Search

As one of the largest and most complex artificial systems in the world, power systems present challenges for statistical analysis under multi-fault contingencies that alter the network topology. This paper proposes a meta-transfer learning-based (MTL) method, where base-learners are designed to learn the power flow mapping relationships for different network topologies, and a meta-learner is developed to guide the updating of structural parameters in base-learners in response to topological changes. The proposed MTL model quickly generates base-learners to adapt to new topologies and enables large-scale statistical analysis of multi-fault contingencies in power systems. The efficacy of the proposed method has been validated using the benchmark Institute of Electrical and Electronics Engineers (IEEE) 39-bus and 300-bus systems, specifically focusing on multi-fault contingencies. The results indicate that the proposed MTL model not only achieves high accuracy in dynamic topologies but also provides adaptive initial parameters for few/zero-shot learning within each topology. [ABSTRACT FROM AUTHOR]

To address the challenges of accurately diagnosing few-shot fault samples obtained from rolling bearings under variable operating conditions, as well as the issues of black box nature and delayed feedback to guide fault handling in intelligent diagnostic models, this paper proposes an interpretable multi-domain meta-transfer learning method. Firstly, vibration monitoring data of rolling bearings under different operating conditions are collected, and time–frequency domain features are extracted to construct multi-channel one-dimensional temporal samples as inputs. A multi-domain meta-transfer learning framework based on deep convolutional neural networks is then built to perform few-shot learning with multiple tasks under different operating conditions. The output results are reverse-reconstructed through a fusion hierarchical class activation mapping, and the feature maps are assigned different weights to obtain saliency maps corresponding to the inputs, thus improving the interpretability of the output results. Finally, the dataset of bearing vibration data under time-varying rotational speed conditions is used to validate the effectiveness of the proposed method. Experimental results show that the proposed method can achieve accurate fault diagnosis results under variable operating conditions with few-shot samples, and the diagnosis results can be fed back to the input for decision-making, enhancing the interpretability of the model. Compared with other models, it also demonstrates better robustness and accuracy. [ABSTRACT FROM AUTHOR]

To solve the problems of insufficient samples and weak fault features of audio signals in the fault diagnosis of plunger pump, this paper proposes a fault diagnosis method of plunger pump based on audio signal combined with meta-transfer learning (MTL-PAFD). The method takes the audio signals of the plunger pump as samples, which are acquired by a single sensor. Through the Gammatone filter bank processing, the representation ability of the audio signal under strong noise interference is effectively improved. Then combined with meta-transfer learning, the few-shot fault diagnosis of plunger pump is realized. In addition, according to the actual needs of fault diagnosis of plunger pump, the test method of meta-transfer learning in fault diagnosis application is improved, which can process unknown fault classes adaptively. Experimental results show that MTL-PAFD has a fault diagnosis accuracy of 91.41% for seen classes. After fast adaptive learning, it can achieve an accuracy of 89.64% when identifying unseen fault classes. [ABSTRACT FROM AUTHOR]

Summary: Person re‐identification (Re‐ID) aims to retrieve a person of interest across multiple nonoverlapping cameras. In recent years, to enable person Re‐ID technology to play out its application value in real‐world scenarios, visual surveillance through unmanned aerial vehicle (UAV) platforms has received intense attention, and aerial person datasets have been constructed. However, the pedestrian images captured by ground cameras and those captured by UAVs exhibit great differences. Person Re‐ID methods based on ground person images have difficulty performing Re‐ID on aerial person images. In this paper, we first use a meta‐transfer method to learn to generalize the aerial person Re‐ID task. Specifically, combining the ideas of meta‐learning and transfer learning, a meta‐learning strategy is introduced to study a feature extractor, and a transfer learning strategy is introduced to utilize and further improve the acquired meta‐knowledge. To prevent the catastrophic forgetting and overfitting problems caused by large‐scale model parameters, we freeze the lower‐layer neurons with powerful generalization and fine‐tune the higher‐layer neurons with strong specialization to transfer and represent the feature extractor. In addition, during the model training process, it is observed that the presence of difficult categories in the given dataset significantly affects the convergence speed and recognition accuracy of the utilized meta‐learning method, and the loss function based on the general Euclidean distance measure tends to mislead the model to optimize in a suboptimal direction. Therefore, we introduce a curriculum sampling based learning strategy that is harmonized with our meta‐transfer learning framework and a new metric formulation of sample similarity based on the Mahalanobis distance to improve the model. In the experimental part, when our method is adopted, a Rank‐1 accuracy of 63.63% and a mean average precision (mAP) of 38.02% are achieved on an aerial Re‐ID dataset, demonstrating its potential for completing person Re‐ID with aerial images. The results obtained on two commonly used ground pedestrian datasets show the generalization of the proposed method. Ablation studies also validate that each component contributes to improving the performance of the model. [ABSTRACT FROM AUTHOR]

