Search

Brain imaging plays a crucial role in the diagnosis and treatment of various neurological disorders, providing valuable insights into the structure and function of the brain. Techniques such as magnetic resonance imaging (MRI) and computed tomography (CT) enable non-invasive visualization of the brain, aiding in the understanding of brain anatomy, abnormalities, and functional connectivity. However, cost and radiation dose may limit the acquisition of specific image modalities, so medical image synthesis can be used to generate required medical images without actual addition. CycleGAN and other GANs are valuable tools for generating synthetic images across various fields. In the medical domain, where obtaining labeled medical images is labor-intensive and expensive, addressing data scarcity is a major challenge. Recent studies propose using transfer learning to overcome this issue. This involves adapting pre-trained CycleGAN models, initially trained on non-medical data, to generate realistic medical images. In this work, transfer learning was applied to the task of MR-CT image translation and vice versa using 18 pre-trained non-medical models, and the models were fine-tuned to have the best result. The models' performance was evaluated using four widely used image quality metrics: Peak-signal-to-noise-ratio, Structural Similarity Index, Universal Quality Index, and Visual Information Fidelity. Quantitative evaluation and qualitative perceptual analysis by radiologists demonstrate the potential of transfer learning in medical imaging and the effectiveness of the generic pre-trained model. The results provide compelling evidence of the model's exceptional performance, which can be attributed to the high quality and similarity of the training images to actual human brain images. These results underscore the significance of carefully selecting appropriate and representative training images to optimize performance in brain image analysis tasks. • Pre-trained CycleGANs can effectively deal with limited labeled medical images. • Models performance were evaluated comprehensively with Peak SNR, Structural Similarity, UQI, and Visual Fidelity. • Models latent space was illustratively analyzed to explain class separability for paired image translation. • Improved results achieved when carefully selecting non-medical representative training datasets related to underlying physiology. [ABSTRACT FROM AUTHOR]

Code cloning is a common practice in software development, where developers reuse existing code to accelerate programming speed and enhance work efficiency. Existing clone-detection methods mainly focus on code clones within a single programming language. To address the challenge of code clone instances in cross-platform development, we propose a novel method called TCCCD, which stands for Triplet-Based Cross-Language Code Clone Detection. Our approach is based on machine learning and can accurately detect code clone instances between different programming languages. We used the pre-trained model UniXcoder to map programs written in different languages into the same vector space and learn their code representations. Then, we fine-tuned TCCCD using triplet learning to improve its effectiveness in cross-language clone detection. To assess the effectiveness of our proposed approach, we conducted thorough comparative experiments using the dataset provided by the paper titled CLCDSA (Cross Language Code Clone Detection using Syntactical Features and API Documentation). The experimental results demonstrated a significant improvement of our approach over the state-of-the-art baselines, with precision, recall, and F1-measure scores of 0.96, 0.91, and 0.93, respectively. In summary, we propose a novel cross-language code-clone-detection method called TCCCD. TCCCD leverages the pre-trained model UniXcode for source code representation and fine-tunes the model using triplet learning. In the experimental results, TCCCD outperformed the state-of-the-art baselines in terms of the precision, recall, and F1-measure.

Since the current remote sensing pre-trained models trained on optical images are not as effective when applied to SAR image tasks, it is crucial to create sensor-specific SAR models with generalized feature representations and to demonstrate with evidence the limitations of optical pre-trained models in downstream SAR tasks. The following aspects are the focus of this study: pre-training, fine-tuning, and explaining. First, we collect the current large-scale open-source SAR scene image classification datasets to pre-train a series of deep neural networks, including convolutional neural networks (CNNs) and vision transformers (ViT). A novel dynamic range adaptive enhancement method and a mini-batch class-balanced loss are proposed to tackle the challenges in SAR scene image classification. Second, the pre-trained models are transferred to various SAR downstream tasks compared with optical ones. Lastly, we propose a novel knowledge point interpretation method to reveal the benefits of the SAR pre-trained model with comprehensive and quantifiable explanations. This study is reproducible using open-source code and datasets, demonstrates generalization through extensive experiments on a variety of tasks, and is interpretable through qualitative and quantitative analyses. The codes and models are open source.

