In response to the scarcity of infrared aircraft samples and the tendency of traditional deep learning to overfit,a few-shot infrared aircraft classification method based on cross-correlation networks is proposed.This...In response to the scarcity of infrared aircraft samples and the tendency of traditional deep learning to overfit,a few-shot infrared aircraft classification method based on cross-correlation networks is proposed.This method combines two core modules:a simple parameter-free self-attention and cross-attention.By analyzing the self-correlation and cross-correlation between support images and query images,it achieves effective classification of infrared aircraft under few-shot conditions.The proposed cross-correlation network integrates these two modules and is trained in an end-to-end manner.The simple parameter-free self-attention is responsible for extracting the internal structure of the image while the cross-attention can calculate the cross-correlation between images further extracting and fusing the features between images.Compared with existing few-shot infrared target classification models,this model focuses on the geometric structure and thermal texture information of infrared images by modeling the semantic relevance between the features of the support set and query set,thus better attending to the target objects.Experimental results show that this method outperforms existing infrared aircraft classification methods in various classification tasks,with the highest classification accuracy improvement exceeding 3%.In addition,ablation experiments and comparative experiments also prove the effectiveness of the method.展开更多
Predicting the productivity of multistage fractured horizontal wells plays an important role in exploiting unconventional resources.In recent years,machine learning(ML)models have emerged as a new approach for such st...Predicting the productivity of multistage fractured horizontal wells plays an important role in exploiting unconventional resources.In recent years,machine learning(ML)models have emerged as a new approach for such studies.However,the scarcity of sufficient real data for model training often leads to imprecise predictions,even though the models trained with real data better characterize geological and engineering features.To tackle this issue,we propose an ML model that can obtain reliable results even with a small amount of data samples.Our model integrates the synthetic minority oversampling technique(SMOTE)to expand the data volume,the support vector machine(SVM)for model training,and the particle swarm optimization(PSO)algorithm for optimizing hyperparameters.To enhance the model performance,we conduct feature fusion and dimensionality reduction.Additionally,we examine the influences of different sample sizes and ML models for training.The proposed model demonstrates higher prediction accuracy and generalization ability,achieving a predicted R^(2)value of up to 0.9 for the test set,compared to the traditional ML techniques with an R^(2)of 0.13.This model accurately predicts the production of fractured horizontal wells even with limited samples,supplying an efficient tool for optimizing the production of unconventional resources.Importantly,the model holds the potential applicability to address similar challenges in other fields constrained by scarce data samples.展开更多
Image classification is crucial for various applications,including digital construction,smart manu-facturing,and medical imaging.Focusing on the inadequate model generalization and data privacy concerns in few-shot im...Image classification is crucial for various applications,including digital construction,smart manu-facturing,and medical imaging.Focusing on the inadequate model generalization and data privacy concerns in few-shot image classification,in this paper,we propose a federated learning approach that incorporates privacy-preserving techniques.First,we utilize contrastive learning to train on local few-shot image data and apply various data augmentation methods to expand the sample size,thereby enhancing the model’s generalization capabilities in few-shot contexts.Second,we introduce local differential privacy techniques and weight pruning methods to safeguard model parameters,perturbing the transmitted parameters to ensure user data privacy.Finally,numerical simulations are conducted to demonstrate the effectiveness of our proposed method.The results indicate that our approach significantly enhances model generalization and test accuracy compared to several popular federated learning algorithms while maintaining data privacy,highlighting its effectiveness and practicality in addressing the challenges of model generalization and data privacy in few-shot image scenarios.展开更多
Although conventional object detection methods achieve high accuracy through extensively annotated datasets,acquiring such large-scale labeled data remains challenging and cost-prohibitive in numerous real-world appli...Although conventional object detection methods achieve high accuracy through extensively annotated datasets,acquiring such large-scale labeled data remains challenging and cost-prohibitive in numerous real-world applications.Few-shot object detection presents a new research idea that aims to localize and classify objects in images using only limited annotated examples.However,the inherent challenge in few-shot object detection lies in the insufficient sample diversity to fully characterize the sample feature distribution,which consequently impacts model performance.Inspired by contrastive learning principles,we propose an Implicit Feature Contrastive Learning(IFCL)module to address this limitation and augment feature diversity for more robust representational learning.This module generates augmented support sample features in a mixed feature space and implicitly contrasts them with query Region of Interest(RoI)features.This approach facilitates more comprehensive learning of both intra-class feature similarity and inter-class feature diversity,thereby enhancing the model’s object classification and localization capabilities.Extensive experiments on PASCAL VOC show that our method achieves a respective improvement of 3.2%,1.8%,and 2.3%on 10-shot of three Novel Sets compared to the baseline model FPD.展开更多
Air quality estimation assesses the pollution level in the air,supports public health warnings,and is a valuable tool in environmental management.Although air sensors have proven helpful in this task,sensors are often...Air quality estimation assesses the pollution level in the air,supports public health warnings,and is a valuable tool in environmental management.Although air sensors have proven helpful in this task,sensors are often expensive and difficult to install,while cameras are becoming more popular and accessible,from which images can be collected as data for deep learning models to solve the above task.This leads to another problem:several labeled images are needed to achieve high accuracy when deep-learningmodels predict air quality.In this research,we have threemain contributions:(1)Collect and publish an air quality estimation dataset,namely PTIT_AQED,including environmental image data and air quality;(2)Propose a deep learning model to predict air quality with few data,called PTIT_FAQE(PTIT Few-shot air quality estimation).We build PTIT_FAQE based on EfficientNet-a CNN architecture that ensures high performance in deep learning applications and Few-shot Learning with Prototypical Networks.This helps the model use only a fewtraining data but still achieve high accuracy in air quality estimation.And(3)conduct experiments to prove the superiority of PTIT_FAQE compared to other studies on both PTIT_AQED and APIN datasets.The results show that our model achieves an accuracy of 0.9278 and an F1-Score of 0.9139 on the PTIT_AQED dataset and an accuracy of 0.9467 and an F1-Score of 0.9371 on the APIN dataset,which demonstrate a significant performance improvement compared to previous studies.We also conduct detailed experiments to evaluate the impact of each component on model performance.展开更多