Recently, deep learning has been applied to many categories of wireless communications. Channel estimation, which plays a critical role inadequate reception and decoding of the information signal has been one of the application domains of learning techniques. Although the training of neural networks to learn through a massive number of dataset samples gains the knowledge to deal with different situations, it comes with a cost for both computational resources and processing time. This paper proposes a combination of transfer learning and meta-learning techniques to overcome this problem. We propose a super-resolution-based channel estimation where transfer learning is used to obtain a good initialization point for new environments while meta-learning enables learning and fast adaptation of different channel tasks to learn using fewer samples. The proposed technique is tested in a vehicular channel environment and the simulation results confirm the improved performance achieved compared to the state-of-the-art techniques. [ABSTRACT FROM AUTHOR]

In practical industrial applications, it is crucial to train a robust fault diagnosis (FD) model that can quickly adapt to new working conditions or fault modes using a few labeled fault samples. Therefore, a novel convolutional multi-head self-attention network-based meta-transfer learning approach (CMS-MTL) for few-shot fault diagnosis (FSFD) is proposed. Firstly, a convolutional multi-head self-attention network (CMHSAN) is designed, which ingeniously combines the multi-head self-attention (MHSA) blocks and convolution blocks. The local and global feature information of the input time–frequency images are fully considered through the mutual cooperation of MHSA and convolution, so as to fully extract the discriminative features among various fault classes. Secondly, a three-stage CMHSAN-based meta-transfer learning (MTL) scheme is proposed, which provides a good initialization state for the meta-training of the CMHSAN model through the pre-training stage, updates the pre-trained model with the scaling and shifting parameters in the meta-training stage, and fine-tunes the updated model in the meta-testing stage, so as to quickly adapt to new FSFD tasks from the target domain. Thirdly, aiming at the fault classes that are difficult to be diagnosed during meta-training, a meta-task re-training (MTRT) strategy is designed to learn more valuable transferable knowledge in the meta-training stage, thereby improving the adaptability of the CMHSAN model to hard FSFD tasks. Finally, extensive experiments are conducted under different FSFD scenarios to verify the effectiveness of the proposed approach. The results prove that the approach can quickly adapt to new FSFD tasks through the learned meta-knowledge and achieve high diagnosis accuracies. • Design a convolutional multi-head self-attention network to extract fault features. • Develop a three-stage CMHSAN-based MTL scheme to adapt fast to new diagnosis tasks. • Design a meta-task re-training strategy to better diagnose difficult fault classes. • Proposed CMS-MTL approach greatly improves the few-shot fault diagnosis accuracy. [ABSTRACT FROM AUTHOR]

Physics-Informed Neural Networks (PINNs) have gained popularity in solving nonlinear partial differential equations (PDEs) via integrating physical laws into the training of neural networks, making them superior in many scientific and engineering applications. However, conventional PINNs still fall short in accurately approximating the solution of complex systems with strong nonlinearity, especially in long temporal domains. Besides, since PINNs are designed to approximate a specific realization of a given PDE system, they lack the necessary generalizability to efficiently adapt to new system configurations. This entails computationally expensive re-training from scratch for any new change in the system. To address these shortfalls, in this work a sequential meta-transfer (SMT) learning framework is proposed, offering a unified solution for both fast training and efficient adaptation of PINNs in highly nonlinear systems with long temporal domains. Specifically, the framework decomposes PDE's time domain into smaller time segments to create "easier" PDE problems for PINNs training. Then for each time interval, a meta-learner is assigned and trained to achieve an optimal initial state for rapid adaptation to a range of related tasks. Transfer learning principles are then leveraged across time intervals to further reduce the computational cost. Through a composites autoclave processing case study, it is shown that SMT is clearly able to enhance the adaptability of PINNs while significantly reducing computational cost, by a factor of 100. [ABSTRACT FROM AUTHOR]