Personality is a collection of striking traits and behaviors of a person. The use of personality models can be applied in employee recruitment systems or to analyze characteristics and potential in more depth. Personality models are usually made using psychological test data or filling out questionnaires. However, this requires a long time. Building a personality classification model using NLP and deep learning is considered one of the best solutions. However, the performance of the classification model still needs to be improved, especially for Indonesian Language data. So, this research makes a personality classification model with Indonesian Language data using BERT-based architectures such as Multilingual BERT, IndoBERT, and Indonesian RoBERTa Base with an ensemble majority voting technique. Data limitations and imbalances were addressed using synonym replacement by incorporating words from a pre-trained model, MBERT. Information contained in social media often has ambiguous meanings because the words conveyed are not standardized, so this study tries to retain the information contained in the text by translating emoticons and slang words at the preprocessing stage to help keep the meaning of words in context. The proposed approach's research results can improve the classification model's results in classifying personality. [ABSTRACT FROM AUTHOR]

There are several disease kinds in global populations that may be related to human lifestyles, social, genetic, economic, and other factors related to the nature of the country they live in. Most of the recent studies have focused on investigating prevalent diseases that spread in the population in order to minimize mortality risks, choose the best method for treatment, and improve community healthcare. Kidney disease is one of the most widespread health problems in modern society. This study focuses on kidney stones, cysts, and tumors, the three most common types of renal illness, using a dataset of 12,446 CT urogram and whole abdomen images Available: https://www.kaggle.com/nazmul0087/ct-kidney-dataset -normalcyst-tumor-and-stone, aiming to move toward an AI-based kidney disease diagnosis system while contributing to the wider field of artificial intelligence research. In this study, a hybrid technique is used by utilizing both pre-trained models for feature extraction and classification using machine learning algorithms for the task of kidney disease image diagnosis. The pre-trained model used in this study is the Densenet-201 model. As well as using Random Forest for classification, the Densenet-201-Random-Forest approach has outperformed many of the previous models used in other studies, having an accuracy rate of 99.44 percent. [ABSTRACT FROM AUTHOR]

Digital humanities aims to use modern computer network technology to help traditional humanities research. Classical Chinese historical books are the important basis for historical research and learning, but since their writing language is classical Chinese, it is quite different from the vernacular Chinese in grammar and meaning, so it is not easy to read and understand. In view of the above problems, the solution to extract entities and relations in historical books based on pre-trained models was proposed to obtain the rich information contained in historical texts effectively. The model usedmulti-level pre-training tasks instead of BERT's original pre-training tasks to fully capture semantic information. And the model added some structures such as convolutional layers and sentence-level aggregations on the basis of the BERT model to optimize the generated word representation further. Then, in view of the scarcity of classical Chinese annotation data, a crowdsourcing system for the task of labeling historical classics was constructed, high-quality, large-scale entity and relation data was obtained and the classical Chinese knowledge extraction dataset was constructed. So it helped to evaluate the performance of the model and fine-tune the model. Experiments on the dataset constructed in this paper and on the GulianNER dataset demonstrated the effectiveness of the model proposed in this paper. [ABSTRACT FROM AUTHOR]

Glaucomatous optic neuropathy is the preeminent cause of incurable vision impairment and blindness across the world. Manual interpretation of the pathological structures in fundus images is time‐consuming and requires the expertise of a competent specialist. With the development of deep learning approaches, automated glaucoma diagnosis is easy and effective for larger screening. Convolutional neural networks, in particular, have emerged as a promising choice for glaucoma detection from fundus images due to their remarkable success in image classification. Transferring the optimized weights from a pre‐trained model expedites and simplifies the training process of deep neural network. In this paper, a deep ensemble model using the stacking ensemble learning technique is developed to attain the optimum performance for the classification of glaucomatous and normal images. Thirteen pre‐trained models such as Alexnet, Googlenet, VGG‐16, VGG‐19, Squeezenet, Resnet‐18, Resnet‐50, Resnet‐101, Efficientnet‐b0, Mobilenet‐v2, Densenet‐201, Inception‐v3, and Xception are implemented. Their performance is compared in 65 different configurations, comprising 13 CNN architectures and five various classification approaches. A two‐stage ensemble selection technique is proposed to select the optimal configurations. Selected configurations are pooled using a probability averaging technique. The final classification is performed using an SVM classifier. In this work, publicly available databases are modified (such as: DRISHTI‐GS1‐R, ORIGA‐R, RIM‐ONE2‐R, LAG‐R, and ACRIMA‐R) based on oversampling data‐level technique for validating the performance of deep ensemble model. Ensembling the best configurations reports an overall classification accuracy of 93.4%, 79.6%, 91.3%, 99.5%, and 99.6% in DRISHTI‐GS1‐R, ORIGA‐R, RIM‐ONE2‐R, ACRIMA‐R, and LAG‐R databases, respectively. [ABSTRACT FROM AUTHOR]

