Few-shot Model Research Articles

AFGN: Adaptive Filtering Graph Neural Network for Few-Shot Learning

The combination of few-shot learning and graph neural networks can effectively solve the issue of extracting more useful information from limited data. However, most graph-based few-shot models only consider the global feature information extracted by the backbone during the construction process, while ignoring the dependency information hidden within the features. Additionally, the essence of graph convolution is the filtering of graph signals, and the majority of graph-based few-shot models construct fixed, single-property filters to process these graph signals. Therefore, in this paper, we propose an Adaptive Filtering Graph Convolutional Neural Network (AFGN) for few-shot classification. AFGN explores the hidden dependency information within the features, providing a new approach for constructing graph tasks in few-shot scenarios. Furthermore, we design an adaptive filter for the graph convolution of AFGN, which can adaptively adjust its strategy for acquiring high and low-frequency information from graph signals based on different few-shot episodic tasks. We conducted experiments on three standard few-shot benchmarks, including image recognition and fine-grained categorization. The experimental results demonstrate that our AFGN performs better compared to other state-of-the-art models.

DMMGNet: A discrimination mapping and memory bank mean guidance-based network for high-performance few-shot industrial anomaly detection

For deep learning-based industrial anomaly detection, it is still challenging to get adequate images for model training and achieve cold start for cross-product migration, restricting their practical application in real industrial production. Herein, an innovative few-shot anomaly detection network DMMGNet based on discrimination mapping and memory bank mean guidance strategies are demonstrated, which is trained by a new two-branch data augmentation technique. By separating the features stored in memory bank from the features used for training, the two-branch data augmentation method can significantly improve the robustness of few-shot model training and reduce the redundance of memory bank. In the elaborately designed discrimination mapping module, new negative samples are generated by adding dynamic Gaussian noise to normal samples along the channel dimension in feature space to solve the problem of sample imbalance. Meanwhile, the discrimination mapping module also helps to map the feature distribution of positive samples to the target domain more efficiently and reduce the deviation of feature domain, conducive to a more precise separation of positive and negative samples. In addition, a novel mean guidance approach with an optimized loss function is developed to guide the positive sample feature mapping by specifying the local feature space center to form a clear feature domain contour and enhance the detection accuracy. The multiple experimental results validate that our DMMGNet outperforms the most advanced anomaly detection counterparts on image-level AUROC, showing an increase by 0.3–3 % on both MVTec AD and MPDD benchmarks under several few-shot scenarios.

Automatic Metric Search for Few-Shot Learning.

Few-shot learning (FSL) aims to learn a model that can identify unseen classes using only a few training samples from each class. Most of the existing FSL methods adopt a manually predefined metric function to measure the relationship between a sample and a class, which usually require tremendous efforts and domain knowledge. In contrast, we propose a novel model called automatic metric search (Auto-MS), in which an Auto-MS space is designed for automatically searching task-specific metric functions. This allows us to further develop a new searching strategy to facilitate automated FSL. More specifically, by incorporating the episode-training mechanism into the bilevel search strategy, the proposed search strategy can effectively optimize the network weights and structural parameters of the few-shot model. Extensive experiments on the miniImageNet and tieredImageNet datasets demonstrate that the proposed Auto-MS achieves superior performance in FSL problems.

Task-Similarity is a Crucial Factor for Few-Shot Meta-Learning of Structure-Activity Relationships.

Machine learning models support computer-aided molecular design and compound optimization. However, the initial phases of drug discovery often face a scarcity of training data for these models. Meta-learning has emerged as a potentially promising strategy, harnessing the wealth of structure-activity data available for known targets to facilitate efficient few-shot model training for the specific target of interest. In this study, we assessed the effectiveness of two different meta-learning methods, namely model-agnostic meta-learning (MAML) and adaptive deep kernel fitting (ADKF), specifically in the regression setting. We investigated how factors such as dataset size and the similarity of training tasks impact predictability. The results indicate that ADKF significantly outperformed both MAML and a single-task baseline model on the inhibition data. However, the performance of ADKF varied across different test tasks. Our findings suggest that considerable enhancements in performance can be anticipated primarily when the task of interest is similar to the tasks incorporated in the meta-learning process.