Multi-horizon time series forecasting is a very challenging task in many fields of research. In the field of machine learning, artificial neural networks have been used to carry out these tasks. However, there are still problems that are of general interest to researchers such as: Loss of data in data acquisition and long-term forecast. In this paper, we propose a hybrid Meta-Transfer Learning technique based on transfer-learning, meta-learning and signal detection by means of the discrete wavelet transform to solve the aforementioned problems in multi-horizon time series forecasting. Input-to-state stability analysis and the strong and weak convergence analysis for the proposed method are included. To validate the effectiveness of the method, the long-term prediction of earthquakes magnitude (M>4.5) in Italy is taken as a case of study, using information from Italy and Mexico. Simulations of classic methods for forecasting time series based on neural models are performed. The forecasting performance obtained is the minimum square error (MSE) is 0.091, while for the meta-transfer learning, the MSE is 0.032.

Radar radio source (RRS) recognition plays an important role in the fields of military electronic support systems (ESM) and civilian autonomous driving. The rapid development of machine learning technology, especially deep learning, has effectively and efficiently improved RRS intelligent recognition performances when operating in the increasingly complex electromagnetic environment. However, the data sampling limitation and computation cost are still severe challenges in real RRS recognition scenarios. In this paper, we propose a novel network based on meta-transfer learning, called RRSARNet, to achieve effective adaptive RRS recognition in the context of low signal-to-noise ratio (SNR). First, by using the short-time Fourier transform, a six-type small samples RRS simulation dataset with different SNR levels is constructed. Then, a novel RRSARNet, based on metric learning, is proposed, which consists of a four-layer embedding module and a four-layer relational module. Finally, the RRS dataset is divided into training, supporting and testing subsets, which are used to train and test the RRSARNet in a meta-transfer learning method. Experiments on the RRS dataset show that the proposed RRSARNet can achieve an overall accuracy (OA) above 96% and 99% when the SNR is above −15 dB and −10 dB, respectively. Even when the SNR is -30 dB, OA can reach more than 70%. For 5-way 1-shot and 5-way 5-shot experiments, the inference time of an image is about 0.043 and 0.140 milliseconds, respectively. Besides, experiments on the RRS simulation dataset and the two benchmark datasets, the RRSARNet performs better or more competitive than many existing state-of-the-art technologies in terms of recognition accuracy. [ABSTRACT FROM AUTHOR]

Bearing skidding is a frequent phenomenon in rotating machinery, which causes equipment motion instability and bearing wear failure. Due to the difficulties in the manufacture of high-speed flexible rotor test rig, measurement of bearing motion parameters and complex structure of angular contact ball bearing (ACBB), which makes the research on double-piece inner ring ACBB skidding challenging. This paper studies the ACBB skidding mechanism from the aspect of rotor characteristics, and first finds the resonance skidding phenomenon. Firstly, a high-speed flexible rotor test rig is designed and built to study the ACBB skidding mechanism, and the influence of foreign matter on the bearing cage slip rate is systematically studied. Then, the bearing cage speed considering the influence of lubricating oil pollution is accurately measured based on the adaptive fractional short-time Fourier transform. Finally, bearing skidding diagnosis under variable working conditions is realized based on the deep meta-transfer learning with feature enhanced generative adversarial network and average deflection power threshold. These proposed strategies systematically solve some problems in the research field of ACBB skidding, which has high practical significance and theoretical guidance value. • The resonance skidding phenomenon is first found by considering flexible rotor characteristics. • Bearing cage speed is accurately measured based on the eddy current sensor. • The influence of foreign matter on ball bearing skidding is first systematically studied. • Bearing skidding diagnosis is realized based on the deep meta-transfer learning. [ABSTRACT FROM AUTHOR]