Full ceramic bearings are mission-critical components in oil-free environments,such as food processing,semiconductor manufacturing,and medical applications.Developing effective fault diagnosis methods for these bearin...Full ceramic bearings are mission-critical components in oil-free environments,such as food processing,semiconductor manufacturing,and medical applications.Developing effective fault diagnosis methods for these bearings is essential to ensuring operational reliability and preventing costly failures.Traditional supervised deep learning approaches have demonstrated promise in fault detection,but their dependence on large labeled datasets poses significant challenges in industrial settings where fault-labeled data is scarce.This paper introduces a few-shot learning approach for full ceramic bearing fault diagnosis by leveraging the pre-trained GPT-2 model.Large language models(LLMs)like GPT-2,pre-trained on diverse textual data,exhibit remarkable transfer learning and few-shot learning capabilities,making them ideal for applications with limited labeled data.In this study,acoustic emission(AE)signals from bearings were processed using empirical mode decomposition(EMD),and the extracted AE features were converted into structured text for fine-tuning GPT-2 as a fault classifier.To enhance its performance,we incorporated a modified loss function and softmax activation with cosine similarity,ensuring better generalization in fault identification.Experimental evaluations on a laboratory-collected full ceramic bearing dataset demonstrated that the proposed approach achieved high diagnostic accuracy with as few as five labeled samples,outperforming conventional methods such as k-nearest neighbor(KNN),large memory storage and retrieval(LAMSTAR)neural network,deep neural network(DNN),recurrent neural network(RNN),long short-term memory(LSTM)network,and model-agnostic meta-learning(MAML).The results highlight LLMs’potential to revolutionize fault diagnosis,enabling faster deployment,reduced reliance on extensive labeled datasets,and improved adaptability in industrial monitoring systems.展开更多
Intelligent fault diagnosis technology plays an indispensable role in ensuring the safety,stability,and efficiency of railway operations.However,existing studies have the following limitations.1)They are typical black-...Intelligent fault diagnosis technology plays an indispensable role in ensuring the safety,stability,and efficiency of railway operations.However,existing studies have the following limitations.1)They are typical black-box models that lacks interpretability as well as they fuse features by simply stacking them,overlooking the discrepancies in the importance of different features,which reduces the credibility and diagnosis accuracy of the models.2)They ignore the effects of potentially mistaken labels in the training datasets disrupting the ability of the models to learn the true data distribution,which degrades the generalization performance of intelligent diagnosis models,especially when the training samples are limited.To address the above items,an interpretable few-shot framework for fault diagnosis with noisy labels is proposed for train transmission systems.In the proposed framework,a feature extractor is constructed by stacked frequency band focus modules,which can capture signal features in different frequency bands and further adaptively concentrate on the features corresponding to the potential fault characteristic frequency.Then,according to prototypical network,a novel metric-based classifier is developed that is tolerant to mislabeled support samples in the case of limited samples.Besides,a new loss function is designed to decrease the impact of label mistakes in query datasets.Finally,fault simulation experiments of subway train transmission systems are designed and conducted,and the effectiveness as well as superiority of the proposed method are proved by ablation experiments and comparison with the existing methods.展开更多
Semantic segmentation of novel object categories with limited labeled data remains a challenging problem in computer vision.Few-shot segmentation methods aim to address this problem by recognizing objects from specifi...Semantic segmentation of novel object categories with limited labeled data remains a challenging problem in computer vision.Few-shot segmentation methods aim to address this problem by recognizing objects from specific target classes with a few provided examples.Previous approaches for few-shot semantic segmentation typically represent target classes using class prototypes.These prototypes are matched with the features of the query set to get segmentation results.However,class prototypes are usually obtained by applying global average pooling on masked support images.Global pooling discards much structural information,which may reduce the accuracy of model predictions.To address this issue,we propose a Category-Guided Frequency Modulation(CGFM)method.CGFM is designed to learn category-specific information in the frequency space and leverage it to provide a twostage guidance for the segmentation process.First,to self-adaptively activate class-relevant frequency bands while suppressing irrelevant ones,we leverage the Dual-Perception Gaussian Band Pre-activation(DPGBP)module to generate Gaussian filters using class embedding vectors.Second,to further enhance category-relevant frequency components in activated bands,we design a Support-Guided Category Response Enhancement(SGCRE)module to effectively introduce support frequency components into the modulation of query frequency features.Experiments on the PASCAL-5^(i) and COCO-20^(i) datasets demonstrate the promising performance of our model.展开更多
Few-shot point cloud 3D object detection(FS3D)aims to identify and locate objects of novel classes within point clouds using knowledge acquired from annotated base classes and a minimal number of samples from the nove...Few-shot point cloud 3D object detection(FS3D)aims to identify and locate objects of novel classes within point clouds using knowledge acquired from annotated base classes and a minimal number of samples from the novel classes.Due to imbalanced training data,existing FS3D methods based on fully supervised learning can lead to overfitting toward base classes,which impairs the network’s ability to generalize knowledge learned from base classes to novel classes and also prevents the network from extracting distinctive foreground and background representations for novel class objects.To address these issues,this thesis proposes a category-agnostic contrastive learning approach,enhancing the generalization and identification abilities for almost unseen categories through the construction of pseudo-labels and positive-negative sample pairs unrelated to specific classes.Firstly,this thesis designs a proposal-wise context contrastive module(CCM).By reducing the distance between foreground point features and increasing the distance between foreground and background point features within a region proposal,CCM aids the network in extracting more discriminative foreground and background feature representations without reliance on categorical annotations.Secondly,this thesis utilizes a geometric contrastive module(GCM),which enhances the network’s geometric perception capability by employing contrastive learning on the foreground point features associated with various basic geometric components,such as edges,corners,and surfaces,thereby enabling these geometric components to exhibit more distinguishable representations.This thesis also combines category-aware contrastive learning with former modules to maintain categorical distinctiveness.Extensive experimental results on FS-SUNRGBD and FS-ScanNet datasets demonstrate the effectiveness of this method with average precision exceeding the baseline by up to 8%.展开更多
The key to the success of few-shot semantic segmentation(FSS)depends on the efficient use of limited annotated support set to accurately segment novel classes in the query set.Due to the few samples in the support set...The key to the success of few-shot semantic segmentation(FSS)depends on the efficient use of limited annotated support set to accurately segment novel classes in the query set.Due to the few samples in the support set,FSS faces challenges such as intra-class differences,background(BG)mismatches between query and support sets,and ambiguous segmentation between the foreground(FG)and BG in the query set.To address these issues,The paper propose a multi-module network called CAMSNet,which includes four modules:the General Information Module(GIM),the Class Activation Map Aggregation(CAMA)module,the Self-Cross Attention(SCA)Block,and the Feature Fusion Module(FFM).In CAMSNet,The GIM employs an improved triplet loss,which concatenates word embedding vectors and support prototypes as anchors,and uses local support features of FG and BG as positive and negative samples to help solve the problem of intra-class differences.Then for the first time,the Class Activation Map(CAM)from the Weakly Supervised Semantic Segmentation(WSSS)is applied to FSS within the CAMA module.This method replaces the traditional use of cosine similarity to locate query information.Subsequently,the SCA Block processes the support and query features aggregated by the CAMA module,significantly enhancing the understanding of input information,leading to more accurate predictions and effectively addressing BG mismatch and ambiguous FG-BG segmentation.Finally,The FFM combines general class information with the enhanced query information to achieve accurate segmentation of the query image.Extensive Experiments on PASCAL and COCO demonstrate that-5i-20ithe CAMSNet yields superior performance and set a state-of-the-art.展开更多