The Internet of Things is a paradigm that interconnects several smart devices through the internet to provide ubiquitous services to users. This paradigm and Web 2.0 platforms generate countless amounts of textual data. Thus, a significant challenge in this context is automatically performing text classification. State-of-the-art outcomes have recently been obtained by employing language models trained from scratch on corpora made up from news online to handle text classification better. A language model that we can highlight is BERT (Bidirectional Encoder Representations from Transformers) and also DistilBERT is a pre-trained smaller general-purpose language representation model. In this context, through a case study, we propose performing the text classification task with two previously mentioned models for two languages (English and Brazilian Portuguese) in different datasets. The results show that DistilBERT's training time for English and Brazilian Portuguese was about 45% faster than its larger counterpart, it was also 40% smaller, and preserves about 96% of language comprehension skills for balanced datasets. [ABSTRACT FROM AUTHOR]

BERT, a pre-trained language model on the large-scale corpus, has made breakthrough progress in NLP tasks. However, the experimental data shows that the BERT model's application effect in Chinese tasks is not ideal. The reason is that we believe that only character-level embedding can be obtained through BERT. However, a single Chinese character often cannot express their comprehensive meaning. To improve the model's ability to understand phrase-level semantic information, this paper proposes an enhanced BERT based on the average pooling(AP-BERT). Our model uses an average pooling layer to act on token embedding and reconstructs the model's input embedding, which can effectively improve BERT's application effect in Chinese natural language processing. Experimental data show that our proposed method has been enhanced in the four tasks of Chinese text classification, named entity recognition, reading comprehension, and summary generation. This method can not only improve the application effect of the BERT model in Chinese tasks but also can be well applied to other pre-trained language models. [ABSTRACT FROM AUTHOR]

The year 2020 and 2021 was the witness of Covid 19 and it was the leading cause of death throughout the world during this time period. It has an impact on a large geographic area, particularly in countries with a large population. Due to the fact that this novel coronavirus has been detected in all countries around the world, the World Health Organization (WHO) has declared Covid-19 to be a pandemic. This novel coronavirus spread quickly from person to person through the saliva droplets and direct or indirect contact with an infected person. The tests carried out to detect the Covid-19 are time-consuming and the primary cause of rapid growth in Covid19 cases. Early detection of Covid patient can play a significant role in controlling the Covid chain by isolation the patient and proper treatment at the right time. Recent research on Covid-19 claim that Chest CT and X-ray images can be used as the preliminary screening for Covid-19 detection. This paper suggested an Artificial Intelligence (AI) based approach for detecting Covid-19 by using X-ray and CT scan images. Due to the availability of the small Covid dataset, we are using a pre-trained model. In this paper, four pre-trained models named VGGNet-19, ResNet50, InceptionResNetV2 and MobileNet are trained to classify the X-ray images into the Covid and Normal classes. A model is tuned in such a way that a smaller percentage of Covid cases will be classified as Normal cases by employing normalization and regularization techniques. The updated binary cross entropy loss (BCEL) function imposes a large penalty for classifying any Covid class to Normal class. The experimental results reveal that the proposed InceptionResNetV2 model outperforms the other pre-trained model with training, validation and test accuracy of 99.2%, 98% and 97% respectively. [ABSTRACT FROM AUTHOR]

Cross-language sentence similarity computation is among the focuses of research in natural language processing (NLP). At present, some researchers have introduced fine-grained word and character features to help models understand sentence meanings, but they do not consider coarse-grained prior knowledge at the sentence level. Even if two cross-linguistic sentence pairs have the same meaning, the sentence representations extracted by the baseline approach may have language-specific biases. Considering the above problems, in this paper, we construct a Chinese–Uyghur cross-lingual sentence similarity dataset and propose a method to compute cross-lingual sentence similarity by fusing multiple features. The method is based on the cross-lingual pretraining model XLM-RoBERTa and assists the model in similarity calculation by introducing two coarse-grained prior knowledge features, i.e., sentence sentiment and length features. At the same time, to eliminate possible language-specific biases in the vectors, we whitened the sentence vectors of different languages to ensure that they were all represented under the standard orthogonal basis. Considering that the combination of different vectors has different effects on the final performance of the model, we introduce different vector features for comparison experiments based on the basic feature splicing method. The results show that the absolute value feature of the difference between two vectors can reflect the similarity of two sentences well. The final F1 value of our method reaches 98.97%, which is 19.81% higher than that of the baseline.

