Search

Tobacco is a valuable crop, but its disease identification is rarely involved in existing works. In this work, we use few-shot learning (FSL) to identify abnormalities in tobacco. FSL is a solution for the data deficiency that has been an obstacle to using deep learning. However, weak feature representation caused by limited data is still a challenging issue in FSL. The weak feature representation leads to weak generalization and troubles in cross-domain. In this work, we propose a feature representation enhancement network (FREN) that enhances the feature representation through instance embedding and task adaptation. For instance embedding, global max pooling, and global average pooling are used together for adding more features, and Gaussian-like calibration is used for normalizing the feature distribution. For task adaptation, self-attention is adopted for task contextualization. Given the absence of publicly available data on tobacco, we created a tobacco leaf abnormality dataset (TLA), which includes 16 categories, two settings, and 1,430 images in total. In experiments, we use PlantVillage, which is the benchmark dataset for plant disease identification, to validate the superiority of FREN first. Subsequently, we use the proposed method and TLA to analyze and discuss the abnormality identification of tobacco. For the multi-symptom diseases that always have low accuracy, we propose a solution by dividing the samples into subcategories created by symptom. For the 10 categories of tomato in PlantVillage, the accuracy achieves 66.04% in 5-way, 1-shot tasks. For the two settings of the tobacco leaf abnormality dataset, the accuracies were achieved at 45.5% and 56.5%. By using the multisymptom solution, the best accuracy can be lifted to 60.7% in 16-way, 1-shot tasks and achieved at 81.8% in 16-way, 10-shot tasks. The results show that our method improves the performance greatly by enhancing feature representation, especially for tasks that contain categories with high similarity. The desensitization of data when crossing domains also validates that the FREN has a strong generalization ability.

Depression, a prevalent adult symptom, can arise from various sources, including mental health conditions and social interactions. With the rise of social media, adults often share their daily experiences, potentially revealing their emotional state on social platforms, like X (formerly Twitter) and Facebook. In this study, we present Ensemble (E) of Convolutional Neural Network (C), Attention-based Long Short-Term Memory (L) Network, and Support Vector Machine (S) (E-CLS), utilizing Term Frequency-Inverse Document Frequency (TF-IDF) vectors, Global Vectors for Word Representation (GloVe) and Bidirectional Encoder Representations from Transformers (BERT) word embeddings. This model effectively identifies depressive posts. Validated with a Twitter-derived depressive dataset, E-CLS achieves an impressive F 1 -score of 0.91, surpassing existing machine-learning and deep-learning models by 2%. This research advances the detection of depression in social media posts, holding promise for enhanced mental health monitoring. Furthermore, our work contributes to the burgeoning field of mental health informatics by leveraging state-of-the-art techniques in natural language processing. The ensemble approach synergizes the strengths of Convolutional Neural Network (CNN) for local pattern recognition, Long Short-Term Memory (LSTM) Network for sequential context understanding, and Support Vector Machine (SVM) for robust classification. The incorporation of TF-IDF vectors and GloVe embeddings enriches feature representation, enhancing the model's ability to discern nuanced linguistic cues associated with depression. By demonstrating superior performance over established models, E-CLS showcases its potential as a valuable tool in digital mental health interventions. [ABSTRACT FROM AUTHOR]

The field of natural language processing (NLP) plays a pivotal role in discerning unstructured data from diverse origins. This study employs advanced techniques rooted in machine learning and deep learning to effectively categorize news articles. Notably, deep learning models have demonstrated superior performance over traditional machine learning algorithms, rendering them a popular choice for a range of NLP tasks. The research employs feature extraction techniques to identify multiword tokens, negation words, and out-of-vocabulary words and replace them. Additionally, convolutional neural network models leverage embedding, convolutional layers, and max pooling layers to capture intricate features. For tasks requiring an understanding of dependencies among long phrases, long short-term memory models come into play. The evaluation of the proposed model hinges on training it with datasets like AG News, BBC, and 20 Newsgroup, gauging its efficacy. The study delves into the myriad challenges inherent to text classification. These challenges are thoughtfully discussed, shedding light on the intricacies of the process. Furthermore, the research furnishes comprehensive test outcomes for both conventional machine learning and deep learning models. The significance of this proposed model is that it uses a multiword expression lexicon, wordnet synset, and word embedding techniques for feature extraction. The performance of the models is increased when using these feature extraction techniques. [ABSTRACT FROM AUTHOR]

The ability to extract high‐quality features from data is critical for machine learning applications. With the development of deep learning, various methods have been developed for image feature extraction, and unsupervised techniques have gained popularity due to their ability to operate without response variables. Autoencoders with encoder–decoder architectures are a common example of such techniques, but they are limited by a lack of proportional relationship between model reconstruction and encoder feature extraction performance. If the decoder is composed of multiple layers and mapping to a higher dimension is easier, the feature extraction performance of the encoder is likely to decrease. However, previous research has not adequately addressed this limitation. This study identifies the limitations of conventional unsupervised feature extraction techniques that utilize the encoder–decoder architecture, and proposes a novel feature extraction technique called SAFE, which utilizes a self‐attention mechanism to eliminate decoder effects and improve the performance of encoder. To validate the effectiveness of the proposed model, we conducted experiments using diverse datasets (MNIST, Fashion MNIST, SVHN, and WM811K). The results of the experiments demonstrated that our proposed method exhibited, on average, 2%–10% higher performance in terms of accuracy and F‐measure compared to the existing feature extraction techniques in the classification problem. While our research has limitations, specifically in its applicability only to the selection of image features, future studies should be undertaken to explore its potential application in various fields. [ABSTRACT FROM AUTHOR]

As a physiological signal reflecting the state of muscle activation, surface electromyography (sEMG) plays a vital role in the assessment of neuromuscular health, human–computer interaction, and gait analysis. Inspired by the audio signal analysis outcome that features extracted with Mel Frequency Cepstral Coefficient (MFCC) empower better representation, this paper proposes a comparative study of a gesture recognition method by using and improving with the MFCC features of sEMG. Comparing and combining with the conventional time-domain and frequency-domain features, different learning-based techniques are deployed to evaluate the performance of the proposed approach on the NinaPro datasets. The proposed approach was evaluated on the NinaPro-DB1 and NinaPro-DB2 datasets, achieving the improvements of 3.42% and 3.67%, respectively, in terms of the highest accuracy using the standard MFCC method. Correspondingly, when combined with the improved MFCC, the accuracy was further increased, reaching the maximum values of 89.82% and 87.82%, respectively, on the two datasets. The impact on the performance reveals the effectiveness of MFCC, and the results show that the proposed method has the potential to realize high-precision gesture recognition. [ABSTRACT FROM AUTHOR]