This paper focuses on the task of few-shot 3D point cloud semantic segmentation.Despite some progress,this task still encounters many issues due to the insufficient samples given,e.g.,incomplete object segmentation an...This paper focuses on the task of few-shot 3D point cloud semantic segmentation.Despite some progress,this task still encounters many issues due to the insufficient samples given,e.g.,incomplete object segmentation and inaccurate semantic discrimination.To tackle these issues,we first leverage part-whole relationships into the task of 3D point cloud semantic segmentation to capture semantic integrity,which is empowered by the dynamic capsule routing with the module of 3D Capsule Networks(CapsNets)in the embedding network.Concretely,the dynamic routing amalgamates geometric information of the 3D point cloud data to construct higher-level feature representations,which capture the relationships between object parts and their wholes.Secondly,we designed a multi-prototype enhancement module to enhance the prototype discriminability.Specifically,the single-prototype enhancement mechanism is expanded to the multi-prototype enhancement version for capturing rich semantics.Besides,the shot-correlation within the category is calculated via the interaction of different samples to enhance the intra-category similarity.Ablation studies prove that the involved part-whole relations and proposed multi-prototype enhancement module help to achieve complete object segmentation and improve semantic discrimination.Moreover,under the integration of these two modules,quantitative and qualitative experiments on two public benchmarks,including S3DIS and ScanNet,indicate the superior performance of the proposed framework on the task of 3D point cloud semantic segmentation,compared to some state-of-the-art methods.展开更多
In the metric-based meta-learning detection model,the distribution of training samples in the metric space has great influence on the detection performance,and this influence is usually ignored by traditional meta-det...In the metric-based meta-learning detection model,the distribution of training samples in the metric space has great influence on the detection performance,and this influence is usually ignored by traditional meta-detectors.In addition,the design of metric space might be interfered with by the background noise of training samples.To tackle these issues,we propose a metric space optimisation method based on hyperbolic geometry attention and class-agnostic activation maps.First,the geometric properties of hyperbolic spaces to establish a structured metric space are used.A variety of feature samples of different classes are embedded into the hyperbolic space with extremely low distortion.This metric space is more suitable for representing tree-like structures between categories for image scene analysis.Meanwhile,a novel similarity measure function based on Poincarédistance is proposed to evaluate the distance of various types of objects in the feature space.In addition,the class-agnostic activation maps(CCAMs)are employed to re-calibrate the weight of foreground feature information and suppress background information.Finally,the decoder processes the high-level feature information as the decoding of the query object and detects objects by predicting their locations and corresponding task encodings.Experimental evaluation is conducted on Pascal VOC and MS COCO datasets.The experiment results show that the effectiveness of the authors’method surpasses the performance baseline of the excellent few-shot detection models.展开更多
Large Language Models(LLMs)are increasingly demonstrating their ability to understand natural language and solve complex tasks,especially through text generation.One of the relevant capabilities is contextual learning...Large Language Models(LLMs)are increasingly demonstrating their ability to understand natural language and solve complex tasks,especially through text generation.One of the relevant capabilities is contextual learning,which involves the ability to receive instructions in natural language or task demonstrations to generate expected outputs for test instances without the need for additional training or gradient updates.In recent years,the popularity of social networking has provided a medium through which some users can engage in offensive and harmful online behavior.In this study,we investigate the ability of different LLMs,ranging from zero-shot and few-shot learning to fine-tuning.Our experiments show that LLMs can identify sexist and hateful online texts using zero-shot and few-shot approaches through information retrieval.Furthermore,it is found that the encoder-decoder model called Zephyr achieves the best results with the fine-tuning approach,scoring 86.811%on the Explainable Detection of Online Sexism(EDOS)test-set and 57.453%on the Multilingual Detection of Hate Speech Against Immigrants and Women in Twitter(HatEval)test-set.Finally,it is confirmed that the evaluated models perform well in hate text detection,as they beat the best result in the HatEval task leaderboard.The error analysis shows that contextual learning had difficulty distinguishing between types of hate speech and figurative language.However,the fine-tuned approach tends to produce many false positives.展开更多
With the successive application of deep learning(DL)in classification tasks,the DL-based modulation classification method has become the preference for its state-of-the-art performance.Nevertheless,once the DL recogni...With the successive application of deep learning(DL)in classification tasks,the DL-based modulation classification method has become the preference for its state-of-the-art performance.Nevertheless,once the DL recognition model is pre-trained with fixed classes,the pre-trained model tends to predict incorrect results when identifying incremental classes.Moreover,the incremental classes are usually emergent without label information or only a few labeled samples of incremental classes can be obtained.In this context,we propose a graphbased semi-supervised approach to address the fewshot classes-incremental(FSCI)modulation classification problem.Our proposed method is a twostage learning method,specifically,a warm-up model is trained for classifying old classes and incremental classes,where the unlabeled samples of incremental classes are uniformly labeled with the same label to alleviate the damage of the class imbalance problem.Then the warm-up model is regarded as a feature extractor for constructing a similar graph to connect labeled samples and unlabeled samples,and the label propagation algorithm is adopted to propagate the label information from labeled nodes to unlabeled nodes in the graph to achieve the purpose of incremental classes recognition.Simulation results prove that the proposed method is superior to other finetuning methods and retrain methods.展开更多
Due to the limited computational capability and the diversity of the Internet of Things devices working in different environment,we consider fewshot learning-based automatic modulation classification(AMC)to improve it...Due to the limited computational capability and the diversity of the Internet of Things devices working in different environment,we consider fewshot learning-based automatic modulation classification(AMC)to improve its reliability.A data enhancement module(DEM)is designed by a convolutional layer to supplement frequency-domain information as well as providing nonlinear mapping that is beneficial for AMC.Multimodal network is designed to have multiple residual blocks,where each residual block has multiple convolutional kernels of different sizes for diverse feature extraction.Moreover,a deep supervised loss function is designed to supervise all parts of the network including the hidden layers and the DEM.Since different model may output different results,cooperative classifier is designed to avoid the randomness of single model and improve the reliability.Simulation results show that this few-shot learning-based AMC method can significantly improve the AMC accuracy compared to the existing methods.展开更多