Unlocking the capabilities of explainable few-shot learning in remote sensing

Recent advancements have significantly improved the efficiency and effectiveness of deep learning methods for image-based remote sensing tasks. However, the requirement for large amounts of labeled data can limit the applicability of deep neural networks to existing remote sensing datasets. To overcome this challenge, few-shot learning has emerged as a valuable approach for enabling learning with limited data. While previous research has evaluated the effectiveness of few-shot learning methods on satellite-based datasets, little attention has been paid to exploring the applications of these methods to datasets obtained from Unmanned Aerial Vehicles (UAVs), which are increasingly used in remote sensing studies. In this review, we provide an up-to-date overview of both existing and newly proposed few-shot classification techniques, along with appropriate datasets that are used for both satellite-based and UAV-based data. We demonstrate few-shot learning can effectively handle the diverse perspectives in remote sensing data. As an example application, we evaluate state-of-the-art approaches on a UAV disaster scene dataset, yielding promising results. Furthermore, we highlight the significance of incorporating explainable AI (XAI) techniques into few-shot models. In remote sensing, where decisions based on model predictions can have significant consequences, such as in natural disaster response or environmental monitoring, the transparency provided by XAI is crucial. Techniques like attention maps and prototype analysis can help clarify the decision-making processes of these complex models, enhancing their reliability. We identify key challenges including developing flexible few-shot methods to handle diverse remote sensing data effectively. This review aims to equip researchers with an improved understanding of few-shot learning’s capabilities and limitations in remote sensing, while pointing out open issues to guide progress in efficient, reliable and interpretable data-efficient techniques.

Towards interactive AI-authoring with prototypical few-shot classifiers in histopathology

A vast multitude of tasks in histopathology could potentially benefit from the support of artificial intelligence (AI). Many examples have been shown in the literature and first commercial products with FDA or CE-IVDR clearance are available. However, two key challenges remain: (1) a scarcity of thoroughly annotated images, respectively the laboriousness of this task, and (2) the creation of robust models that can cope with the data heterogeneity in the field (domain generalization). In this work, we investigate how the combination of prototypical few-shot classification models and data augmentation can address both of these challenges. Based on annotated data sets that include multiple centers, multiple scanners, and two tumor entities, we examine the robustness and the adaptability of few-shot classifiers in multiple scenarios. We demonstrate that data from one scanner and one site are sufficient to train robust few-shot classification models by applying domain-specific data augmentation. The models achieved classification performance of around 90% on a multiscanner and multicenter database, which is on par with the accuracy achieved on the primary single-center single-scanner data. Various convolutional neural network (CNN) architectures can be used for feature extraction in the few-shot model. A comparison of nine state-of-the-art architectures yielded that EfficientNet B0 provides the best trade-off between accuracy and inference time. The classification of prototypical few-shot models directly relies on class prototypes derived from example images of each class. Therefore, we investigated the influence of prototypes originating from images from different scanners and evaluated their performance also on the multiscanner database. Again, our few-shot model showed a stable performance with an average absolute deviation in accuracy compared to the primary prototypes of 1.8% points. Finally, we examined the adaptability to a new tumor entity: classification of tissue sections containing urothelial carcinoma into normal, tumor, and necrotic regions. Only three annotations per subclass (e.g., muscle and adipose tissue are subclasses of normal tissue) were provided to adapt the few-shot model, which obtained an overall accuracy of 93.6%. These results demonstrate that prototypical few-shot classification is an ideal technology for realizing an interactive AI authoring system as it only requires few annotations and can be adapted to new tasks without involving retraining of the underlying feature extraction CNN, which would in turn require a selection of hyper-parameters based on data science expert knowledge. Similarly, it can be regarded as a guided annotation system. To this end, we realized a workflow and user interface that targets non-technical users.

GlyReShot: A glyph-aware model with label refinement for few-shot Chinese agricultural named entity recognition

Chinese agricultural named entity recognition (NER) has been studied with supervised learning for many years. However, considering the scarcity of public datasets in the agricultural domain, exploring this task in the few-shot scenario is more practical for real-world demands. In this paper, we propose a novel model named GlyReShot, integrating the knowledge of Chinese character glyph into few-shot NER models. Although the utilization of glyph has been proven successful in supervised models, two challenges still persist in the few-shot setting, i.e., how to obtain glyph representations and when to integrate them into the few-shot model. GlyReShot handles the two challenges by introducing a lightweight glyph representation obtaining module and a training-free label refinement strategy. Specifically, the glyph representations are generated based on the descriptive sentences by filling the predefined template. As most steps come before training, this module aligns well with the few-shot setting. Furthermore, by computing the confidence values for draft predictions, the refinement strategy selectively utilizes the glyph information only when the confidence values are relatively low, thus mitigating the influence of noise. Finally, we annotate a new agricultural NER dataset and the experimental results demonstrate effectiveness of GlyReShot for few-shot Chinese agricultural NER.