COVID-19 prevention procedures are executed to support public services and business continuity in a pandemic situation. Manual mask use monitoring is not efficient as it requires resources to monitor people at all times. Therefore, this task can be supported by automated surveillance systems based on Deep Learning. We performed mask detection and face recognition for a real-environment dataset. YOLOV3 as a one-stage detector was implemented to simultaneously generate a bounding box of the face area and class prediction. In face recognition, we compared the performance of three pre-trained models, namely ResNet152V2, InceptionV3, and Xception. The mask detection showed promising results with MAP=0.8960 on training and MAP=0.8957 on validation. We chose the Xception model for face recognition because it has equal quality as ResNet152V2 but has fewer parameters. Xception achieved a minimal loss value in the validation of 0.09157 with perfect accuracy on facial images larger than 100 pixels. Overall the system delivers promising results and can identify faces, even those behind the mask.

There are several disease kinds in global populations that may be related to human lifestyles, social, genetic, economic, and other factors related to the nature of the country they live in. Most of the recent studies have focused on investigating prevalent diseases that spread in the population in order to minimize mortality risks, choose the best method for treatment, and improve community healthcare. Kidney disease is one of the most widespread health problems in modern society. This study focuses on kidney stones, cysts, and tumors, the three most common types of renal illness, using a dataset of 12,446 CT urogram and whole abdomen images, aiming to move toward an AI-based kidney disease diagnosis system while contributing to the wider field of artificial intelligence research. In this study, a hybrid technique is used by utilizing both pre-train models for feature extraction and classification using machine learning algorithms for the task of kidney disease image diagnosis. The pre-trained model used in this study is the Densenet-201 model. As well as using Random Forest for classification, the Densenet-201-Random-Forest approach has outperformed many of the previous models used in other studies, having an accuracy rate of 99.719 percent.

Multimodal emotion classification (MEC) has been extensively studied in human–computer interaction, healthcare, and other domains. Previous MEC research has utilized identical multimodal annotations (IMAs) to train unimodal models, hindering the learning of effective unimodal representations due to differences between unimodal expressions and multimodal perceptions. Additionally, most MEC fusion techniques fail to consider the unimodal–multimodal inconsistencies. This study addresses two important issues in MEC: learning satisfactory unimodal representations of emotion and accounting for unimodal–multimodal inconsistencies during the fusion process. To tackle these challenges, the authors propose the Two-Stage Conformer-based MEC model (Uni2Mul) with two key innovations: (1) in stage one, unimodal models are trained using independent unimodal annotations (IUAs) to optimize unimodal emotion representations; (2) in stage two, a Conformer-based architecture is employed to fuse the unimodal representations learned in stage one and predict IMAs, accounting for unimodal–multimodal differences. The proposed model is evaluated on the CH-SIMS dataset. The experimental results demonstrate that Uni2Mul outperforms baseline models. This study makes two key contributions: (1) the use of IUAs improves unimodal learning; (2) the two-stage approach addresses unimodal–multimodal inconsistencies during Conformer-based fusion. Uni2Mul advances MEC by enhancing unimodal representation learning and Conformer-based fusion.

Capsule networks exhibit the potential to enhance computer vision tasks through their utilization of equivariance for capturing spatial relationships. However, the broader adoption of these networks has been impeded by the computational complexity of their routing mechanism and shallow backbone model. To address these challenges, this paper introduces an innovative hybrid architecture that seamlessly integrates a pretrained backbone model with a task-specific capsule head (CapsHead). Our methodology is extensively evaluated across a range of classification and segmentation tasks, encompassing diverse datasets. The empirical findings robustly underscore the efficacy and practical feasibility of our proposed approach in real-world vision applications. Notably, our approach yields substantial 3.45% and 6.24% enhancement in linear evaluation on the CIFAR10 dataset and segmentation on the VOC2012 dataset, respectively, compared to baselines that do not incorporate the capsule head. This research offers a noteworthy contribution by not only advancing the application of capsule networks, but also mitigating their computational complexities. The results substantiate the feasibility of our hybrid architecture, thereby paving the way for a wider integration of capsule networks into various computer vision tasks.