In order to improve the models capability in expressing features during few-shot learning,a multi-scale features prototypical network(MS-PN)algorithm is proposed.The metric learning algo-rithm is employed to extract i...In order to improve the models capability in expressing features during few-shot learning,a multi-scale features prototypical network(MS-PN)algorithm is proposed.The metric learning algo-rithm is employed to extract image features and project them into a feature space,thus evaluating the similarity between samples based on their relative distances within the metric space.To sufficiently extract feature information from limited sample data and mitigate the impact of constrained data vol-ume,a multi-scale feature extraction network is presented to capture data features at various scales during the process of image feature extraction.Additionally,the position of the prototype is fine-tuned by assigning weights to data points to mitigate the influence of outliers on the experiment.The loss function integrates contrastive loss and label-smoothing to bring similar data points closer and separate dissimilar data points within the metric space.Experimental evaluations are conducted on small-sample datasets mini-ImageNet and CUB200-2011.The method in this paper can achieve higher classification accuracy.Specifically,in the 5-way 1-shot experiment,classification accuracy reaches 50.13%and 66.79%respectively on these two datasets.Moreover,in the 5-way 5-shot ex-periment,accuracy of 66.79%and 85.91%are observed,respectively.展开更多
Deep learning has been fully verified and accepted in the field of electromagnetic signal classification. However, in many specific scenarios, such as radio resource management for aircraft communications, labeled dat...Deep learning has been fully verified and accepted in the field of electromagnetic signal classification. However, in many specific scenarios, such as radio resource management for aircraft communications, labeled data are difficult to obtain, which makes the best deep learning methods at present seem almost powerless, because these methods need a large amount of labeled data for training. When the training dataset is small, it is highly possible to fall into overfitting, which causes performance degradation of the deep neural network. For few-shot electromagnetic signal classification, data augmentation is one of the most intuitive countermeasures. In this work, a generative adversarial network based on the data augmentation method is proposed to achieve better classification performance for electromagnetic signals. Based on the similarity principle, a screening mechanism is established to obtain high-quality generated signals. Then, a data union augmentation algorithm is designed by introducing spatiotemporally flipped shapes of the signal. To verify the effectiveness of the proposed data augmentation algorithm, experiments are conducted on the RADIOML 2016.04C dataset and real-world ACARS dataset. The experimental results show that the proposed method significantly improves the performance of few-shot electromagnetic signal classification.展开更多
Intrusion Detection Systems(IDSs)have a great interest these days to discover complex attack events and protect the critical infrastructures of the Internet of Things(IoT)networks.Existing IDSs based on shallow and de...Intrusion Detection Systems(IDSs)have a great interest these days to discover complex attack events and protect the critical infrastructures of the Internet of Things(IoT)networks.Existing IDSs based on shallow and deep network architectures demand high computational resources and high volumes of data to establish an adaptive detection engine that discovers new families of attacks from the edge of IoT networks.However,attackers exploit network gateways at the edge using new attacking scenarios(i.e.,zero-day attacks),such as ransomware and Distributed Denial of Service(DDoS)attacks.This paper proposes new IDS based on Few-Shot Deep Learning,named CNN-IDS,which can automatically identify zero-day attacks from the edge of a network and protect its IoT systems.The proposed system comprises two-methodological stages:1)a filtered Information Gain method is to select the most useful features from network data,and 2)one-dimensional Convolutional Neural Network(CNN)algorithm is to recognize new attack types from a network’s edge.The proposed model is trained and validated using two datasets of the UNSW-NB15 and Bot-IoT.The experimental results showed that it enhances about a 3%detection rate and around a 3%–4%falsepositive rate with the UNSW-NB15 dataset and about an 8%detection rate using the BoT-IoT dataset.展开更多
Few-shot semantic segmentation aims at training a model that can segment novel classes in a query image with only a few densely annotated support exemplars.It remains a challenge because of large intra-class variation...Few-shot semantic segmentation aims at training a model that can segment novel classes in a query image with only a few densely annotated support exemplars.It remains a challenge because of large intra-class variations between the support and query images.Existing approaches utilize 4D convolutions to mine semantic correspondence between the support and query images.However,they still suffer from heavy computation,sparse correspondence,and large memory.We propose axial assembled correspondence network(AACNet)to alleviate these issues.The key point of AACNet is the proposed axial assembled 4D kernel,which constructs the basic block for semantic correspondence encoder(SCE).Furthermore,we propose the deblurring equations to provide more robust correspondence for the aforementioned SCE and design a novel fusion module to mix correspondences in a learnable manner.Experiments on PASCAL-5~i reveal that our AACNet achieves a mean intersection-over-union score of 65.9%for 1-shot segmentation and 70.6%for 5-shot segmentation,surpassing the state-of-the-art method by 5.8%and 5.0%respectively.展开更多
Existing almost deep learning methods rely on a large amount of annotated data, so they are inappropriate for forest fire smoke detection with limited data. In this paper, a novel hybrid attention-based few-shot learn...Existing almost deep learning methods rely on a large amount of annotated data, so they are inappropriate for forest fire smoke detection with limited data. In this paper, a novel hybrid attention-based few-shot learning method, named Attention-Based Prototypical Network, is proposed for forest fire smoke detection. Specifically, feature extraction network, which consists of convolutional block attention module, could extract high-level and discriminative features and further decrease the false alarm rate resulting from suspected smoke areas. Moreover, we design a metalearning module to alleviate the overfitting issue caused by limited smoke images, and the meta-learning network enables achieving effective detection via comparing the distance between the class prototype of support images and the features of query images. A series of experiments on forest fire smoke datasets and miniImageNet dataset testify that the proposed method is superior to state-of-the-art few-shot learning approaches.展开更多
基金Supported by the National Pre-research Program during the 14th Five-Year Plan(514010405)。
文摘In response to the scarcity of infrared aircraft samples and the tendency of traditional deep learning to overfit,a few-shot infrared aircraft classification method based on cross-correlation networks is proposed.This method combines two core modules:a simple parameter-free self-attention and cross-attention.By analyzing the self-correlation and cross-correlation between support images and query images,it achieves effective classification of infrared aircraft under few-shot conditions.The proposed cross-correlation network integrates these two modules and is trained in an end-to-end manner.The simple parameter-free self-attention is responsible for extracting the internal structure of the image while the cross-attention can calculate the cross-correlation between images further extracting and fusing the features between images.Compared with existing few-shot infrared target classification models,this model focuses on the geometric structure and thermal texture information of infrared images by modeling the semantic relevance between the features of the support set and query set,thus better attending to the target objects.Experimental results show that this method outperforms existing infrared aircraft classification methods in various classification tasks,with the highest classification accuracy improvement exceeding 3%.In addition,ablation experiments and comparative experiments also prove the effectiveness of the method.