As a highly promising technique in the field of similarity search, the hashing-based image retrieval algorithm has received continued attention in recent years because of its strong ability to efficiently provide accurate results when measuring similarities between data instances in the binary space. At the level of practical applications, it is necessary to consider both the optimization of the hashing algorithm itself and the mining of image semantic features to produce high-quality hash codes with a strong semantic resolution, enabling the performance maximisation of such a neighbour search system on complex scenario images. To this end, a unified deep hashing framework termed Dual Enhanced Semantic Hashing (DESH) is proposed in this work. Specifically, it benefits the fast image retrieval in two aspects: 1) By taking advantage of dynamic multi-scale fusion and graph encoding networks, the dual enhanced feature learning significantly strengthens the semantic feature representation by jointly exploring and encoding the local multi-scale information with the high-order adjacency relationship between original images; 2) With the joint optimization of diverse loss functions, the binary semantic modelling process seamlessly module the image semantic information within the hash function learning in the code generation, aiming to generate discriminative hash codes to refine the retrieval performance eventually. By conducting extensive experiments on public datasets, the retrieval results further validate the claims of the proposed DESH by exhibiting its superior performance against competitive state-of-the-art baselines. [ABSTRACT FROM AUTHOR]

The precise tie-points (TPs) on synthetic aperture radar (SAR) images are a critical cornerstone in the global digital elevation model (DEM) and digital ortho map (DOM) production process. While there are abundant studies on SAR TPs matching, improvement opportunities persist in large areas. The correspondences have pixel-level errors during geocoding, which result in misalignment between global products. Consequently, this paper proposed a robust method for SAR images TPs matching, which consists of three key steps: (1) interest point extraction based on the dynamic Harris area entropy (DHAE) grid; (2) adaptive determination of template size; (3) normalized cross correlation (NCC) template matching. DHAE is a regional texture information grid based on the SAR-Harris map, and it is achieved through dynamic block division. Generating the DHAE grid over SAR images enables the extraction of interest points that have regional feature representation and distribution uniformity. A variable-size matching template is adaptively determined based on DHAE to enhance template quality while maintaining computational efficiency. Subsequently, the NCC algorithm is employed to find subpixel-precise correspondences. The proposed method is applied on TPs matching in 57 Terra-SAR images, which cover a large geographical area. Furthermore, the overlapping area is partitioned into five segments according to different coverage types. The experimental results demonstrate that the proposed method outperforms other template matching methods. For all coverage types, the proposed method exhibits high-precision sub-pixel results that reach up to 38.64% in terms of the relative positioning error (RPE), particularly in texture-weak and large areas. [ABSTRACT FROM AUTHOR]

Deep learning models have been shown to be vulnerable to critical attacks under adversarial conditions. Attackers are able to generate powerful adversarial examples by searching for adversarial perturbations, without interfering with model training or directly modifying the model. This phenomenon indicates an endogenous problem in existing deep learning frameworks. Therefore, optimizing individual models for defense is often limited and can always be defeated by new attack methods. Ensemble defense has been shown to be effective in defending against adversarial attacks by combining diverse models. However, the problem of insufficient differentiation among existing models persists. Active defense in cyberspace security has successfully defended against unknown vulnerabilities by integrating subsystems with multiple different implementations to achieve a unified mission objective. Inspired by this, we propose exploring the feasibility of achieving model differentiation by changing the data features used in training individual models, as they are the core factor of functional implementation. We utilize several feature extraction methods to preprocess the data and train differentiated models based on these features. By generating adversarial perturbations to attack different models, we demonstrate that the feature representation of the data is highly resistant to adversarial perturbations. The entire ensemble is able to operate normally in an error-bearing environment. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Frontiers of Computer Science & Technology is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Deep learning has been widely applied in the field of electrocardiogram (ECG) signal analysis due to its powerful data representation capability. However, supervised methods require a large amount of labeled data, and ECG data annotation is typically time-consuming and costly. Additionally, supervised methods are limited by the finite data types in the training set, resulting in limited generalization performance. Therefore, how to leverage massive unlabeled ECG signals for data mining and universal feature representation has become an urgent problem to be addressed. Self-supervised learning (SSL) is an effective approach to address the issue of missing annotated ECG data and improve the transfer ability of the model by learning generalized features from unlabeled data using pre-defined proxy tasks. However, existing surveys on self-supervised learning mostly focus on the domains of images or temporal signals, and there is a relative lack of comprehensive reviews on self-supervised learning in the ECG domain. To fill this gap, this paper provides a comprehensive review of advanced self-supervised learning methods used in the field of ECG. Firstly, a systematic summary and classification of self-supervised learning methods for ECG are presented, starting from two learning paradigms—contrastive and predictive. The basic principles of different categories of methods are elaborated, and the characteristics of each method are analyzed in detail, highlighting the advantages and limitations of each approach. Subsequently, a summary is provided for the commonly used datasets and application scenarios in ECG self-supervised learning, along with a review of data augmentation methods frequently applied in the ECG domain, offering a systematic reference for subsequent research. Finally, an in-depth discussion is presented on the current challenges of self-supervised learning within the ECG field, and future directions for the development of ECG self-supervised learning are explored, providing guidance for subsequent research in the field.