Few-shot learning based on hierarchical feature fusion via relation networks

Few-shot learning, which aims to identify new classes with few samples, is an increasingly popular and crucial research topic in the machine learning. Recently, the development of deep learning has deepened the network structure of a few-shot model, thereby obtaining deeper features from the samples. This trend led to an increasing number of few-shot learning models pursuing more complex structures and deeper features. However, discarding shallow features and blindly pursuing the depth of sample feature levels is not reasonable. The features at different levels of the sample have different information and characteristics. In this paper, we propose a few-shot image classification model based on deep and shallow feature fusion and a coarse-grained relationship score network (HFFCR). First, we utilize networks with different depth structures as feature extractors and then fuse the two kinds of sample features. The fused sample features collect sample information at different levels. Second, we condense the fused features into a coarse-grained prototype point. Prototype points can better represent the information in this class and improve classification efficiency. Finally, we construct a relationship score network, concatenating the prototype points and query samples into a feature map and sending it into the network to calculate the relationship score. The classification criteria for learnable relationship scores reflect the information difference between the two samples. Experiments on three datasets show that HFFCR has advanced performance.

MeshID: Few-Shot Finger Gesture Based User Identification Using Orthogonal Signal Interference.

Radio frequency (RF) technology has been applied to enable advanced behavioral sensing in human-computer interaction. Due to its device-free sensing capability and wide availability on Internet of Things devices. Enabling finger gesture-based identification with high accuracy can be challenging due to low RF signal resolution and user heterogeneity. In this paper, we propose MeshID, a novel RF-based user identification scheme that enables identification through finger gestures with high accuracy. MeshID significantly improves the sensing sensitivity on RF signal interference, and hence is able to extract subtle individual biometrics through velocity distribution profiling (VDP) features from less-distinct finger motions such as drawing digits in the air. We design an efficient few-shot model retraining framework based on first component reverse module, achieving high model robustness and performance in a complex environment. We conduct comprehensive real-world experiments and the results show that MeshID achieves a user identification accuracy of 95.17% on average in three indoor environments. The results indicate that MeshID outperforms the state-of-the-art in identification performance with less cost.

Subject-independent meta-learning framework towards optimal training of EEG-based classifiers

Advances in deep learning have shown great promise towards the application of performing high-accuracy Electroencephalography (EEG) signal classification in a variety of tasks. However, many EEG-based datasets are often plagued by the issue of high inter-subject signal variability. Robust deep learning models are notoriously difficult to train under such scenarios, often leading to subpar or widely varying performance across subjects under the leave-one-subject-out paradigm. Recently, the model agnostic meta-learning framework was introduced as a way to increase the model’s ability to generalize towards new tasks. While the original framework focused on task-based meta-learning, this research aims to show that the meta-learning methodology can be modified towards subject-based signal classification while maintaining the same task objectives and achieve state-of-the-art performance. Namely, we propose the novel implementation of a few/zero-shot subject-independent meta-learning framework towards multi-class inner speech and binary class motor imagery classification. Compared to current subject-adaptive methods which utilize large number of labels from the target, the proposed framework shows its effectiveness in training zero-calibration and few-shot models for subject-independent EEG classification. The proposed few/zero-shot subject-independent meta-learning mechanism performs well on both small and large datasets and achieves robust, generalized performance across subjects. The results obtained shows a significant improvement over the current state-of-the-art, with the binary class motor imagery achieving 88.70% and the accuracy of multi-class inner speech achieving an average of 31.15%. Codes will be made available to public upon publication.

Multi-level Few-Shot Model with Selective Aggregation Feature for Bearing Fault Diagnosis under Limited Data Condition

Multi-level Few-Shot Model with Selective Aggregation Feature for Bearing Fault Diagnosis under Limited Data Condition

ECEA: Extensible Co-Existing Attention for Few-Shot Object Detection.