Machine reading comprehension (MRC) is an important research topic in the field of Natural Language Processing (NLP). However, traditional MRC models often face challenges of information loss, lack of capability to retain long-distance dependence, and inability to deal with unanswerable questions where answers are not available in given texts. In this paper, a Chinese reading comprehension algorithm, called the Attention and Conditional Random Filed (AT-CRF) Reader, is proposed to address the above challenges. Firstly, RoBERTa, a pre-trained language model, is introduced to obtain the embedding representations of input. Then, a depthwise separable convolution neural network and attention mechanisms are used to replace the recurrent neural network for encoding. Next, the attention flow and self-attention mechanisms are used to obtain the context–query internal relation. Finally, the conditional random field is used to handle unanswerable questions and predict the correct answer. Experiments were conducted on the two Chinese machine reading comprehension datasets, CMRC2018 and DuReader-checklist, and the results showed that, compared with the baseline model, the F1 scores achieved by our AT-CRF Reader model has improved by 2.65% and 2.68%, and the EM values increased by 4.45% and 3.88%. [ABSTRACT FROM AUTHOR]

Background: Skin cancer can quickly become fatal. An examination and biopsy of dermoscopic pictures are required to determine if skin cancer is malignant or benign. However, these examinations can be costly. Objective: In this research, we proposed deep learning (DL)-based approach to identify a melanoma, the most dangerous kind of skin cancer. DL is particularly excellent in learning traits and predicting cancer. However, DL requires a vast number of images. Method: We used image augmentation and transferring learning to categorise images into benign and malignant. We used the public ISIC 2020 database to train and test our models. The ISIC 2020 dataset classifies melanoma as malignant. Along with the categorization, the dataset was examined for variation. The training and validation accuracy of three of the best pre-trained models were compared. To minimise the loss, three optimizers were used: RMSProp, SGD, and ADAM. Results: We attained training accuracy of 98.73%, 99.12%, and 99.76% using ResNet, VGG16, and MobileNetV2, respectively. We achieved a validation accuracy of 98.39% using these three pre-trained models. Conclusion: The validation accuracy of 98.39% outperforms the prior pre-trained model. The findings of this study can be applied in medical science to help physicians diagnose skin cancer early and save lives. [ABSTRACT FROM AUTHOR]

Utilizing pre-trained models involves fully or partially using pre-trained parameters as initialization. In general, configuring a pre-trained model demands practitioners' knowledge about problems or an exhaustive trial–error experiment according to a given task. In this paper, we propose tuning trainable layers using a genetic algorithm on a pre-trained model that is fine-tuned on single-channel image datasets for a classification task. The single-channel dataset comprises images from grayscale and preprocessed audio signals transformed into a log-Mel spectrogram. Four deep-learning models used in the experimental evaluation employed the pre-trained model with the ImageNet dataset. The proposed genetic algorithm was applied to find the highest fitness for every generation to determine the selective layer tuning of the pre-trained models. Compared to the conventional fine-tuning method and random layer search, our proposed selective layer search with a genetic algorithm achieves higher accuracy, on average, by 9.7% and 1.88% (MNIST-Fashion), 1.31% and 1.14% (UrbanSound8k), and 2.2% and 0.29% (HospitalAlarmSound), respectively. In addition, our searching method can naturally be applied to various datasets of the same task without prior knowledge about the dataset of interest. [ABSTRACT FROM AUTHOR]

A plethora of software vulnerabilities are exposed daily, posing a severe threat to the Internet. It is almost impossible for security experts or software developers to deal with all vulnerabilities. Therefore, it is imperative to rapidly assess the severity of the vulnerability to be able to select which one should be given preferential attention. CVSS is now the industry's de facto evaluation standard, which is calculated with a quantitative formula to measure the severity of a vulnerability. The CVSS formula consists of several metrics related to the vulnerability's features. Security experts need to determine the values of each metric, which is tedious and time-consuming, therefore hindering the efficiency of severity assessment. To address this problem, in this paper, we propose a method based on a pre-trained model for the prediction of CVSS metric values. More specifically, this method utilizes the XLNet model that is fine-tuned with a self-built corpus to predict the metric values from the vulnerability description text, thus reducing the burden of the assessment procedure. To verify the performance of our method, we compare the XLNet model with other pre-trained models and conventional machine learning techniques. The experimental results show that the method outperforms these models on evaluation metrics, reaching state-of-the-art performance levels. [ABSTRACT FROM AUTHOR]

Protein–protein interaction (PPI) plays an important role in biological processes such as signal transduction, immune response and membrane transport, etc. In this work, a protein sequence-based machine learning model, SeBPPI, to predict protein–protein binding was proposed. In this model, the descriptors were generated from three pre-trained models: Unirep, ESM and TAPE. The performance of SeBPPI with these pre-trained models was evaluated on several different test datasets. The accuracy of our binary prediction model shows improvement over the existing methods. We also compared the performance of two classification heads: The Recurrent convolution neural network (RCNN) and the fully connected neural network (FNN) and found that the use of RCNN is beneficial for the overall improvement in the accuracy of the model. This study helps to improve the accuracy in sequence-based protein–protein binding predictions. The model used in this work is integrated in the web server http://www.icdrug.com/ICDrug/SeBPPI. In this paper, a sequence-based protein-protein binding predictor was proposed. This machine learning model combined the RCNN model and the pre-trained models. Comparison between other machine learning models illustrated the SeBPPI model was a powerful tool. [ABSTRACT FROM AUTHOR]