Recently, large language models have shown great success on various abstractive summarization datasets. These datasets consist of numerous data that are enough to train a large number of parameters. However, for a new domain, there is a lack of labeled data to train those parameters and the model is easily overfitted to a small amount of data. In addition, because annotating document-summary pairs is too expensive and transfer learning using high-resource datasets causes a domain shifting problem, a low-resource abstractive summarization task is becoming necessary. Herein, we propose an efficient framework for low-resource abstractive summarization using a pointer-generator network and a meta-learning technique to address the above problems. Meta-learning using existing high-resource datasets enables our model to rapidly adapt to a new domain using limited data to solve the domain shifting problem. In addition, we explore the copy mechanism using a pointer-generator network that can copy words from a source document when generating a summary. The experimental results on 11 different datasets show that the proposed model outperforms the previous state-of-the-art models in low-resource abstractive summarization on most of the datasets. [ABSTRACT FROM AUTHOR]

Deep learning often requires large amounts of labeled data to train the model, which is not always readily available in the field of speech emotion recognition (SER). Related research work on SER in few shot conditions has reported problem with overfifitting and domain transfer of training. In this study, a few-shot learning method based on meta-transfer learning with domain adaption (MTLDA) is proposed for SER. It not only effectively reduces the over-fitting phenomenon of deep neural networks (DNN) trained with a small number of samples, but also solves the forgetting problem in meta-learning and the target domain adaptability problem in transfer learning. Experiments on three databases (i.e., CASIA is used for pre-training, Emo-DB and SAVEE are used for few-shot learning) are performed for few-shot learning of SER, from which the WAR is 65.12% and UAR is 64.50% on Emo-DB, and the WAR is 58.84% and UAR is 53.26% on SAVEE. • It effectively reduces the over-fitting phenomenon of deep neural networks (DNN) trained with a small number of samples. • It solves the forgetting problem in meta-learning and the target domain adaptability problem in transfer learning. • It makes the convergence fast with fewer tasks. [ABSTRACT FROM AUTHOR]

Screening baggage against potential threats has become one of the prime aviation security concerns all over the world, where manual detection of prohibited items is a time-consuming and hectic process. Many researchers have developed autonomous systems to recognize baggage threats using security X-ray scans. However, all of these frameworks are vulnerable against screening cluttered and concealed contraband items. Furthermore, to the best of our knowledge, no framework possesses the capacity to recognize baggage threats across multiple scanner specifications without an explicit retraining process. To overcome this, we present a novel meta-transfer learning-driven tensor-shot detector that decomposes the candidate scan into dual-energy tensors and employs a meta-one-shot classification backbone to recognize and localize the cluttered baggage threats. In addition, the proposed detection framework can be well-generalized to multiple scanner specifications due to its capacity to generate object proposals from the unified tensor maps rather than diversified raw scans. We have rigorously evaluated the proposed tensor-shot detector on the publicly available SIXray and GDXray datasets (containing a cumulative of 1,067,381 grayscale and colored baggage X-ray scans). On the SIXray dataset, the proposed framework achieved a mean average precision (mAP) of 0.6457, and on the GDXray dataset, it achieved the precision and F1 score of 0.9441 and 0.9598, respectively. Furthermore, it outperforms state-of-the-art frameworks by 8.03% in terms of mAP, 1.49% in terms of precision, and 0.573% in terms of F1 on the SIXray and GDXray dataset, respectively.

Fault diagnosis of bearing under variable working conditions is widely required in practice, and the combination of working conditions and fault fluctuations increases the complexity of addressing its related problems. By developing a virtual simulation model, a digital twin (DT) can obtain the same or even more information than the physical object at a lower cost. Furthermore, it has great potential in the application of bearing fault diagnosis. In this paper, the DT model of the bearing test rig is robustly established, and the fault diagnosis bearing between the simulation and physical object is realized using the proposed enhanced meta-transfer learning (EMTL). First, the DT model is established through parameter identification and modal testing, and the modeling accuracy of DT model reaching 95.685%. The bearing simulation and experimental data are then collected under the same conditions using the DT model and bearing test rig, and the simulation data with little deviation from the experimental data is obtained. Finally, an attention mechanism and domain adaptation are introduced into the EMTL, with the average accuracy of fault diagnosis of bearing reaching 95.18% with few-label target domain data. The proposed strategy is both theoretically significant and practically useful. The experiment results demonstrate that our method outperforms a series of state-of-the-art methods on the bearing fault diagnosis across various limited data conditions. The proposed strategy effectively solves the few-shot problem, which is both theoretically significant and practically useful. • The digital twin model of the bearing test rig is established based on multidisciplinary simulation. • The complete-modal decomposition algorithm is used to identify closely-spaced modal parameter. • The enhanced meta-transfer learning is used for bearing fault diagnosis across different domains. [ABSTRACT FROM AUTHOR]