基金supported by the National Natural Science Foundation of China(52274055)the Shandong Provincial Natural Science Foundation(ZR2022YQ50)the Taishan Scholar Program of Shandong Province(tsqn202408088)。
文摘Predicting the productivity of multistage fractured horizontal wells plays an important role in exploiting unconventional resources.In recent years,machine learning(ML)models have emerged as a new approach for such studies.However,the scarcity of sufficient real data for model training often leads to imprecise predictions,even though the models trained with real data better characterize geological and engineering features.To tackle this issue,we propose an ML model that can obtain reliable results even with a small amount of data samples.Our model integrates the synthetic minority oversampling technique(SMOTE)to expand the data volume,the support vector machine(SVM)for model training,and the particle swarm optimization(PSO)algorithm for optimizing hyperparameters.To enhance the model performance,we conduct feature fusion and dimensionality reduction.Additionally,we examine the influences of different sample sizes and ML models for training.The proposed model demonstrates higher prediction accuracy and generalization ability,achieving a predicted R^(2)value of up to 0.9 for the test set,compared to the traditional ML techniques with an R^(2)of 0.13.This model accurately predicts the production of fractured horizontal wells even with limited samples,supplying an efficient tool for optimizing the production of unconventional resources.Importantly,the model holds the potential applicability to address similar challenges in other fields constrained by scarce data samples.
基金supported by Suzhou Science and Technology Plan(Basic Research)Project under Grant SJC2023002Postgraduate Research&Practice Innovation Program of Jiangsu Province under Grant KYCX23_3322.
文摘Image classification is crucial for various applications,including digital construction,smart manu-facturing,and medical imaging.Focusing on the inadequate model generalization and data privacy concerns in few-shot image classification,in this paper,we propose a federated learning approach that incorporates privacy-preserving techniques.First,we utilize contrastive learning to train on local few-shot image data and apply various data augmentation methods to expand the sample size,thereby enhancing the model’s generalization capabilities in few-shot contexts.Second,we introduce local differential privacy techniques and weight pruning methods to safeguard model parameters,perturbing the transmitted parameters to ensure user data privacy.Finally,numerical simulations are conducted to demonstrate the effectiveness of our proposed method.The results indicate that our approach significantly enhances model generalization and test accuracy compared to several popular federated learning algorithms while maintaining data privacy,highlighting its effectiveness and practicality in addressing the challenges of model generalization and data privacy in few-shot image scenarios.
基金funded by the China Chongqing Municipal Science and Technology Bureau,grant numbers CSTB2024TIAD-CYKJCXX0009,CSTB2024NSCQ-LZX0043,CSTB2022NSCQ-MSX0288Chongqing Municipal Commission of Housing and Urban-Rural Development,grant number CKZ2024-87+3 种基金the Chongqing University of Technology Graduate Education High-Quality Development Project,grant number gzlsz202401the Chongqing University of Technology—Chongqing LINGLUE Technology Co.,Ltd.Electronic Information(Artificial Intelligence)Graduate Joint Training Basethe Postgraduate Education and Teaching Reform Research Project in Chongqing,grant number yjg213116the Chongqing University of Technology-CISDI Chongqing Information Technology Co.,Ltd.Computer Technology Graduate Joint Training Base.
文摘Although conventional object detection methods achieve high accuracy through extensively annotated datasets,acquiring such large-scale labeled data remains challenging and cost-prohibitive in numerous real-world applications.Few-shot object detection presents a new research idea that aims to localize and classify objects in images using only limited annotated examples.However,the inherent challenge in few-shot object detection lies in the insufficient sample diversity to fully characterize the sample feature distribution,which consequently impacts model performance.Inspired by contrastive learning principles,we propose an Implicit Feature Contrastive Learning(IFCL)module to address this limitation and augment feature diversity for more robust representational learning.This module generates augmented support sample features in a mixed feature space and implicitly contrasts them with query Region of Interest(RoI)features.This approach facilitates more comprehensive learning of both intra-class feature similarity and inter-class feature diversity,thereby enhancing the model’s object classification and localization capabilities.Extensive experiments on PASCAL VOC show that our method achieves a respective improvement of 3.2%,1.8%,and 2.3%on 10-shot of three Novel Sets compared to the baseline model FPD.
文摘Air quality estimation assesses the pollution level in the air,supports public health warnings,and is a valuable tool in environmental management.Although air sensors have proven helpful in this task,sensors are often expensive and difficult to install,while cameras are becoming more popular and accessible,from which images can be collected as data for deep learning models to solve the above task.This leads to another problem:several labeled images are needed to achieve high accuracy when deep-learningmodels predict air quality.In this research,we have threemain contributions:(1)Collect and publish an air quality estimation dataset,namely PTIT_AQED,including environmental image data and air quality;(2)Propose a deep learning model to predict air quality with few data,called PTIT_FAQE(PTIT Few-shot air quality estimation).We build PTIT_FAQE based on EfficientNet-a CNN architecture that ensures high performance in deep learning applications and Few-shot Learning with Prototypical Networks.This helps the model use only a fewtraining data but still achieve high accuracy in air quality estimation.And(3)conduct experiments to prove the superiority of PTIT_FAQE compared to other studies on both PTIT_AQED and APIN datasets.The results show that our model achieves an accuracy of 0.9278 and an F1-Score of 0.9139 on the PTIT_AQED dataset and an accuracy of 0.9467 and an F1-Score of 0.9371 on the APIN dataset,which demonstrate a significant performance improvement compared to previous studies.We also conduct detailed experiments to evaluate the impact of each component on model performance.