Kazakh is one of the multilingual languages of China and is widely spoken in some areas of Xinjiang, China. However, due to the fact that Kazakh is a language in which several characters are glued together to form a continuous word with a unique shape and complex structural combinations of relationships. This paper explores a solution for offline image retrieval of handwritten Kazakh words, which is a challenging task because, due to the lack of relevant datasets and the special writing morphology of the Kazakh language, traditional text image retrieval algorithms often struggle to achieve satisfactory results when dealing with writing styles that are varied and adherent to the language. Therefore, a dataset of offline Kazakh handwritten document images was created in this paper. The dataset contains 300 pages of document images with 20 500 words. Then, a new model called the 'AENet' is proposed. The model utilizes an attention mechanism to focus more finely on focal regions such as centers, inflection points, and contours of handwritten word images and to capture important local features from different scales. Fusion space pyramid pooling, feature aggregation, encoding operations, and feature downscaling and reconstruction are used to extract and reconstruct more representative feature representations from local to global to capture the overall information in the word images. Through experimental evaluation on Kazak‐80, Zilla‐64, and HWDB1.1‐375 datasets, it is verified that the method significantly improves the mAP for image retrieval of handwritten words, which is especially applicable to adhesive languages like Kazakh. © 2024 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Object trackers based on Siamese networks view tracking as a similarity-matching process. However, the correlation operation operates as a local linear matching process, limiting the tracker's ability to capture the intricate nonlinear relationship between the template and search region branches. Moreover, most trackers don't update the template and often use the first frame of an image as the initial template, which will easily lead to poor tracking performance of the algorithm when facing instances of deformation, scale variation, and occlusion of the tracking target. To this end, we propose a Simases tracking network with a multi-attention mechanism, including a template branch and a search branch. To adapt to changes in target appearance, we integrate dynamic templates and multi-attention mechanisms in the template branch to obtain more effective feature representation by fusing the features of initial templates and dynamic templates. To enhance the robustness of the tracking model, we utilize a multi-attention mechanism in the search branch that shares weights with the template branch to obtain multi-scale feature representation by fusing search region features at different scales. In addition, we design a lightweight and simple feature fusion mechanism, in which the Transformer encoder structure is utilized to fuse the information of the template area and search area, and the dynamic template is updated online based on confidence. Experimental results on publicly tracking datasets show that the proposed method achieves competitive results compared to several state-of-the-art trackers. [ABSTRACT FROM AUTHOR]

Software vulnerability detection aims to proactively reduce the risk to software security and reliability. Despite advancements in deep-learning-based detection, a semantic gap still remains between learned features and human-understandable vulnerability semantics. In this paper, we present an XAI-based framework to assess program code in a graph context as feature representations and their effect on code vulnerability classification into multiple Common Weakness Enumeration (CWE) types. Our XAI framework is deep-learning-model-agnostic and programming-language-neutral. We rank the feature importance of 40 syntactic constructs for each of the top 20 distributed CWE types from three datasets in Java and C++. By means of four metrics of information retrieval, we measure the similarity of human-understandable CWE types using each CWE type’s feature contribution ranking learned from XAI methods. We observe that the subtle semantic difference between CWE types occurs after the variation in neighboring features’ contribution rankings. Our study shows that the XAI explanation results have approximately 78% Top-1 to 89% Top-5 similarity hit rates and a mean average precision of 0.70 compared with the baseline of CWE similarity identified by the open community experts. Our framework allows for code vulnerability patterns to be learned and contributing factors to be assessed at the same stage.

Abstract Background Protein residue–residue distance maps are used for remote homology detection, protein information estimation, and protein structure research. However, existing prediction approaches are time-consuming, and hundreds of millions of proteins are discovered each year, necessitating the development of a rapid and reliable prediction method for protein residue–residue distances. Moreover, because many proteins lack known homologous sequences, a waiting-free and alignment-free deep learning method is needed. Result In this study, we propose a learning framework named FreeProtMap. In terms of protein representation processing, the proposed group pooling in FreeProtMap effectively mitigates issues arising from high-dimensional sparseness in protein representation. In terms of model structure, we have made several careful designs. Firstly, it is designed based on the locality of protein structures and triangular inequality distance constraints to improve prediction accuracy. Secondly, inference speed is improved by using additive attention and lightweight design. Besides, the generalization ability is improved by using bottlenecks and a neural network block named local microformer. As a result, FreeProtMap can predict protein residue–residue distances in tens of milliseconds and has higher precision than the best structure prediction method. Conclusion Several groups of comparative experiments and ablation experiments verify the effectiveness of the designs. The results demonstrate that FreeProtMap significantly outperforms other state-of-the-art methods in accurate protein residue–residue distance prediction, which is beneficial for lots of protein research works. It is worth mentioning that we could scan all proteins discovered each year based on FreeProtMap to find structurally similar proteins in a short time because the fact that the structure similarity calculation method based on distance maps is much less time-consuming than algorithms based on 3D structures.

Aim: The wealth of data generated by continuous glucose monitoring (CGM) provides new opportunities for revealing heterogeneities in patients with type 2 diabetes mellitus (T2DM). We aimed to develop a method using CGM data to discover T2DM subtypes and investigate their relationship with clinical phenotypes and microvascular complications. Methods: The data from 3119 patients with T2DM who wore blinded CGM at an academic medical centre was collected, and a glucose symbolic pattern (GSP) metric was created that combined knowledge‐based temporal abstraction with numerical vectorization. The k‐means clustering was applied to GSP to obtain subgroups of patients with T2DM. Clinical characteristics and the presence of diabetic retinopathy and albuminuria were compared among the subgroups. The findings were validated in an independent population comprising 773 patients with T2DM. Results: By using GSP, four subgroups were identified with distinct features in CGM profiles and parameters. Moreover, the clustered subgroups differed significantly in clinical phenotypes, including indices of pancreatic β‐cell function and insulin resistance (all p <.001). After adjusting for confounders, group C (the most insulin resistant) had a significantly higher risk of albuminuria (odds ratio = 1.24, 95% confidence interval: 1.03‐1.39) relative to group D, which had the best glucose control. These findings were confirmed in the validation set. Conclusion: Subtyping patients with T2DM using CGM data may help identify high‐risk patients for microvascular complications and provide insights into the underlying pathophysiology. This method may help refine clinically meaningful stratification of patients with T2DM and inform personalized diabetes care. [ABSTRACT FROM AUTHOR]

At present, as a research hotspot for time series data (TSD), the deep clustering analysis of TSD has huge research value and practical significance. However, there still exist the following three problems: (1) For deep clustering based on joint optimization, the inevitably mutual interference existing between deep feature representation learning progress and clustering progress leads to difficult model training especially in the initial stage, the possible feature space distortion, inaccurate and weak feature representation; (2) Existing deep clustering methods are difficult to intuitively define the similarity of time series and rely heavily on complex feature extraction networks and clustering algorithms. (3) Multidimensional time series have the characteristics of high dimensions, complex relationships between dimensions, and variable data forms, thus generating a huge feature space. It is difficult for existing methods to select discriminative features, resulting in generally low accuracy of methods. Accordingly, to address the above three problems, we proposed a novel general two-stage multi-dimensional spatial features based multi-view deep clustering method 1DCAE-TSSAMC (One-dimensional deep convolutional auto-encoder based two-stage stepwise amplification multi-clustering). We conducted verification and analysis based on real-world important multi-scenario, and compared with many other benchmarks ranging from the most classic approaches such as K-means and Hierarchical to the state-of-the-art approaches based on deep learning such as Deep Temporal Clustering (DTC) and Temporal Clustering Network (TCN). Experimental results show that the new method outperforms the other benchmarks, and provides more accurate, richer, and more reliable analysis results, more importantly, with significant improvement in accuracy and spatial linear separability. [ABSTRACT FROM AUTHOR]