Few-shot object detection (FSOD) identifies objects from extremely few annotated samples. Most existing FSOD methods, recently, apply the two-stage learning paradigm, which transfers the knowledge learned from abundant base classes to assist the few-shot detectors by learning the global features. However, such existing FSOD approaches seldom consider the localization of objects from local to global. Limited by the scarce training data in FSOD, the training samples of novel classes typically capture part of objects, resulting in such FSOD methods being unable to detect the completely unseen object during testing. To tackle this problem, we propose an Extensible Co-Existing Attention (ECEA) module to enable the model to infer the global object according to the local parts. Specifically, we first devise an extensible attention mechanism that starts with a local region and extends attention to co-existing regions that are similar and adjacent to the given local region. We then implement the extensible attention mechanism in different feature scales to progressively discover the full object in various receptive fields. In the training process, the model learns the extensible ability on the base stage with abundant samples and transfers it to the novel stage of continuous extensible learning, which can assist the few-shot model to quickly adapt in extending local regions to co-existing regions. Extensive experiments on the PASCAL VOC and COCO datasets show that our ECEA module can assist the few-shot detector to completely predict the object despite some regions failing to appear in the training samples and achieve the new state-of-the-art compared with existing FSOD methods. Code is released at https://github.com/zhimengXin/ECEA.

Early fault diagnosis strategy for high-speed train suspension systems based on model-agnostic meta-learning

Early fault diagnosis of suspension systems is essential for the safe operation of high-speed trains. However, neural network-based fault diagnosis methods have two remaining issues: the fault samples are difficult to obtain in practice and the early fault features are too weak to be extracted directly from the raw vibration signals using neural networks. A novel strategy is proposed for early faults diagnosis in suspension systems (i.e. component performance degradation within 20%) by integrating a new sample reconstruction method, a new grouping normalisation method, and model-agnostic meta-learning (MAML) algorithm. First, the 1D raw vibration signals are converted to 2D feature matrices consisting of artificial features using the sample reconstruction method; meanwhile, the grouping normalisation method is used to enhance the early weak fault features in the feature matrices. Second, MAML specialises in few-shot model training for early fault diagnosis, with the feature matrices as the training samples. Finally, the results are compared with those obtained using other current methods. The numerical results show that the proposed strategy yielded excellent performance in the few-shot early faults diagnosis of suspension systems, achieving a maximum accuracy of 94.75%.

EEG-fest: few-shot based attention network for driver's drowsiness estimation with EEG signals

The leading factor behind most vehicular accidents is the driver’s inattentiveness. To accurately determine driver’s drowsiness, Electroencephalography (EEG) has been proven to be a reliable and effective method. Even though previous studies have developed accurate driver’s drowsiness detection algorithms, certain challenges still persist, such as (a) limited training sample sizes, (b) detecting anomalous signals, and (c) achieving subject-independent classification. In this paper we propose a novel solution, names as EEG-Fest, which is a generalized few-shot model aimed at addressing the aforementioned limitations. The EEG-Fest has the ability to (a) classify a query sample’s level of drowsiness with only a few support sample inputs (b) identify whether a query sample is anomalous signals or not, and (c) perform subject-independent classification. During the evaluation, our proposed EEG-Fest algorithm demonstrates better performance compared to other two conventional EEG algorithms in cross-subject validation.

Can Triplet Loss Be Used for Multi-Label Few-Shot Classification? A Case Study

Few-shot learning is a deep learning subfield that is the focus of research nowadays. This paper addresses the research question of whether a triplet-trained Siamese network, initially designed for multi-class classification, can effectively handle multi-label classification. We conducted a case study to identify any limitations in its application. The experiments were conducted on a dataset containing Hungarian legal decisions of administrative agencies in tax matters belonging to a major legal content provider. We also tested how different Siamese embeddings compare on classifying a previously non-existing label on a binary and a multi-label setting. We found that triplet-trained Siamese networks can be applied to perform classification but with a sampling restriction during training. We also found that the overlap between labels affects the results negatively. The few-shot model, seeing only ten examples for each label, provided competitive results compared to models trained on tens of thousands of court decisions using tf-idf vectorization and logistic regression.

ENInst: Enhancing weakly-supervised low-shot instance segmentation

We address a weakly-supervised low-shot instance segmentation, an annotation-efficient training method to deal with novel classes effectively. Since it is an under-explored problem, we first investigate the difficulty of the problem and identify the performance bottleneck by conducting systematic analyses of model components and individual sub-tasks with a simple baseline model. Based on the analyses, we propose ENInst with sub-task enhancement methods: instance-wise mask refinement for enhancing pixel localization quality and novel classifier composition for improving classification accuracy. Our proposed method lifts the overall performance by enhancing the performance of each sub-task. We demonstrate that our ENInst is 7.5 times more efficient in achieving comparable performance to the existing fully-supervised few-shot models and even outperforms them at times.

A Multi-Layer Feature Fusion Method for Few-Shot Image Classification.