文摘Full ceramic bearings are mission-critical components in oil-free environments,such as food processing,semiconductor manufacturing,and medical applications.Developing effective fault diagnosis methods for these bearings is essential to ensuring operational reliability and preventing costly failures.Traditional supervised deep learning approaches have demonstrated promise in fault detection,but their dependence on large labeled datasets poses significant challenges in industrial settings where fault-labeled data is scarce.This paper introduces a few-shot learning approach for full ceramic bearing fault diagnosis by leveraging the pre-trained GPT-2 model.Large language models(LLMs)like GPT-2,pre-trained on diverse textual data,exhibit remarkable transfer learning and few-shot learning capabilities,making them ideal for applications with limited labeled data.In this study,acoustic emission(AE)signals from bearings were processed using empirical mode decomposition(EMD),and the extracted AE features were converted into structured text for fine-tuning GPT-2 as a fault classifier.To enhance its performance,we incorporated a modified loss function and softmax activation with cosine similarity,ensuring better generalization in fault identification.Experimental evaluations on a laboratory-collected full ceramic bearing dataset demonstrated that the proposed approach achieved high diagnostic accuracy with as few as five labeled samples,outperforming conventional methods such as k-nearest neighbor(KNN),large memory storage and retrieval(LAMSTAR)neural network,deep neural network(DNN),recurrent neural network(RNN),long short-term memory(LSTM)network,and model-agnostic meta-learning(MAML).The results highlight LLMs’potential to revolutionize fault diagnosis,enabling faster deployment,reduced reliance on extensive labeled datasets,and improved adaptability in industrial monitoring systems.
基金supported in part by the National Key R&D Program of China under Grant 2022YFB4300601in part by the State Key Laboratory of Advanced Rail Autonomous Operation under Grant RAO2023ZZ003.
文摘Intelligent fault diagnosis technology plays an indispensable role in ensuring the safety,stability,and efficiency of railway operations.However,existing studies have the following limitations.1)They are typical black-box models that lacks interpretability as well as they fuse features by simply stacking them,overlooking the discrepancies in the importance of different features,which reduces the credibility and diagnosis accuracy of the models.2)They ignore the effects of potentially mistaken labels in the training datasets disrupting the ability of the models to learn the true data distribution,which degrades the generalization performance of intelligent diagnosis models,especially when the training samples are limited.To address the above items,an interpretable few-shot framework for fault diagnosis with noisy labels is proposed for train transmission systems.In the proposed framework,a feature extractor is constructed by stacked frequency band focus modules,which can capture signal features in different frequency bands and further adaptively concentrate on the features corresponding to the potential fault characteristic frequency.Then,according to prototypical network,a novel metric-based classifier is developed that is tolerant to mislabeled support samples in the case of limited samples.Besides,a new loss function is designed to decrease the impact of label mistakes in query datasets.Finally,fault simulation experiments of subway train transmission systems are designed and conducted,and the effectiveness as well as superiority of the proposed method are proved by ablation experiments and comparison with the existing methods.
文摘Semantic segmentation of novel object categories with limited labeled data remains a challenging problem in computer vision.Few-shot segmentation methods aim to address this problem by recognizing objects from specific target classes with a few provided examples.Previous approaches for few-shot semantic segmentation typically represent target classes using class prototypes.These prototypes are matched with the features of the query set to get segmentation results.However,class prototypes are usually obtained by applying global average pooling on masked support images.Global pooling discards much structural information,which may reduce the accuracy of model predictions.To address this issue,we propose a Category-Guided Frequency Modulation(CGFM)method.CGFM is designed to learn category-specific information in the frequency space and leverage it to provide a twostage guidance for the segmentation process.First,to self-adaptively activate class-relevant frequency bands while suppressing irrelevant ones,we leverage the Dual-Perception Gaussian Band Pre-activation(DPGBP)module to generate Gaussian filters using class embedding vectors.Second,to further enhance category-relevant frequency components in activated bands,we design a Support-Guided Category Response Enhancement(SGCRE)module to effectively introduce support frequency components into the modulation of query frequency features.Experiments on the PASCAL-5^(i) and COCO-20^(i) datasets demonstrate the promising performance of our model.
文摘Few-shot point cloud 3D object detection(FS3D)aims to identify and locate objects of novel classes within point clouds using knowledge acquired from annotated base classes and a minimal number of samples from the novel classes.Due to imbalanced training data,existing FS3D methods based on fully supervised learning can lead to overfitting toward base classes,which impairs the network’s ability to generalize knowledge learned from base classes to novel classes and also prevents the network from extracting distinctive foreground and background representations for novel class objects.To address these issues,this thesis proposes a category-agnostic contrastive learning approach,enhancing the generalization and identification abilities for almost unseen categories through the construction of pseudo-labels and positive-negative sample pairs unrelated to specific classes.Firstly,this thesis designs a proposal-wise context contrastive module(CCM).By reducing the distance between foreground point features and increasing the distance between foreground and background point features within a region proposal,CCM aids the network in extracting more discriminative foreground and background feature representations without reliance on categorical annotations.Secondly,this thesis utilizes a geometric contrastive module(GCM),which enhances the network’s geometric perception capability by employing contrastive learning on the foreground point features associated with various basic geometric components,such as edges,corners,and surfaces,thereby enabling these geometric components to exhibit more distinguishable representations.This thesis also combines category-aware contrastive learning with former modules to maintain categorical distinctiveness.Extensive experimental results on FS-SUNRGBD and FS-ScanNet datasets demonstrate the effectiveness of this method with average precision exceeding the baseline by up to 8%.
基金supported by funding from the following sources:National Natural Science Foundation of China(U1904119)Research Programs of Henan Science and Technology Department(232102210033,232102210054)+3 种基金Chongqing Natural Science Foundation(CSTB2023NSCQ-MSX0070)Henan Province Key Research and Development Project(231111212000)Aviation Science Foundation(20230001055002)supported by Henan Center for Outstanding Overseas Scientists(GZS2022011).