• A novel prior knowledge-embedded meta -transfer learning (PKEMTL) is proposed for few-shot fault diagnosis. • The representation of distinguishable and condition-invariant features is enhanced by embedding prior knowledge. • Massive experiments under variable operating conditions with limited data have been conducted. In recent years, intelligent fault diagnosis based on deep learning has achieved vigorous development thanks to its powerful feature representation ability. However, scarcity of high-quality data, especially samples under severe fault states, and variable operating conditions have limited the industrial application of intelligent fault diagnosis. To alleviate this predicament, a novel prior knowledge-embedded meta -transfer learning (PKEMTL) is proposed for few-shot fault diagnosis with limited training data and scarce test data. The method focuses on the problem of few-shot fault diagnosis under variable operating conditions to improve adaptability. Different from traditional models, the PKEMTL employs a metric-based meta -learning framework and embeds prior knowledge to enable cross-task learning under variable operating conditions. Specifically, order tracking is firstly introduced as preliminary prior information for data augmentation, and then the augmented data are divided into a series of meta -tasks. Secondly, the meta -tasks are performed by lightweight multiscale feature encoding to obtain high-level feature representations. Next, the meta -learning module based on diagnostic knowledge embedding guides the model to acquire meta -knowledge of speed generalization by constructing the self-supervised task to embed additional prior knowledge into the meta -training process. The generalization performance of the model is further improved by adaptive information fusion learning as a comprehensive decision-making module. Two case studies under variable operating conditions are implemented to validate the effectiveness and superiority of the proposed few-shot fault diagnosis method. [ABSTRACT FROM AUTHOR]

Healthcare has benefited from the implementation of deep-learning models to solve medical image classification tasks. For example, White Blood Cell (WBC) image analysis is used to diagnose different pathologies like leukemia. However, medical datasets are mostly imbalanced, inconsistent, and costly to collect. Hence, it is difficult to select an adequate model to overcome the mentioned drawbacks. Therefore, we propose a novel methodology to automatically select models to solve WBC classification tasks. These tasks contain images collected using different staining methods, microscopes, and cameras. The proposed methodology includes meta- and base-level learnings. At the meta-level, we implemented meta-models based on prior-models to acquire meta-knowledge by solving meta-tasks using the shades of gray color constancy method. To determine the best models to solve new WBC tasks we developed an algorithm that uses the meta-knowledge and the Centered Kernel Alignment metric. Next, a learning rate finder method is employed to adapt the selected models. The adapted models (base-models) are used in an ensemble learning approach achieving accuracy and balanced accuracy scores of 98.29 and 97.69 in the Raabin dataset; 100 in the BCCD dataset; 99.57 and 99.51 in the UACH dataset, respectively. The results in all datasets outperform most of the state-of-the-art models, which demonstrates our methodology's advantage of automatically selecting the best model to solve WBC tasks. The findings also indicate that our methodology can be extended to other medical image classification tasks where is difficult to select an adequate deep-learning model to solve new tasks with imbalanced, limited, and out-of-distribution data., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper, (Copyright © 2023 Elsevier Ltd. All rights reserved.)