Software vulnerability detection aims to proactively reduce the risk to software security and reliability. Despite advancements in deep-learning-based detection, a semantic gap still remains between learned features and human-understandable vulnerability semantics. In this paper, we present an XAI-based framework to assess program code in a graph context as feature representations and their effect on code vulnerability classification into multiple Common Weakness Enumeration (CWE) types. Our XAI framework is deep-learning-model-agnostic and programming-language-neutral. We rank the feature importance of 40 syntactic constructs for each of the top 20 distributed CWE types from three datasets in Java and C++. By means of four metrics of information retrieval, we measure the similarity of human-understandable CWE types using each CWE type's feature contribution ranking learned from XAI methods. We observe that the subtle semantic difference between CWE types occurs after the variation in neighboring features' contribution rankings. Our study shows that the XAI explanation results have approximately 78% Top-1 to 89% Top-5 similarity hit rates and a mean average precision of 0.70 compared with the baseline of CWE similarity identified by the open community experts. Our framework allows for code vulnerability patterns to be learned and contributing factors to be assessed at the same stage. [ABSTRACT FROM AUTHOR]

Background: Protein residue–residue distance maps are used for remote homology detection, protein information estimation, and protein structure research. However, existing prediction approaches are time-consuming, and hundreds of millions of proteins are discovered each year, necessitating the development of a rapid and reliable prediction method for protein residue–residue distances. Moreover, because many proteins lack known homologous sequences, a waiting-free and alignment-free deep learning method is needed. Result: In this study, we propose a learning framework named FreeProtMap. In terms of protein representation processing, the proposed group pooling in FreeProtMap effectively mitigates issues arising from high-dimensional sparseness in protein representation. In terms of model structure, we have made several careful designs. Firstly, it is designed based on the locality of protein structures and triangular inequality distance constraints to improve prediction accuracy. Secondly, inference speed is improved by using additive attention and lightweight design. Besides, the generalization ability is improved by using bottlenecks and a neural network block named local microformer. As a result, FreeProtMap can predict protein residue–residue distances in tens of milliseconds and has higher precision than the best structure prediction method. Conclusion: Several groups of comparative experiments and ablation experiments verify the effectiveness of the designs. The results demonstrate that FreeProtMap significantly outperforms other state-of-the-art methods in accurate protein residue–residue distance prediction, which is beneficial for lots of protein research works. It is worth mentioning that we could scan all proteins discovered each year based on FreeProtMap to find structurally similar proteins in a short time because the fact that the structure similarity calculation method based on distance maps is much less time-consuming than algorithms based on 3D structures. [ABSTRACT FROM AUTHOR]

Fine-grained entity typing assigns fine-grained types to entities in the text, which can provide entities with rich semantic information through type information, and plays important roles in downstream tasks such as relation extraction, entity linking, and question answering systems. Since the length and position of entities in sentences are not uniform, the representation of entities in context can not be calculated. Existing fine-grained entity typing models process entity mentions and their contexts separately into individual feature representations, which separates the semantic relationship between them. This paper proposes a context-aware feature representation method in fine-grained entity typing, which places entities back into their contexts and solves the problem of computing entity feature representation when the entity length and position are not uniform. Experimental results demonstrate that this method can extract the feature representation of entities in their contexts, and significantly improve the performance of fine-grained entity typing. The Macro-F1 value of this method on the Chinese finegrained entity classification dataset CFET is improved by more than 10%. [ABSTRACT FROM AUTHOR]

Network traffic classification is crucial for identifying network applications and defending against network threats. Traditional traffic classification approaches struggle to extract structural features and suffer from poor interpretability of feature representations. The high symmetry between network traffic classification and its interpretable feature representation is vital for network traffic analysis. To address these issues, this paper proposes a traffic classification and feature representation model named the attention mechanism autoencoder (AMAE). The AMAE model extracts the global spatial structural features of network traffic through attention mechanisms and employs an autoencoder to extract local structural features and perform dimensionality reduction. This process maps different network traffic features into one-dimensional coordinate systems in the form of spectra, termed FlowSpectrum. The spectra of different network traffic represent different intervals in the coordinate system. This paper tests the interpretability and classification performance of network traffic features of the AMAE model using the ISCX-VPN2016 dataset. Experimental results demonstrate that by analyzing the overall distribution of attention weights and local weight values of network traffic, the model effectively explains the differences in the spectral representation intervals of different types of network traffic. Furthermore, our approach achieves the highest classification accuracy of up to 100% for non-VPN-encrypted traffic and 99.69% for VPN-encrypted traffic, surpassing existing traffic classification schemes. [ABSTRACT FROM AUTHOR]

Deep learning-based object detection methods have achieved great performance improvement. However, the current mainstream object detectors focus on normal illumination images, while low-illumination object detection is often ignored. It is still a challenging task to detect objects in low-illumination scenes due to insufficient illumination and low visibility. To address this issue, we propose a low-illumination object detection network based on feature representation refinement and semantic-aware enhancement, called FRSE-Net. There are two key components in the proposed network, including a feature capture module (FCM) and a semantic aggregation module (SAM). First, the FCM is designed to enhance the feature representation of the feature map, thus making the object features more discriminative. This is beneficial to capture more effective feature information for subsequent prediction tasks. Furthermore, the SAM is introduced to enhance the semantic-aware ability of the model in low-light images, which makes the detection network focus on the objects of interest to learn rich semantic information. Finally, the experimental results on two low-light image datasets demonstrate the effectiveness and superiority of the proposed network when compared with other advanced low-illumination detection methods. [ABSTRACT FROM AUTHOR]

