Leak Event Diagnosis for Power Plants: Generative Anomaly Detection Using Prototypical Networks.

Anomaly detection systems based on artificial intelligence (AI) have demonstrated high performance and efficiency in a wide range of applications such as power plants and smart factories. However, due to the inherent reliance of AI systems on the quality of training data, they still demonstrate poor performance in certain environments. Especially in hazardous facilities with constrained data collection, deploying these systems remains a challenge. In this paper, we propose Generative Anomaly Detection using Prototypical Networks (GAD-PN) designed to detect anomalies using only a limited number of normal samples. GAD-PN is a structure that integrates CycleGAN with Prototypical Networks (PNs), learning from metadata similar to the target environment. This approach enables the collection of data that are difficult to gather in real-world environments by using simulation or demonstration models, thus providing opportunities to learn a variety of environmental parameters under ideal and normal conditions. During the inference phase, PNs can classify normal and leak samples using only a small number of normal data from the target environment by prototypes that represent normal and abnormal features. We also complement the challenge of collecting anomaly data by generating anomaly data from normal data using CycleGAN trained on anomaly features. It can also be adapted to various environments that have similar anomalous scenarios, regardless of differences in environmental parameters. To validate the proposed structure, data were collected specifically targeting pipe leakage scenarios, which are significant problems in environments such as power plants. In addition, acoustic ultrasound signals were collected from the pipe nozzles in three different environments. As a result, the proposed model achieved a leak detection accuracy of over 90% in all environments, even with only a small number of normal data. This performance shows an average improvement of approximately 30% compared with traditional unsupervised learning models trained with a limited dataset.

Post-stroke hand gesture recognition via one-shot transfer learning using prototypical networks.

BACKGROUND: In-home rehabilitation systems are a promising, potential alternative to conventional therapy for stroke survivors. Unfortunately, physiological differences between participants and sensor displacement in wearable sensors pose a significant challenge to classifier performance, particularly for people with stroke who may encounter difficulties repeatedly performing trials. This makes it challenging to create reliable in-home rehabilitation systems that can accurately classify gestures. METHODS: Twenty individuals who suffered a stroke performed seven different gestures (mass flexion, mass extension, wrist volar flexion, wrist dorsiflexion, forearm pronation, forearm supination, and rest) related to activities of daily living. They performed these gestures while wearing EMG sensors on the forearm, as well as FMG sensors and an IMU on the wrist. We developed a model based on prototypical networks for one-shot transfer learning, K-Best feature selection, and increased window size to improve model accuracy. Our model was evaluated against conventional transfer learning with neural networks, as well as subject-dependent and subject-independent classifiers: neural networks, LGBM, LDA, and SVM. RESULTS: Our proposed model achieved 82.2% hand-gesture classification accuracy, which was better (P<0.05) than one-shot transfer learning with neural networks (63.17%), neural networks (59.72%), LGBM (65.09%), LDA (63.35%), and SVM (54.5%). In addition, our model performed similarly to subject-dependent classifiers, slightly lower than SVM (83.84%) but higher than neural networks (81.62%), LGBM (80.79%), and LDA (74.89%). Using K-Best features improved the accuracy in 3 of the 6 classifiers used for evaluation, while not affecting the accuracy in the other classifiers. Increasing the window size improved the accuracy of all the classifiers by an average of 4.28%. CONCLUSION: Our proposed model showed significant improvements in hand-gesture recognition accuracy in individuals who have had a stroke as compared with conventional transfer learning, neural networks and traditional machine learning approaches. In addition, K-Best feature selection and increased window size can further improve the accuracy. This approach could help to alleviate the impact of physiological differences and create a subject-independent model for stroke survivors that improves the classification accuracy of wearable sensors. TRIAL REGISTRATION NUMBER: The study was registered in Chinese Clinical Trial Registry with registration number CHiCTR1800017568 in 2018/08/04.

DiffFSRE: Diffusion-Enhanced Prototypical Network for Few-Shot Relation Extraction.

Supervised learning methods excel in traditional relation extraction tasks. However, the quality and scale of the training data heavily influence their performance. Few-shot relation extraction is gradually becoming a research hotspot whose objective is to learn and extract semantic relationships between entities with only a limited number of annotated samples. In recent years, numerous studies have employed prototypical networks for few-shot relation extraction. However, these methods often suffer from overfitting of the relation classes, making it challenging to generalize effectively to new relationships. Therefore, this paper seeks to utilize a diffusion model for data augmentation to address the overfitting issue of prototypical networks. We propose a diffusion model-enhanced prototypical network framework. Specifically, we design and train a controllable conditional relation generation diffusion model on the relation extraction dataset, which can generate the corresponding instance representation according to the relation description. Building upon the trained diffusion model, we further present a pseudo-sample-enhanced prototypical network, which is able to provide more accurate representations for prototype classes, thereby alleviating overfitting and better generalizing to unseen relation classes. Additionally, we introduce a pseudo-sample-aware attention mechanism to enhance the model's adaptability to pseudo-sample data through a cross-entropy loss, further improving the model's performance. A series of experiments are conducted to prove our method's effectiveness. The results indicate that our proposed approach significantly outperforms existing methods, particularly in low-resource one-shot environments. Further ablation analyses underscore the necessity of each module in the model. As far as we know, this is the first research to employ a diffusion model for enhancing the prototypical network through data augmentation in few-shot relation extraction.