Traditional grain size determination in materials characterization involves microscopy images and a laborious process requiring significant manual input and human expertise. In recent years, the development of computer vision (CV) has provided an alternative approach to microstructural characterization with preliminary implementations greatly simplifying the grain size determination process. Here, an end-to-end workflow to measure grain size in microscopy images without any manual input is presented. Following the ASTM standards for grain size determination, results from the line intercept (Heyn's method) and planimetric (Saltykov's method) approaches are used as the baseline. A pre-trained holistically nested edge detection (HED) model is used for CV-based edge detection, and the results are further compared to the classic Canny edge detection method. Post-processing was performed using open-source image processing packages to extract the grain size. In optical microscope images, the pre-trained HED model achieves much higher accuracy than the Canny edge detection method while reducing the image processing time by one to two orders of magnitude compared to traditional methods. The effects of morphological operations on the predicted grain size accuracy are also explored. Overall, the proposed end-to-end convolutional neural network (CNN)-based workflow can significantly reduce the processing time while maintaining the same accuracy as the traditional manual method. [ABSTRACT FROM AUTHOR]

Gait is a pattern of a person's walking. The body movements of a person while walking makes the gait unique. Regardless of the uniqueness, the gait recognition process suffers under various factors, namely the viewing angle, carrying condition, and clothing. In this paper, a pre-trained VGG-16 model is incorporated with a multilayer perceptron to enhance the performance under various covariates. At first, the gait energy image is obtained by averaging the silhouettes over a gait cycle. Transfer learning and fine-tuning techniques are then applied on the pre-trained VGG-16 model to learn the gait features of the attained gait energy image. Subsequently, a multilayer perceptron is utilized to determine the relationship among the gait features and the corresponding subject. Lastly, the classification layer identifies the corresponding subject. Experiments are conducted to evaluate the performance of the proposed method on the CASIA-B dataset, the OU-ISIR dataset D, and the OU-ISIR large population dataset. The comparison with the state-of-the-art methods shows that the proposed method outperforms the methods on all the datasets. [ABSTRACT FROM AUTHOR]

The last decade has seen increased interest in environmental sound classification (ESC) due to the increased complexity and rich information of ambient sounds. The state-of-the-art methods for ESC are based on transfer learning paradigms that often utilize learned representations from common image-classification problems. This paper aims to determine the effectiveness of employing pre-trained convolutional neural networks (CNNs) for audio categorization and the feasibility of retraining. This study investigated various hyper-parameters and optimizers, such as optimal learning rate, epochs, and Adam, Adamax, and RMSprop optimizers for several pre-trained models, such as Inception, and VGG, ResNet, etc. Firstly, the raw sound signals were transferred into an image format (log-Mel spectrogram). Then, the selected pre-trained models were applied to the obtained spectrogram data. In addition, the effect of essential retraining factors on classification accuracy and processing time was investigated during CNN training. Various optimizers (such as Adam, Adamax, and RMSprop) and hyperparameters were utilized for evaluating the proposed method on the publicly accessible sound dataset UrbanSound8K. The proposed method achieves 97.25% and 95.5% accuracy on the provided dataset using the pre-trained DenseNet201 and the ResNet50V2 CNN models, respectively. [ABSTRACT FROM AUTHOR]

Deep learning has broad applications in various fields.However,users still need to face problems from two aspects when applying deep learning.First,deep learning has a complex theoretical background,non-professional users lack background knowledge in modeling and tuning.It is difficult for them to build performance-optimized models.Second,modules such as data preprocessing,model training,and prediction often involve more complicated programming implementations,which bring some difficulties in getting started for non-professional users who have no programming skill background.In view of the above two issues of usability,this paper proposes an easy-to-use graphical deep learning model construction system,DragDL.The purpose of DragDL is to reduce the difficulty of data preprocessing,model training,monitoring,online prediction and other tasks for users.The system is based on the PaddlePaddle framework and supports building a deep learning network structure on the canvas by dragging graphical operators,supporting inference and prediction functions,and abstracting the data preprocessing operation process into a dataflow graph,which is convenient for users to understand and debug.The system also provides visualization functions for performance monitoring during the training process.At the same time,DragDL provides a classic model library,which allows users to build new DL network by tuning the existing classic model network.DragDL is deployed based on a centralized server and Web client.The server provides a virtual machine service for submitted tasks and supports large-scale asynchronous task scheduling to have concurrent processing capabilities.