Tobacco is a valuable crop, but its disease identification is rarely involved in existing works. In this work, we use few-shot learning (FSL) to identify abnormalities in tobacco. FSL is a solution for the data deficiency that has been an obstacle to using deep learning. However, weak feature representation caused by limited data is still a challenging issue in FSL. The weak feature representation leads to weak generalization and troubles in crossdomain. In this work, we propose a feature representation enhancement network (FREN) that enhances the feature representation through instance embedding and task adaptation. For instance embedding, global max pooling, and global average pooling are used together for adding more features, and Gaussian-like calibration is used for normalizing the feature distribution. For task adaptation, self-attention is adopted for task contextualization. Given the absence of publicly available data on tobacco, we created a tobacco leaf abnormality dataset (TLA), which includes 16 categories, two settings, and 1,430 images in total. In experiments, we use PlantVillage, which is the benchmark dataset for plant disease identification, to validate the superiority of FREN first. Subsequently, we use the proposed method and TLA to analyze and discuss the abnormality identification of tobacco. For the multi-symptom diseases that always have low accuracy, we propose a solution by dividing the samples into subcategories created by symptom. For the 10 categories of tomato in PlantVillage, the accuracy achieves 66.04% in 5-way, 1-shot tasks. For the two settings of the tobacco leaf abnormality dataset, the accuracies were achieved at 45.5% and 56.5%. By using the multisymptom solution, the best accuracy can be lifted to 60.7% in 16-way, 1-shot tasks and achieved at 81.8% in 16-way, 10-shot tasks. The results show that our method improves the performance greatly by enhancing feature representation, especially for tasks that contain categories with high similarity. The desensitization of data when crossing domains also validates that the FREN has a strong generalization ability. [ABSTRACT FROM AUTHOR]

Detecting small and densely distributed (SDD) objects in large-size images is quite challenging due to the facts that directly inputting such images to the detection network would result in severe geometric deformation and the overcrowded objects would lead to unclear feature expression. In this paper, we propose a two-level data augmentation method, referred to as MultiCut-MultiMix, to solve the problems, in which MultiCut is developed to avoid feature distortion at the physical level and MultiMix is designed to enrich background at the pixel level. Specifically, according to the images size required by the detection network, MultiCut, by appropriately tuning two introduced parameters, cuts large-size images into a series of image chips that are suitable for training, and meanwhile ensuring that the objects at the cutting edge would not lose. Furthermore, to strengthen feature information in these obtained image chips, MultiMix fuses different chips into new ones, in which the chip with most SDD objects will be remained as the major information and the others as the background. The fused chips from MultiMix, together with the original chips from MultiCut, then serve as the new data to train the detection network, by which the dataset is enlarged and thus overfitting can be effectively avoided. Extensive ablation experiments show that, compared with existing approaches, our method usually significantly assists detection networks to identify SDD objects in large-size images. For example, based on the bacterial dataset, 14.69% improvement in the mean average precision is achieved over the classical CutMix on Darknet53. [ABSTRACT FROM AUTHOR]

The reliable way to discern human emotions in various circumstances has been proven to be through facial expressions. Facial expression recognition (FER) has emerged as a research topic to identify various essential emotions in the present exponential rise in research for emotion detection. Happiness is one of these basic emotions everyone may experience, and facial expressions are better at detecting it than other emotionmeasuring methods. Most techniques have been designed to recognize various emotions to achieve the highest level of general precision. Maximizing the recognition accuracy for a particular emotion is challenging for researchers. Some techniques exist to identify a single happy mood recorded in unrestricted video. Still, they are all limited by the processing of extreme head posture fluctuations that they need to consider, and their accuracy still needs to be improved. This research proposes a novel hybrid facial emotion recognition using unconstraint video to improve accuracy. Here, a Deep Belief Network (DBN) with long short-term memory (LSTM) is employed to extract dynamic data from the video frames. The experiments conducted uses decision-level and feature-level fusion techniques are applied unconstrained video dataset. The outcomes show that the proposed hybrid approach may be more precise than some existing facial expression models. [ABSTRACT FROM AUTHOR]

Abstract The voltage-gated sodium (Nav) channel is a crucial molecular component responsible for initiating and propagating action potentials. While the α subunit, forming the channel pore, plays a central role in this function, the complete physiological function of Nav channels relies on crucial interactions between the α subunit and auxiliary proteins, known as protein–protein interactions (PPI). Nav blocking peptides (NaBPs) have been recognized as a promising and alternative therapeutic agent for pain and itch. Although traditional experimental methods can precisely determine the effect and activity of NaBPs, they remain time-consuming and costly. Hence, machine learning (ML)-based methods that are capable of accurately contributing in silico prediction of NaBPs are highly desirable. In this study, we develop an innovative meta-learning-based NaBP prediction method (MetaNaBP). MetaNaBP generates new feature representations by employing a wide range of sequence-based feature descriptors that cover multiple perspectives, in combination with powerful ML algorithms. Then, these feature representations were optimized to identify informative features using a two-step feature selection method. Finally, the selected informative features were applied to develop the final meta-predictor. To the best of our knowledge, MetaNaBP is the first meta-predictor for NaBP prediction. Experimental results demonstrated that MetaNaBP achieved an accuracy of 0.948 and a Matthews correlation coefficient of 0.898 over the independent test dataset, which were 5.79% and 11.76% higher than the existing method. In addition, the discriminative power of our feature representations surpassed that of conventional feature descriptors over both the training and independent test datasets. We anticipate that MetaNaBP will be exploited for the large-scale prediction and analysis of NaBPs to narrow down the potential NaBPs.

This study introduces the concept of relative age position learning to supplement the learning process for age estimation models. Drawing inspiration from feature recalibration modules that prioritize features based on their importance, a novel age-based reweighting module is developed to enhance feature representation in the proposed age estimation method. The proposed reweighting module obtains the features of the selected references for each age and further exploits them to reweight the features of the input images based on age importance. The resulting recalibrated features are then assessed for relative age position prediction. In addition, to achieve better generalization performance in age estimation, a gender prediction head is added to create a multi-task learning network that simultaneously predicts the ages and genders of the input images. Through extensive experiments, we demonstrate that the proposed approach outperforms other state-of-the-art age estimation methods on three challenging benchmark datasets for facial age estimation: AgeDB, AFAD, and CACD.