文摘The key to the success of few-shot semantic segmentation(FSS)depends on the efficient use of limited annotated support set to accurately segment novel classes in the query set.Due to the few samples in the support set,FSS faces challenges such as intra-class differences,background(BG)mismatches between query and support sets,and ambiguous segmentation between the foreground(FG)and BG in the query set.To address these issues,The paper propose a multi-module network called CAMSNet,which includes four modules:the General Information Module(GIM),the Class Activation Map Aggregation(CAMA)module,the Self-Cross Attention(SCA)Block,and the Feature Fusion Module(FFM).In CAMSNet,The GIM employs an improved triplet loss,which concatenates word embedding vectors and support prototypes as anchors,and uses local support features of FG and BG as positive and negative samples to help solve the problem of intra-class differences.Then for the first time,the Class Activation Map(CAM)from the Weakly Supervised Semantic Segmentation(WSSS)is applied to FSS within the CAMA module.This method replaces the traditional use of cosine similarity to locate query information.Subsequently,the SCA Block processes the support and query features aggregated by the CAMA module,significantly enhancing the understanding of input information,leading to more accurate predictions and effectively addressing BG mismatch and ambiguous FG-BG segmentation.Finally,The FFM combines general class information with the enhanced query information to achieve accurate segmentation of the query image.Extensive Experiments on PASCAL and COCO demonstrate that-5i-20ithe CAMSNet yields superior performance and set a state-of-the-art.
基金This work is supported by the National Natural Science Foundation of China under Grant No.62001341the National Natural Science Foundation of Jiangsu Province under Grant No.BK20221379the Jiangsu Engineering Research Center of Digital Twinning Technology for Key Equipment in Petrochemical Process under Grant No.DTEC202104.
文摘This paper focuses on the task of few-shot 3D point cloud semantic segmentation.Despite some progress,this task still encounters many issues due to the insufficient samples given,e.g.,incomplete object segmentation and inaccurate semantic discrimination.To tackle these issues,we first leverage part-whole relationships into the task of 3D point cloud semantic segmentation to capture semantic integrity,which is empowered by the dynamic capsule routing with the module of 3D Capsule Networks(CapsNets)in the embedding network.Concretely,the dynamic routing amalgamates geometric information of the 3D point cloud data to construct higher-level feature representations,which capture the relationships between object parts and their wholes.Secondly,we designed a multi-prototype enhancement module to enhance the prototype discriminability.Specifically,the single-prototype enhancement mechanism is expanded to the multi-prototype enhancement version for capturing rich semantics.Besides,the shot-correlation within the category is calculated via the interaction of different samples to enhance the intra-category similarity.Ablation studies prove that the involved part-whole relations and proposed multi-prototype enhancement module help to achieve complete object segmentation and improve semantic discrimination.Moreover,under the integration of these two modules,quantitative and qualitative experiments on two public benchmarks,including S3DIS and ScanNet,indicate the superior performance of the proposed framework on the task of 3D point cloud semantic segmentation,compared to some state-of-the-art methods.
基金National Natural Science Foundation of China,Grant/Award Number:61602157Henan scientific and technological project,Grant/Award Number:242102210020Basal Research Fund,Grant/Award Number:NSFRF240618。
文摘In the metric-based meta-learning detection model,the distribution of training samples in the metric space has great influence on the detection performance,and this influence is usually ignored by traditional meta-detectors.In addition,the design of metric space might be interfered with by the background noise of training samples.To tackle these issues,we propose a metric space optimisation method based on hyperbolic geometry attention and class-agnostic activation maps.First,the geometric properties of hyperbolic spaces to establish a structured metric space are used.A variety of feature samples of different classes are embedded into the hyperbolic space with extremely low distortion.This metric space is more suitable for representing tree-like structures between categories for image scene analysis.Meanwhile,a novel similarity measure function based on Poincarédistance is proposed to evaluate the distance of various types of objects in the feature space.In addition,the class-agnostic activation maps(CCAMs)are employed to re-calibrate the weight of foreground feature information and suppress background information.Finally,the decoder processes the high-level feature information as the decoding of the query object and detects objects by predicting their locations and corresponding task encodings.Experimental evaluation is conducted on Pascal VOC and MS COCO datasets.The experiment results show that the effectiveness of the authors’method surpasses the performance baseline of the excellent few-shot detection models.
基金This work is part of the research projects LaTe4PoliticES(PID2022-138099OBI00)funded by MICIU/AEI/10.13039/501100011033the European Regional Development Fund(ERDF)-A Way of Making Europe and LT-SWM(TED2021-131167B-I00)funded by MICIU/AEI/10.13039/501100011033the European Union NextGenerationEU/PRTR.Mr.Ronghao Pan is supported by the Programa Investigo grant,funded by the Region of Murcia,the Spanish Ministry of Labour and Social Economy and the European Union-NextGenerationEU under the“Plan de Recuperación,Transformación y Resiliencia(PRTR).”。
文摘Large Language Models(LLMs)are increasingly demonstrating their ability to understand natural language and solve complex tasks,especially through text generation.One of the relevant capabilities is contextual learning,which involves the ability to receive instructions in natural language or task demonstrations to generate expected outputs for test instances without the need for additional training or gradient updates.In recent years,the popularity of social networking has provided a medium through which some users can engage in offensive and harmful online behavior.In this study,we investigate the ability of different LLMs,ranging from zero-shot and few-shot learning to fine-tuning.Our experiments show that LLMs can identify sexist and hateful online texts using zero-shot and few-shot approaches through information retrieval.Furthermore,it is found that the encoder-decoder model called Zephyr achieves the best results with the fine-tuning approach,scoring 86.811%on the Explainable Detection of Online Sexism(EDOS)test-set and 57.453%on the Multilingual Detection of Hate Speech Against Immigrants and Women in Twitter(HatEval)test-set.Finally,it is confirmed that the evaluated models perform well in hate text detection,as they beat the best result in the HatEval task leaderboard.The error analysis shows that contextual learning had difficulty distinguishing between types of hate speech and figurative language.However,the fine-tuned approach tends to produce many false positives.
基金supported in part by the National Natural Science Foundation of China under Grant No.62171334,No.11973077 and No.12003061。
文摘With the successive application of deep learning(DL)in classification tasks,the DL-based modulation classification method has become the preference for its state-of-the-art performance.Nevertheless,once the DL recognition model is pre-trained with fixed classes,the pre-trained model tends to predict incorrect results when identifying incremental classes.Moreover,the incremental classes are usually emergent without label information or only a few labeled samples of incremental classes can be obtained.In this context,we propose a graphbased semi-supervised approach to address the fewshot classes-incremental(FSCI)modulation classification problem.Our proposed method is a twostage learning method,specifically,a warm-up model is trained for classifying old classes and incremental classes,where the unlabeled samples of incremental classes are uniformly labeled with the same label to alleviate the damage of the class imbalance problem.Then the warm-up model is regarded as a feature extractor for constructing a similar graph to connect labeled samples and unlabeled samples,and the label propagation algorithm is adopted to propagate the label information from labeled nodes to unlabeled nodes in the graph to achieve the purpose of incremental classes recognition.Simulation results prove that the proposed method is superior to other finetuning methods and retrain methods.