Text matching is one of the key technologies in the retrieval system.Aiming at the problem that the existing text ma-tching models can't capture the semantic differences of texts accurately,this paper proposes a text matching method based on fine-grained difference features.Firstly,the pre-trained model is used as the basic model to extract the matching text semantics and preliminarily match them.Then,the idea of adversarial learning is introduced in the embedding layer,and by constructing the virtual confrontation samples artificially for training,the learning ability and generalization ability of the model are improved.Finally,by introducing the fine-grained difference feature of the text to correct the preliminary prediction results of the text ma-tching,the capture ability of the model for fine-grained difference features is effectively improved,and then the performance of the text matching model is improved.In this paper,two datasets are tested,and the experiment on LCQMC dataset shows that the performance index of ACC is 88.96%,which is better than the best known model.

Context: Recently, many studies have been proposed to address the threat posed by Android malware. However, the continuous evolution of malware poses challenges to the task of representing application features in current detection methods. Objective: This paper introduces a novel Android malware detection approach based on the source code and binary code of software by leveraging large pre-trained models with a fine-grained multimodal fusion strategy. Method: Specifically, the approach treats the source code and binary code as the programming language modality (PM) and machine language modality (MM), respectively. Then, domain-specific knowledge (sensitive API) combined with large pre-trained model is further applied to extract PM features; while the binary code is transformed into RGB images, from which MM features are extracted using a pre-trained image processing model. Furthermore, a fine-grained fusion strategy is implemented using a multi-head self-attention mechanism to effectively capture the correlations among features across different modalities and generate comprehensive features for application malware detection. Results and Conclusion: The detection performance and generalization ability of the proposed method were validated on two experimental datasets. The results demonstrate that our method can accurately distinguish malware, achieving an accuracy of 98.28% and an F1-score of 98.66%. Additionally, it performs well on unseen data, with an accuracy of 92.86% and an F1-score of 94.49%. Meanwhile, ablation experiments confirm the contributions of sensitive API knowledge and the fine-grained multimodal fusion strategy to the success of malware detection. • The study detects Android malware by fusing information in source and binary code. • Large pre-trained model is used with task-specific knowledge for feature analysis. • A fine-grained feature fusion strategy is designed for describing software. [ABSTRACT FROM AUTHOR]

Future capacity prediction of lithium-ion batteries is a highly researched topic in the field of battery management systems, owing to the gradual degradation of battery capacity over time due to various factors such as chemical changes within the battery, usage patterns, and operating conditions. The accurate prediction of battery capacity can aid in optimizing its usage, extending its lifespan, and mitigating the risk of unforeseen failures. In this paper, we proposed a novel fine-tuning model based on a deep learning model with a transfer learning approach comprising of two key components: offline training and online prediction. Model weights and prediction parameters were transferred from offline training using source data to the online prediction stage. The transferred Bi-directional Long Short-Term Memory with an Attention Mechanism model weights and prediction parameters were utilized to fine-tune the model by partial target data in the online prediction phase. Three battery batches with different charging policy were used to evaluate the proposed approach’s robustness, reliability, usability, and accuracy for the three charging policy batteries’ real-world data. The experiment results show that the proposed method’s efficacy improved, with an increase in the cycle number of the starting point, exhibiting a linear relationship with the starting point. The proposed method yields relative error values of 8.70%, 6.38%, 9.52%, 7.58%, 1.94%, and 2.29%, respectively, for the six target batteries in online prediction. Thus, the proposed method is effective in predicting the future capacity of lithium-ion batteries and holds potential for use in predictive maintenance applications.

This study presents a novel method for assessing landslide hazards along highways using remote sensing and machine learning. We extract geospatial features such as slope, aspect, and rainfall over Guangxi, China, and apply an extreme gradient boosting model pre-trained on contiguous United States datasets. The model produces susceptibility maps that indicate landslide probability at different scales. However, the lack of accurate data on historical landslides in Guangxi challenges the model evaluation and comparison between regions. To overcome this, we calibrate the model to fit the local conditions in Guangxi. The calibrated model agrees with the observed landslide locations, implying its capability to capture regional variations in landslide mechanisms. We apply the model at a 30 m resolution along the Heba Expressway and validate it against reports from July 2021 to March 2022. The model correctly predicts five of seven landslide events in this period with a reasonable alarm rate. This framework has the potential for large-scale landslide risk management by informing transportation planning and infrastructure maintenance decisions. More data on landslide timing and human disturbance events may improve the model’s accuracy across diverse geographical areas and terrains.