Feature representation is a crucial issue in multimodal image registration. The handcrafted features extracted by traditional methods are highly sensitive to nonlinear radiation differences, while supervised learning methods are limited by deficient labeled samples in the remote sensing field. Therefore, this article proposes a consistency feature representation learning method for multimodal image registration, which involves mapping data into a common feature space to realize the accurate alignment of remote sensing images. First, a contrastive network with a spatial attention mechanism is driven to enhance the capability to highlight high-level features of images. Second, a positive sample augmentation strategy is implemented with contrastive loss, which helps the model learn the inherent features better, and imposes constraints on the sample similarity to optimize the feature projection. Finally, a multimodal image registration framework is proposed to enhance the stability of feature matching. The proposed framework achieves accurate feature extraction and consistency feature description for multimodal images, ensuring robustness against nonlinear radiometric differences. Experimental results demonstrate that the proposed method obtains more reliable registration results on the SEN1-2 dataset. Furthermore, the proposed algorithm achieves superior performance on data from other modalities, indicating strong generalization ability.

Polarimetric synthetic aperture radar (PolSAR) has attracted more attentions because of its excellent observation ability, and PolSAR image classification has become one of the significant tasks in remote sensing interpretation. Various types and sizes of land cover objects lead to misclassification, especially in the boundaries of different categories. To solve these issues, a multiscale superpixel-guided weighted graph convolutional network (MSGWGCN) is proposed for classifying PolSAR images. In the proposed MSGWGCN, multiscale superpixel features are imported into the weighted graph convolutional network to obtain higher level representation, which can make full use of land cover object information in PolSAR images. Moreover, to fuse pixel-level features at different scales, a multiscale feature cascade fusion module is built, which plays an important role in preserving classification details. Experiments on three PolSAR datasets indicate that the proposed MSGWGCN performs better than other advanced methods on PolSAR classification task.

ABSTRACTThe difficult, time-consuming, and imbalanced acquisition of landslide inventories in complex and heterogeneous large areas often results in limited predictive performance for most statistical landslide spatial prediction methods. Although partial transfer learning methods have produced reliable predictive results and successfully implemented knowledge transfer between data-rich and data-scarce areas, most of these methods generally only extract inadequate environmental features and lack the ability to interpret when, what, and how to effectively transfer knowledge to other regions with limited data. In this paper, a heterogeneous transfer learning method considering feature representations and environmental consistency, which features robustness, similarity, and transferability, is proposed for landslide spatial prediction. Specifically, we trained a stacked autoencoder (SAE) to extract more nonlinear features among environmental factors, and added an environmental similarity criterion to the transfer adaptation boosting (TrAdaBoost) algorithm to minimize feature dissimilarities and avoid negative transfer in different scenarios. To evaluate the robustness of the proposed method, we first selected two target areas (Lushan County and Luding County, China) and a source area (Wenchuan County, China) as case study areas. Then, we directly combined the source area and target area as an additional dataset without considering transfer learning to validate the significance and necessity of the proposed method. The area under the receiver operating characteristic curve (AUC) of the two target regions for the proposed method were 0.920 and 0.972, respectively, which were greater than those of the traditional TrAdaBoost (0.909 and 0.969, respectively), SAE (0.790 and 0.937, respectively), and random forest (0.915 and 0.966, respectively) methods. Furthermore, the AUC values of the SAE (0.851 and 0.900) and random forest (0.890 and 0.935) models based on the expanded datasets were also lower than those of the proposed method. Therefore, the experimental results show that the proposed method can be generalized well due to its efficient utilization and high adaptability. Moreover, the proposed method can not only be applied to emergency rescue and disaster prevention but can also offer a promising way to improve landslide predictions of models with incomplete landslide inventories.

Visual affordance detection aims to understand the functional attributes of objects, which is crucial for robots to achieve interactive tasks. Most existing affordance detection methods mainly utilize the global image features for affordance detection while do not fully exploit the features of local relevant objects in the image, which often leads to suboptimal detection accuracy under the interference of cluttered backgrounds and neighbour objects. Numerous researches have proved that the accuracy of affordance detection largely depends on the quality of extracted image features. In this paper, we propose a novel affordance detection network with object shape mask guided feature encoders. The masks play as an attention mechanism that enforce the network to focus on the shape regions of target objects in the image, which facilitate to obtain high-quality features. Specifically, we first propose a shape mask guided encoder, which uses masks to effectively locate all target objects so as to extract more expressive features. Based on the encoder, we then propose a dual enhance feature aggregation module, which consists of two branches. The first branch encodes the global features of the original image, while the second branch locates each local relevant object and encodes its precise features. Aggregating these features enhances the feature representation of each object, further improving feature quality and suppressing interference. Quantitative and qualitative evaluations compared with state-of-the-art methods demonstrate that the proposed method achieves superior performance on the two commonly used affordance detection datasets. [ABSTRACT FROM AUTHOR]

Alzheimer's disease is a cognitive disorder that threatens human lives. Computer-based diagnosis are recommended for Ad progression prediction. Capsule Networks receive investigators' attention which owes to the capability for the efficient features of the fuse multi-dimensionality and the model's coefficients among the samples. Various deep network models are adopted for node classification and utilization of a complete dataset for constructing the dense network, which is not used for independent testing. In addition, network maintenance and construction benefit matrix decomposition to overcome the demerits. An improved CapsNetwork ( i - C N ) model was proposed for AD diagnosis. The sample data provided among the routing neurons works from the lower level to higher level, i.e. it enhances the prediction accuracy. The model intends to recognize the complete entity by analysing the features of input samples. Capsules are vectors that represent the sample features. The improved network model tries to reduce the computational complexity identified in various existing learning approaches. The under-fitting issues due to smaller dataset are also rectified with the proposed model. Primarily, the learning model is realized by introducing the metric-based feature representation to test independently via the different classification tasks. Appropriate samples are needed to generalize the model without sensitivity and to enhance the performance. Further, designing the i - C N layer poses the probability of realizing the efficient fuse of multi-dimensional data. The model needs to be verified using TADPOLE where mild cognitive impairment (MCI) prediction and the earlier AD diagnosis are achieved. The proposed system gives better performance having an accuracy of 95% and AUC of 98.7%. The outcomes proved that the proposed system enhances flexibility to ensure better performance in the classification. It helps to promote the enhancement of disease diagnosis for deep learning algorithms. [ABSTRACT FROM AUTHOR]