In image classification, few-shot learning deals with recognizing visual categories from a few tagged examples. The degree of expressiveness of the encoded features in this scenario is a crucial question that needs to be addressed in the models being trained. Recent approaches have achieved encouraging results in improving few-shot models in deep learning, but designing a competitive and simple architecture is challenging, especially considering its requirement in many practical applications. This work proposes an improved few-shot model based on a multi-layer feature fusion (FMLF) method. The presented approach includes extended feature extraction and fusion mechanisms in the Convolutional Neural Network (CNN) backbone, as well as an effective metric to compute the divergences in the end. In order to evaluate the proposed method, a challenging visual classification problem, maize crop insect classification with specific pests and beneficial categories, is addressed, serving both as a test of our model and as a means to propose a novel dataset. Experiments were carried out to compare the results with ResNet50, VGG16, and MobileNetv2, used as feature extraction backbones, and the FMLF method demonstrated higher accuracy with fewer parameters. The proposed FMLF method improved accuracy scores by up to 3.62% in one-shot and 2.82% in five-shot classification tasks compared to a traditional backbone, which uses only global image features.

Dual adversarial network with meta-learning for domain-generalized few-shot text classification

Meta-learning-based methods prevail in few-shot text classification. Current methods perform meta-training and meta-testing on two parts of a dataset in the same or similar domains. This results in a significant limit in model performance when faced with data from different domains, limiting the generalization of few-shot models. To solve this problem, this study proposes a new setting, namely, domain-generalized few-shot text classification. First, meta-training is conducted on a multi-domain dataset to learn a generalizable model. Subsequently, the model is meta-tested on a target dataset. In addition, a domain-generalized model, namely, a dual adversarial network, is designed to improve the meta-learning-based methods under domain drift between different datasets and domains. Unlike previous meta-learning methods, two N-way-K-shot tasks were input from different domains for a dual adversarial network at each episode. Dual adversarial networks leverage the features from two different domains for adversarial training to improve the domain adaptability of the model. The proposed model utilizes a domain-knowledge generator during adversarial training to produce domain-specific knowledge, and a domain discriminator to recognize the domain label of the produced knowledge. Extensive experiments are conducted to verify the effectiveness of the proposed settings and model. The experimental results show that the model performance in our proposed setting is improved by an average of 3.84% compared to that in cross-domain few-shot text classification. Furthermore, the dual adversarial network significantly outperforms the five competitive baseline models, with an average improvement of 7.20%. The proposed model achieves an average performance improvement of 2.69% compared with the best baseline method.

Few-shot Named Entity Recognition with Joint Token and Sentence Awareness

ABSTRACT Few-shot learning has been proposed and rapidly emerging as a viable means for completing various tasks. Recently, few-shot models have been used for Named Entity Recognition (NER). Prototypical network shows high efficiency on few-shot NER. However, existing prototypical methods only consider the similarity of tokens in query sets and support sets and ignore the semantic similarity among the sentences which contain these entities. We present a novel model, Few-shot Named Entity Recognition with Joint Token and Sentence Awareness (JTSA), to address the issue. The sentence awareness is introduced to probe the semantic similarity among the sentences. The Token awareness is used to explore the similarity of the tokens. To further improve the robustness and results of the model, we adopt the joint learning scheme on the few-shot NER. Experimental results demonstrate that our model outperforms state-of-the-art models on two standard Few-shot NER datasets.

Optimized matrix ordering of sparse linear solver using a few-shot model for circuit simulation

The sparse linear solver has become the bottleneck in a SPICE-like circuit simulator. A general sparse linear solver comprises pre-analysis, numeric factorization, and right-hand solving. The matrix ordering method in pre-analysis determines fill-ins and matrix structure, which is critical for the performance of numeric factorization and right-hand solving. Current leading solvers use a sequential (e.g., KLU) or completely parallel (e.g., NICSLU) framework to order the matrix, which produces non-optimal options or consumes tremendous memory. Furthermore, the criterion for judging the optimal ordering method is also controversial. In this paper, we propose a new pre-analysis framework. It redefines the criterion of the optimal ordering method and utilizes a few-shot model for inference. By performing up to 2 ordering methods from the proposed pre-analysis, our work can implement an optimal ordering method and consume much less memory than a completely parallel framework. Experimental results on 70 circuit matrices show that compared to KLU, our work achieves an average time improvement of 27.09%. Compared to NICSLU, our work achieves an average time improvement of 14.39% and an average memory improvement of 61.52% in the sequential mode, and an average time improvement of 15.74% and an average memory improvement of 55.11% in the parallel (16 threads) mode.