基金supported in part by National Key Research and Development Program of China under Grant 2021YFB2900404.
文摘Due to the limited computational capability and the diversity of the Internet of Things devices working in different environment,we consider fewshot learning-based automatic modulation classification(AMC)to improve its reliability.A data enhancement module(DEM)is designed by a convolutional layer to supplement frequency-domain information as well as providing nonlinear mapping that is beneficial for AMC.Multimodal network is designed to have multiple residual blocks,where each residual block has multiple convolutional kernels of different sizes for diverse feature extraction.Moreover,a deep supervised loss function is designed to supervise all parts of the network including the hidden layers and the DEM.Since different model may output different results,cooperative classifier is designed to avoid the randomness of single model and improve the reliability.Simulation results show that this few-shot learning-based AMC method can significantly improve the AMC accuracy compared to the existing methods.
基金the Scientific Research Foundation of Liaoning Provincial Department of Education(No.LJKZ0139)the Program for Liaoning Excellent Talents in University(No.LR15045).
文摘In order to improve the models capability in expressing features during few-shot learning,a multi-scale features prototypical network(MS-PN)algorithm is proposed.The metric learning algo-rithm is employed to extract image features and project them into a feature space,thus evaluating the similarity between samples based on their relative distances within the metric space.To sufficiently extract feature information from limited sample data and mitigate the impact of constrained data vol-ume,a multi-scale feature extraction network is presented to capture data features at various scales during the process of image feature extraction.Additionally,the position of the prototype is fine-tuned by assigning weights to data points to mitigate the influence of outliers on the experiment.The loss function integrates contrastive loss and label-smoothing to bring similar data points closer and separate dissimilar data points within the metric space.Experimental evaluations are conducted on small-sample datasets mini-ImageNet and CUB200-2011.The method in this paper can achieve higher classification accuracy.Specifically,in the 5-way 1-shot experiment,classification accuracy reaches 50.13%and 66.79%respectively on these two datasets.Moreover,in the 5-way 5-shot ex-periment,accuracy of 66.79%and 85.91%are observed,respectively.
基金supported by the National Natural Science Foundation of China(No.61772401)the Fundamental Research Funds for the Central Universities(No.RW180177)supported by the Science and Technology on Communication Information Security Control Laboratory。
文摘Deep learning has been fully verified and accepted in the field of electromagnetic signal classification. However, in many specific scenarios, such as radio resource management for aircraft communications, labeled data are difficult to obtain, which makes the best deep learning methods at present seem almost powerless, because these methods need a large amount of labeled data for training. When the training dataset is small, it is highly possible to fall into overfitting, which causes performance degradation of the deep neural network. For few-shot electromagnetic signal classification, data augmentation is one of the most intuitive countermeasures. In this work, a generative adversarial network based on the data augmentation method is proposed to achieve better classification performance for electromagnetic signals. Based on the similarity principle, a screening mechanism is established to obtain high-quality generated signals. Then, a data union augmentation algorithm is designed by introducing spatiotemporally flipped shapes of the signal. To verify the effectiveness of the proposed data augmentation algorithm, experiments are conducted on the RADIOML 2016.04C dataset and real-world ACARS dataset. The experimental results show that the proposed method significantly improves the performance of few-shot electromagnetic signal classification.
基金This work has been supported by the Australian Research Data Common(ARDC),project code–RG192500.
文摘Intrusion Detection Systems(IDSs)have a great interest these days to discover complex attack events and protect the critical infrastructures of the Internet of Things(IoT)networks.Existing IDSs based on shallow and deep network architectures demand high computational resources and high volumes of data to establish an adaptive detection engine that discovers new families of attacks from the edge of IoT networks.However,attackers exploit network gateways at the edge using new attacking scenarios(i.e.,zero-day attacks),such as ransomware and Distributed Denial of Service(DDoS)attacks.This paper proposes new IDS based on Few-Shot Deep Learning,named CNN-IDS,which can automatically identify zero-day attacks from the edge of a network and protect its IoT systems.The proposed system comprises two-methodological stages:1)a filtered Information Gain method is to select the most useful features from network data,and 2)one-dimensional Convolutional Neural Network(CNN)algorithm is to recognize new attack types from a network’s edge.The proposed model is trained and validated using two datasets of the UNSW-NB15 and Bot-IoT.The experimental results showed that it enhances about a 3%detection rate and around a 3%–4%falsepositive rate with the UNSW-NB15 dataset and about an 8%detection rate using the BoT-IoT dataset.
基金supported in part by the Key Research and Development Program of Guangdong Province(2021B0101200001)the Guangdong Basic and Applied Basic Research Foundation(2020B1515120071)。
文摘Few-shot semantic segmentation aims at training a model that can segment novel classes in a query image with only a few densely annotated support exemplars.It remains a challenge because of large intra-class variations between the support and query images.Existing approaches utilize 4D convolutions to mine semantic correspondence between the support and query images.However,they still suffer from heavy computation,sparse correspondence,and large memory.We propose axial assembled correspondence network(AACNet)to alleviate these issues.The key point of AACNet is the proposed axial assembled 4D kernel,which constructs the basic block for semantic correspondence encoder(SCE).Furthermore,we propose the deblurring equations to provide more robust correspondence for the aforementioned SCE and design a novel fusion module to mix correspondences in a learnable manner.Experiments on PASCAL-5~i reveal that our AACNet achieves a mean intersection-over-union score of 65.9%for 1-shot segmentation and 70.6%for 5-shot segmentation,surpassing the state-of-the-art method by 5.8%and 5.0%respectively.
基金The work was supported by the National Key R&D Program of China(Grant No.2020YFC1511601)Fundamental Research Funds for the Central Universities(Grant No.2019SHFWLC01).
文摘Existing almost deep learning methods rely on a large amount of annotated data, so they are inappropriate for forest fire smoke detection with limited data. In this paper, a novel hybrid attention-based few-shot learning method, named Attention-Based Prototypical Network, is proposed for forest fire smoke detection. Specifically, feature extraction network, which consists of convolutional block attention module, could extract high-level and discriminative features and further decrease the false alarm rate resulting from suspected smoke areas. Moreover, we design a metalearning module to alleviate the overfitting issue caused by limited smoke images, and the meta-learning network enables achieving effective detection via comparing the distance between the class prototype of support images and the features of query images. A series of experiments on forest fire smoke datasets and miniImageNet dataset testify that the proposed method is superior to state-of-the-art few-shot learning approaches.