As the application of large language models (LLMs) has broadened into the realm of biological predictions, leveraging their capacity for self-supervised learning to create feature representations of amino acid sequences, these models have set a new benchmark in tackling downstream challenges, such as subcellular localization. However, previous studies have primarily focused on either the structural design of models or differing strategies for fine-tuning, largely overlooking investigations into the nature of the features derived from LLMs. In this research, we propose different ESM2 representation extraction strategies, considering both the character type and position within the ESM2 input sequence. Using model dimensionality reduction, predictive analysis and interpretability techniques, we have illuminated potential associations between diverse feature types and specific subcellular localizations. Particularly, the prediction of Mitochondrion and Golgi apparatus prefer segments feature closer to the N-terminal, and phosphorylation site-based features could mirror phosphorylation properties. We also evaluate the prediction performance and interpretability robustness of Random Forest and Deep Neural Networks with varied feature inputs. This work offers novel insights into maximizing LLMs' utility, understanding their mechanisms, and extracting biological domain knowledge. Furthermore, we have made the code, feature extraction API, and all relevant materials available at https://github.com/yujuan-zhang/feature-representation-for-LLMs. [ABSTRACT FROM AUTHOR]

Three-dimensional voxel models are widely applied in various fields such as 3D imaging, industrial design, and medical imaging. The advancement of 3D modeling techniques and measurement devices has made the generation of three-dimensional models more convenient. The exponential increase in the number of 3D models presents a significant challenge for model retrieval. Currently, these models are numerous and typically represented as point clouds or meshes, resulting in sparse data and high feature dimensions within the retrieval database. Traditional methods for 3D model retrieval suffer from high computational complexity and slow retrieval speeds. To address this issue, this paper combines spatial-filling curves with octree structures and proposes a novel approach for representing three-dimensional voxel model sequence data features, along with a similarity measurement method based on symbolic operators. This approach enables efficient similarity calculations and rapid dimensionality reduction for the three-dimensional model database, facilitating efficient similarity calculations and expedited retrieval. [ABSTRACT FROM AUTHOR]

Existing handcrafted keypoint detectors typically focus on designing specific local structures manually while ignoring whether they have enough flexibility to explore diverse visual patterns in an image. Despite the advancement of learning-based approaches in the past few years, most of them still rely on the availability of the outputs of handcrafted detectors as a part of training. In fact, such dependence limits their ability to discover various visual information. Recently, semi-handcrafted methods based on sparse coding have emerged as a promising paradigm to alleviate the above issue. However, the visual relationships between feature points have not been considered in the encoding stage, which may weaken the discriminative capability of feature representations for keypoint recognition. To tackle this problem, we propose a novel sparsity-guided discriminative feature representation (SDFR) method that attempts to explore the intrinsic correlations of keypoint candidates, thus ensuring the validity of characterizing distinctive and diverse structural information. Specifically, we first incorporate an affinity constraint into the feature representation objective, which jointly encodes all the patches in an image while highlighting the similarities and differences between them. Meanwhile, a smoother sparsity regularization with the Frobenius norm is leveraged to further preserve the similarity relationships of patch representations. Due to the differentiable property of this sparsity, SDFR is computationally feasible and effective for representing dense patches. Finally, we treat the SDFR model as multiple optimization sub-problems and introduce an iterative solver. During comprehensive evaluations on five challenging benchmarks, the proposed method achieves favorable performances compared with the state of the art in the literature. [ABSTRACT FROM AUTHOR]

The voltage-gated sodium (Nav) channel is a crucial molecular component responsible for initiating and propagating action potentials. While the α subunit, forming the channel pore, plays a central role in this function, the complete physiological function of Nav channels relies on crucial interactions between the α subunit and auxiliary proteins, known as protein–protein interactions (PPI). Nav blocking peptides (NaBPs) have been recognized as a promising and alternative therapeutic agent for pain and itch. Although traditional experimental methods can precisely determine the effect and activity of NaBPs, they remain time-consuming and costly. Hence, machine learning (ML)-based methods that are capable of accurately contributing in silico prediction of NaBPs are highly desirable. In this study, we develop an innovative meta-learning-based NaBP prediction method (MetaNaBP). MetaNaBP generates new feature representations by employing a wide range of sequence-based feature descriptors that cover multiple perspectives, in combination with powerful ML algorithms. Then, these feature representations were optimized to identify informative features using a two-step feature selection method. Finally, the selected informative features were applied to develop the final meta-predictor. To the best of our knowledge, MetaNaBP is the first meta-predictor for NaBP prediction. Experimental results demonstrated that MetaNaBP achieved an accuracy of 0.948 and a Matthews correlation coefficient of 0.898 over the independent test dataset, which were 5.79% and 11.76% higher than the existing method. In addition, the discriminative power of our feature representations surpassed that of conventional feature descriptors over both the training and independent test datasets. We anticipate that MetaNaBP will be exploited for the large-scale prediction and analysis of NaBPs to narrow down the potential NaBPs. [ABSTRACT FROM AUTHOR]

It proposed an improved artificial bee colony algorithm to better address the flexible job shop scheduling problem with the objective of minimizing the makespan. Firstly, it employed random selection and reverse learning strategies to enhance the quality of initial food sources. Simultaneously, it designed a novel feature representation method to calculate the distance between food sources. During the leading bee phase, it introduced cross-over and mutation strategies to optimize nearby food sources in the population. Moreover, in the exploring bee phase, it incorporate six variable neighborhood methods to expand the search space of solutions. Subsequently, in the scout bee phase, it eliminated local optima individuals based on the potential value of their food sources. This paper conducted extensive experiments on 15 datasets, and the results demonstrate that the performance of the proposed improved algorithm significantly outperforms that of four other well-known swarm intelligence optimization algorithms. This study provides a novel and effective approach for solving the flexible job shop scheduling problem with the goal of minimizing the makespan, holding important practical value for real-world production scheduling. [ABSTRACT FROM AUTHOR]