A Metric-Based Few-Shot Learning Method for Fish Species Identification with Limited Samples.

Fish species identification plays a vital role in marine fisheries resource exploration, yet datasets related to marine fish resources are scarce. In open-water environments, various fish species often exhibit similar appearances and sizes. To solve these issues, we propose a few-shot learning approach to identifying fish species. Our approach involves two key components. Firstly, the embedding module was designed to address the challenges posed by a large number of fish species with similar phenotypes by utilizing the distribution relationships of species in the embedding space. Secondly, a metric function was introduced, effectively enhancing the performance of fish species classification and successfully addressing the issue of limited sample quantity. The proposed model is trained end to end on fish species public datasets including the Croatian fish dataset, Fish4Knowledge and WildFish. Compared with the prototypical networks, our method performs more effectively and improves accuracy by 2% to 10%; it is able to identify fish effectively in small samples sizes and complex scene scenarios. This method provides a valuable technological tool for the development of fisheries resources and the preservation of fish biodiversity.

Few-Shot Learning for WiFi Fingerprinting Indoor Positioning.

In recent years, deep-learning-based WiFi fingerprinting has been intensively studied as a promising technology for providing accurate indoor location services. However, it still demands a time-consuming and labor-intensive site survey and suffers from the fluctuation of wireless signals. To address these issues, we propose a prototypical network-based positioning system, which explores the power of few-shot learning to establish a robust RSSI-position matching model with limited labels. Our system uses a temporal convolutional network as the encoder to learn an embedding of the individual sample, as well as its quality. Each prototype is a weighted combination of the embedded support samples belonging to its position. Online positioning is performed for an embedded query sample by simply finding the nearest position prototype. To mitigate the space ambiguity caused by signal fluctuation, the Kalman Filter estimates the most likely current RSSI based on the historical measurements and current measurement in the online stage. The extensive experiments demonstrate that the proposed system performs better than the existing deep-learning-based models with fewer labeled samples.

Lifelong Adaptive Machine Learning for Sensor-Based Human Activity Recognition Using Prototypical Networks.

Continual learning (CL), also known as lifelong learning, is an emerging research topic that has been attracting increasing interest in the field of machine learning. With human activity recognition (HAR) playing a key role in enabling numerous real-world applications, an essential step towards the long-term deployment of such systems is to extend the activity model to dynamically adapt to changes in people's everyday behavior. Current research in CL applied to the HAR domain is still under-explored with researchers exploring existing methods developed for computer vision in HAR. Moreover, analysis has so far focused on task-incremental or class-incremental learning paradigms where task boundaries are known. This impedes the applicability of such methods for real-world systems. To push this field forward, we build on recent advances in the area of continual learning and design a lifelong adaptive learning framework using Prototypical Networks, LAPNet-HAR, that processes sensor-based data streams in a task-free data-incremental fashion and mitigates catastrophic forgetting using experience replay and continual prototype adaptation. Online learning is further facilitated using contrastive loss to enforce inter-class separation. LAPNet-HAR is evaluated on five publicly available activity datasets in terms of its ability to acquire new information while preserving previous knowledge. Our extensive empirical results demonstrate the effectiveness of LAPNet-HAR in task-free CL and uncover useful insights for future challenges.

Few-Shot Emergency Siren Detection.

It is a well-established practice to build a robust system for sound event detection by training supervised deep learning models on large datasets, but audio data collection and labeling are often challenging and require large amounts of effort. This paper proposes a workflow based on few-shot metric learning for emergency siren detection performed in steps: prototypical networks are trained on publicly available sources or synthetic data in multiple combinations, and at inference time, the best knowledge learned in associating a sound with its class representation is transferred to identify ambulance sirens, given only a few instances for the prototype computation. Performance is evaluated on siren recordings acquired by sensors inside and outside the cabin of an equipped car, investigating the contribution of filtering techniques for background noise reduction. The results show the effectiveness of the proposed approach, achieving AUPRC scores equal to 0.86 and 0.91 in unfiltered and filtered conditions, respectively, outperforming a convolutional baseline model with and without fine-tuning for domain adaptation. Extensive experiments conducted on several recording sensor placements prove that few-shot learning is a reliable technique even in real-world scenarios and gives valuable insights for developing an in-car emergency vehicle detection system.