Facial expression recognition (FER) is crucial in enhancing human-computer interaction. While current FER methods, leveraging various open-source deep learning models and training techniques, have shown promising accuracy and generalizability, their efficacy often diminishes in real-world scenarios that are not extensively studied. Addressing this gap, we introduce a novel in-the-wild balanced testing facial expression dataset designed for cross-domain validation, called BTFER. We rigorously evaluated widely utilized networks and self-designed architectures, adhering to a standardized protocol. Additionally, we explored different configurations, including input resolutions, class balance management, and pre-trained strategies, to ascertain their impact on performance. Through comprehensive testing across three major FER datasets and our in-depth cross-validation, we have ranked these network architectures and formulated a series of practical guidelines for implementing deep learning-based FER solutions in real-life applications. This paper also delves into the ethical considerations, privacy concerns, and regulatory aspects relevant to the deployment of FER technologies in sectors such as marketing, education, entertainment, and healthcare, aiming to foster responsible and effective use. The BTFER dataset and the implementation code are available in Kaggle and Github , respectively. [ABSTRACT FROM AUTHOR]

Within the realm of mathematics education, there exist several challenging supervised tasks that educators and researchers encounter, such as question difficulty prediction and mathematical expression understanding. To address these challenges, researchers have introduced unsupervised pre-trained models specifically tailored for mathematics education, yielding promising outcomes. However, the existing literature fails to consider the domain-specific characteristics of mathematics, particularly the structural features in pre-trained corpora and extensive expressions, which makes them costly expensive and time-consuming. To tackle this problem, we propose a lightweight expression-enhanced large-scale pre-trained language model, called EBERT, for mathematics education. Specifically, we select a large number of expression-enriched exercises to further pre-train the original BERT. To depict the inherent structural features existed in expressions, the initial step involves the creation of an Operator Tree for each expression. Subsequently, each exercise is transformed into a corresponding Question&Answer tree (QAT) to serve as the model input. Notably, to ensure the preservation of semantic integrity within the QAT, a specialized Expression Enhanced Matrix is devised to confine the visibility of individual tokens. Additionally, a new pre-training task, referred to as Question&Answer Matching , is introduced to capture exercise-related structural information at the semantic level. Through three downstream tasks in mathematical education, we prove that EBERT outperforms several state-of-the-art baselines (such as MathBERT and GPT-3) in terms of ACC and F1-score. [ABSTRACT FROM AUTHOR]

As social media has gradually become an indispensable part of people's life, more and more users begin to express their opinions on social media. These opinions contain rich emotional information, as well as many abnormal false emotion. And the opinion target rich in false emotion will have great potential value in the field of movie box office prediction, public opinion guidance and so on. Therefore, we propose a false emotion opinion target extraction task and two stage BERT+CRF model based on the concept of false emotion. Firstly, two stage BERT network includes C-BERT and W-BERT. We use C-BERT to learn the users comment feature, and W-BERT to learn the background content features of microblog. Then, we design five feature interaction fusion methods to fuse the features learned by C-BERT and W-BERT. At last, we transform the false emotion opinion target extraction task into a sequence labeling task, and use the CRF method to learn the label dependency of the context and output the optimal sequence among all candidate sequences. Due to the lack of false emotion opinion target datasets, we collate a false emotion opinion target dataset Weibo23. Comparing with the advanced models used in related research, including some sequence labeling models and named entity recognition models, two stage BERT+CRF model achieves an F1 score of 0.8834, which is about 10%–20% higher. [ABSTRACT FROM AUTHOR]

Idioms are a unique class of words in the Chinese language that can be challenging for Chinese machine reading comprehension due to their formal simplicity and the potential mismatch between their literal and figurative meanings. To address this issue, this paper adopted the “2 + 2” structure as the representation model for idiom structure feature extraction. According to the linguistic theory of idioms, to enhance the model’s learning ability for idiom semantics, we propose a two-stage semantic expansion method that leverages semantic knowledge during the pre-training stage and extracts idiom interpretation information during the fine-tuning stage to improve the model’s understanding of idioms. Moreover, with the consideration of the inferential interaction between global and local information during context fusion, which is neglected by all current works, we propose a method that utilizes a multi-headed attention mechanism to fully extract global and local information by selecting three attention patterns in the fine-tuning stage. This enables the fusion of semantic information extracted by the model at different granularities, leading to improved accuracy. The experimental results demonstrated that our proposed BERT-IDM model outperformed the baseline BERT model by achieving a 4.1% accuracy improvement.