Autoencoders (AEs) are a type of unsupervised neural networks, which can be used to solve various tasks, e.g., dimensionality reduction, image compression, and image denoising. An AE has two goals: (i) compress the original input to a low-dimensional space at the bottleneck of the network topology using an encoder, (ii) reconstruct the input from the representation at the bottleneck using a decoder. Both encoder and decoder are optimized jointly by minimizing a distortion-based loss which implicitly forces the model to keep only the information in input data required to reconstruct them and to reduce redundancies. In this paper, we propose a scheme to explicitly penalize feature redundancies in the bottleneck representation. To this end, we propose an additional loss term, based on the pairwise covariances of the network units, which complements the data reconstruction loss forcing the encoder to learn a more diverse and richer representation of the input. We tested our approach across different tasks, namely dimensionality reduction, image compression, and image denoising. Experimental results show that the proposed loss leads consistently to superior performance compared to using the standard AE loss. • We propose a scheme to avoid redundant features in the bottleneck representation of autoencoders. • We explicitly penalize the pair-wise correlations between the features and learn a diverse compressed embedding. • The proposed penalty acts as an unsupervised regularizer and can be integrated into any autoencoder model. [ABSTRACT FROM AUTHOR]

In recent years, various self-knowledge distillation approaches have been proposed to reduce the cost of training teacher networks. However, these methods often overlook the significance of deep features. To address this limitation and strengthen the capability of deep features while preserving the ability of shallow features, we propose performing Self-Knowledge Distillation via Pyramid Feature Refinement (PR-SKD). Inspired by the representation learning characteristics of deep neural networks, PR-SKD builds a cohort of sub-networks with a pyramid architecture to hierarchically transfer refined information to the target network. According to the different contributions and functions between deep and shallow feature maps, our PR-SKD fully utilizes feature information to improve deep feature representation ability without compromising the capability of shallow feature maps. Extensive experiments on various image classification datasets demonstrate the superiority of our proposed method over widely used state-of-the-art knowledge distillation methods. The code is available at: https://github.com/wo16pao/PR-SKD. • The method recognizes the variation contribution of different feature maps in self-distillation. • A new hierarchical refinement distillation technique via pyramid architecture. • The approach improves the deep feature representative capability of the target network. • The proposed technique achieves highly competitive performance on recent benchmarks by focusing on deep feature representation. [ABSTRACT FROM AUTHOR]

Human visual performance for basic visual dimensions (e.g., contrast sensitivity and acuity) peaks at the fovea and decreases with eccentricity. The eccentricity effect is related to the larger visual cortical surface area corresponding to the fovea, but it is unknown if differential feature tuning contributes to this eccentricity effect. Here, we investigated two system-level computations underlying the eccentricity effect: featural representation (tuning) and internal noise. Observers (both sexes) detected a Gabor embedded in filtered white noise which appeared at the fovea or one of four perifoveal locations. We used psychophysical reverse correlation to estimate the weights assigned by the visual system to a range of orientations and spatial frequencies (SFs) in noisy stimuli, which are conventionally interpreted as perceptual sensitivity to the corresponding features. We found higher sensitivity to task-relevant orientations and SFs at the fovea than that at the perifovea, and no difference in selectivity for either orientation or SF. Concurrently, we measured response consistency using a double-pass method, which allowed us to infer the level of internal noise by implementing a noisy observer model. We found lower internal noise at the fovea than that at the perifovea. Finally, individual variability in contrast sensitivity correlated with sensitivity to and selectivity for task-relevant features as well as with internal noise. Moreover, the behavioral eccentricity effect mainly reflects the foveal advantage in orientation sensitivity compared with other computations. These findings suggest that the eccentricity effect stems from a better representation of task-relevant features and lower internal noise at the fovea than that at the perifovea. [ABSTRACT FROM AUTHOR]

With the advancement of deep neural networks, several methods leveraging convolution neural networks (CNNs) have gained prominence in the field of remote sensing object detection. Acquiring accurate feature representations from feature maps is a critical step in CNN-based object detection methods. Previously, region of interest (RoI)-based methods have been widely used, but of late, deformable convolution network (DCN)-based approaches have started receiving considerable attention. A significant challenge in the use of DCN-based methods is the inefficient distribution patterns of sampling points, stemming from a lack of effective and flexible guidance. To address this, our study introduces Saliency-Guided RepPoints (SGR), an innovative framework designed to enhance feature representation quality in remote sensing object detection. SGR employs a dynamic dual-domain alignment (DDA) training strategy to mitigate potential misalignment issues between spatial and feature domains during the learning process. Furthermore, we propose an interpretable visualization method to assess the alignment between feature representation and classification performance in DCN-based methods, providing theoretical analysis and validation for the effectiveness of sampling points. In this study, we assessed the proposed SGR framework through a series of experiments conducted on four varied and rigorous datasets: DOTA, HRSC2016, DIOR-R, and UCAS-AOD, all of which are widely employed in remote sensing object detection. The outcomes of these experiments substantiate the effectiveness of the SGR framework, underscoring its potential to enhance the accuracy of object detection within remote sensing imagery. [ABSTRACT FROM AUTHOR]

Bioinformatics and genomics are driving a healthcare revolution, particularly in the domain of drug discovery for anticancer peptides (ACPs). The integration of artificial intelligence (AI) has transformed healthcare, enabling personalized and immersive patient care experiences. These advanced technologies, coupled with the power of bioinformatics and genomic data, facilitate groundbreaking developments. The precise prediction of ACPs from complex biological sequences remains an ongoing challenge in the genomic area. Currently, conventional approaches such as chemotherapy, target therapy, radiotherapy, and surgery are widely used for cancer treatment. However, these methods fail to completely eradicate neoplastic cells or cancer stem cells and damage healthy tissues, resulting in morbidity and even mortality. To control such diseases, oncologists and drug designers highly desire to develop new preventive techniques with more efficiency and minor side effects. Therefore, this research provides an optimized computational-based framework for discriminating against ACPs. In addition, the proposed approach intelligently integrates four peptide encoding methods, namely amino acid occurrence analysis (AAOA), dipeptide occurrence analysis (DOA), tripeptide occurrence analysis (TOA), and enhanced pseudo amino acid composition (EPseAAC). To overcome the issue of bias and reduce true error, the synthetic minority oversampling technique (SMOTE) is applied to balance the samples against each class. The empirical results over two datasets, where the accuracy of the proposed model on the benchmark dataset is 97.56% and on the independent dataset is 95.00%, verify the effectiveness of our ensemble learning mechanism and show remarkable performance when compared with state-of-the-art (SOTA) methods. In addition, the application of metaverse technology in healthcare holds promise for transformative innovations, potentially enhancing patient experiences and providing novel solutions in the realm of preventive techniques and patient care. [ABSTRACT FROM AUTHOR]