CT dataset, where 80% of images are utilized for training and 20% for testing purpose. Our proposed TL technique achieves 97.83% classification accuracy with average precision, recall, and F1-score of 98.33, 97.67, and 97.67, respectively.KeywordsCOVID-19Deep learningMachine learningTransfer learning

This paper proposes a novel change-detection-driven transfer learning (TL) approach to update land-cover maps by classifying remote-sensing images acquired on the same area at different times (i.e., image time series). The proposed approach requires that a reliable training set is available only for one of the images (i.e., the source domain) in the time series whereas it is not for another image to be classified (i.e., the target domain). Unlike other literature TL methods, no additional assumptions on either the similarity between class distributions or the presence of the same set of land-cover classes in the two domains are required. The proposed method aims at defining a reliable training set for the target domain, taking advantage of the already available knowledge on the source domain. This is done by applying an unsupervised-change-detection method to target and source domains and transferring class labels of detected unchanged training samples from the source to the target dom...

Online social networks have become extremely popular with the ever-increasing reachability of internet to the common person. There are millions of tweets, Facebook messages, and product reviews posted every day. Such huge amount of data presents an opportunity to analyze the sentiment of masses in order to facilitate the decision making for the betterment of society. Sentiment analysis is the research area that quantitates the opinions expressed in natural language. It is a combination of various research fields such as text mining, natural language processing, artificial intelligence, statistics. The application of supervised machine learning algorithms is limited due to the unavailability of labeled data whereas the unsupervised or lexicon-based methodologies show weak performance. This scenario sets the stage for transfer learning or cross-domain learning approaches where the knowledge is learned from the source domain which is then applied to the target domain. The proposed approach computes the feature weights by the application of cosine similarity measure to SentiWordNet and generates revised sentiment scores. Model learning is performed by support vector machine using two experimental settings, i.e., single source and multiple target domains and multiple source and single target domains (MSST). Nine benchmark datasets have been employed for performance evaluation. Best performance was obtained using the MSST settings with 85.05% accuracy, 85.01% precision, 85.10% recall, and 85.05% F-measure. State-of-the-art performance comparison proved that the cosine similarity-based transfer learning approach outperforms other approaches.

Long-term climate change may strongly affect the aquatic environment in mid-latitude water resources. In particular, it can be demonstrated that temporal variations in surface water temperature in a reservoir have strong responses to air temperature. We adopted deep neural networks (DNNs) to understand the long-term relationships between air temperature and surface water temperature, because DNNs can easily deal with nonlinear data, including uncertainties, that are obtained in complicated climate and aquatic systems. In general, DNNs cannot appropriately predict unexperienced data (i.e., out-of-range training data), such as future water temperature. To improve this limitation, our idea is to introduce a transfer learning (TL) approach. The observed data were used to train a DNN-based model. Continuous data (i.e., air temperature) ranging over 150 years to pre-training to climate change, which were obtained from climate models and include a downscaling model, were used to predict past and future surface water temperatures in the reservoir. The results showed that the DNN-based model with the TL approach was able to approximately predict based on the difference between past and future air temperatures. The model suggested that the occurrences in the highest water temperature increased, and the occurrences in the lowest water temperature decreased in the future predictions.

Purpose Diabetic retinopathy (DR) is one of the dangerous complications of diabetes. Its grade level must be tracked to manage its progress and to start the appropriate decision for treatment in time. Effective automated methods for the detection of DR and the classification of its severity stage are necessary to reduce the burden on ophthalmologists and diagnostic contradictions among manual readers. Design/methodology/approach In this research, convolutional neural network (CNN) was used based on colored retinal fundus images for the detection of DR and classification of its stages. CNN can recognize sophisticated features on the retina and provides an automatic diagnosis. The pre-trained VGG-16 CNN model was applied using a transfer learning (TL) approach to utilize the already learned parameters in the detection. Findings By conducting different experiments set up with different severity groupings, the achieved results are promising. The best-achieved accuracies for 2-class, 3-class, 4-class and 5-class classifications are 86.5, 80.5, 63.5 and 73.7, respectively. Originality/value In this research, VGG-16 was used to detect and classify DR stages using the TL approach. Different combinations of classes were used in the classification of DR severity stages to illustrate the ability of the model to differentiate between the classes and verify the effect of these changes on the performance of the model.

Maturity status classification of papaya fruits based on machine learning and transfer learning approach

Children's speech processing is more challenging than that of adults due to lacking of large scale children's speech corpora. With the developing of the physical speech organ, high inter speaker and intra speaker variabilities are observed in children's speech. On the other hand, data collection on children is difficult as children usually have short attention span and their language proficiency is limited. In this paper, we propose to improve children's automatic speech recognition performance with transfer learning technique. We compare two transfer learning approaches in enhancing children's speech recognition performance with adults' data. The first method is to perform acoustic model adaptation on the pre-trained adult model. The second is to train acoustic model with deep neural network based multi-task learning approach: the adults' and children's acoustic characteristics are learnt jointly in the shared hidden layers, while the output layers are optimized with different speaker groups. Our experiment results show that both transfer learning approaches are effective in transferring rich phonetic and acoustic information from adults' model to children model. The multi-task learning approach outperforms the acoustic adaptation approach. We further show that the speakers' acoustic characteristics in languages can also benefit the target language under the multi-task learning framework.

Sign languages are complex, but there are ongoing research efforts in engineering and data science to recognize, understand, and utilize them in real-time applications. Arabic sign language recognition (ArSL) has been examined and applied using various traditional and intelligent methods. However, there have been limited attempts to enhance this process by utilizing pretrained models and large-sized vision transformers designed for image classification tasks. This study aimed to create robust transfer learning models trained on a dataset of 54,049 images depicting 32 alphabets from an ArSL dataset. The goal was to accurately classify these images into their corresponding Arabic alphabets. This study included two methodological parts. The first one was the transfer learning approach, wherein we utilized various pretrained models namely MobileNet, Xception, Inception, InceptionResNet, DenseNet, and BiT, and two vision transformers namely ViT, and Swin. We evaluated different variants from base-sized to large-sized pretrained models and vision transformers with weights initialized from the ImageNet dataset or otherwise randomly. The second part was the deep learning approach using convolutional neural networks (CNNs), wherein several CNN architectures were trained from scratch to be compared with the transfer learning approach. The proposed methods were evaluated using the accuracy, AUC, precision, recall, F1 and loss metrics. The transfer learning approach consistently performed well on the ArSL dataset and outperformed other CNN models. ResNet and InceptionResNet obtained a comparably high performance of 98%. By combining the concepts of transformer-based architecture and pretraining, ViT and Swin leveraged the strengths of both architectures and reduced the number of parameters required for training, making them more efficient and stable than other models and existing studies for ArSL classification. This demonstrates the effectiveness and robustness of using transfer learning with vision transformers for sign language recognition for other low-resourced languages.