Background: Taiwan has insufficient nursing resources due to the high turnover rate of health care providers. Therefore, reducing the heavy workload of these employees is essential. Herein, speech transcription, which has various potential clinical applications, was employed for the documentation of nursing records. The requirement of including only one speaker per transcription facilitated data collection and system development. Moreover, authorization from patients was unnecessary., Objective: The aim of this study was to construct a speech recognition system for nursing records such that health care providers can complete nursing records without typing or with only a few edits., Methods: Nursing records in Taiwan are mainly written in Mandarin, with technical terms and abbreviations presented in both Mandarin and English. Therefore, the training set consisted of English code-switching information. Next, transfer learning (TL) and meta-TL (MTL) methods, which perform favorably in code-switching scenarios, were applied., Results: As of September 2021, the China Medical University Hospital Artificial Intelligence Speech (CMaiSpeech) data set was established by manually annotating approximately 100 hours of recordings from 525 speakers. The word error rate (WER) of the benchmark model of syllable-based TL was 29.54% in code-switching. The WER of the proposed model of syllable-based MTL was 22.20% in code-switching. The test set comprised 17,247 words. Moreover, in a clinical case, the proposed model of syllable-based MTL yielded a WER of 31.06% in code-switching. The clinical test set contained 1159 words., Conclusions: This paper has two main contributions. First, the CMaiSpeech data set-a Mandarin-English corpus-has been established. Health care providers in Taiwan are often compelled to use a mixture of Mandarin and English in nursing records. Second, an automatic speech recognition system for nursing record document conversion was proposed. The proposed system can shorten the work handover time and further reduce the workload of health care providers., (©Shih-Yen Hou, Ya-Lun Wu, Kai-Ching Chen, Ting-An Chang, Yi-Min Hsu, Su-Jung Chuang, Ying Chang, Kai-Cheng Hsu. Originally published in JMIR Nursing (https://nursing.jmir.org), 07.12.2022.)

In the actual production process of the oil field, the functionality of the oil well pumps will be negatively affected by many factors such as manufacturing quality, installation quality, sand, wax, water, gas, heavy oil, and corrosion, which will cause great loss to the production. Therefore, it is very important to analyze the working conditions of the rod pumping systems. In actual oilfield production, the working conditions of deep well pumps are analyzed based on the measured surface indicator diagrams. However, traditional computer diagnosis of pumping wells relies on necessary mathematical methods, or deep networks with many parameters. These methods require a lot of data, with complex analysis processes, long testing time and low efficiency. This article studies the application of meta-transfer learning in the diagnosis of rod pump wells in few-shot scenarios. Meta-transfer learning combines the advantages of both meta-learning and transfer learning. It can not only provide good initial parameters for learners based on deeper networks through the pre-training stage of transfer learning, but also achieve automatic adjustment of hyperparameters with the help of meta-learning. This enables fast gradient iteration and reduces the probability of overfitting, thereby improving model performance. We also conduct comparative experiments to compare the experimental performance of this method with classical meta-learning methods and deep convolutional networks on the classification problem of indicator diagrams. According to the experimental results, the accuracy rate of meta-transfer learning in the diagnosis of few-shot working conditions in practical problems is close to 80%, which is better than the 70% accuracy rate of the comparative experiments. In the actual oil field, there are not many indicator diagrams for pumping unit diagnosis, so this method can well meet the needs of fault detection. • A novel method for diagnosing pumping unit working conditions is proposed, combining meta-learning and transfer learning. • The proposed method provides a new solution to the few-shot problem of the oil field fault indicator diagrams. • A new batch processing method is proposed, which improves the final accuracy by continuously learning failed tasks. • This model fuses various features of the indicator diagrams for experimental analysis and realizes end-to-end learning. [ABSTRACT FROM AUTHOR]

• Attention module can improve the accuracy of meta-transfer learning. • Our attention module incorporated in meta-transfer learning has better performance than others. • The proposed few-shot learning approach is evaluated for iris recognition. • Our attention approach can be used as a plug-and-play module for convolutional neural networks. Iris recognition is a very important biometric technology. Given sufficient labeled data, iris recognition algorithms combined with deep learning have achieved excellent performance. With limited training samples, however, over-fitting often occurs and affects recognition performance if deep learning methods are directly used for training. The learning problem with insufficient samples may be solved by using few-shot learning methods. In this paper, we propose an attention meta-transfer learning (AttentionMTL) approach for iris recognition through an improved attention network model. Experiments on the publicly available datasets show that AttentionMTL has achieved the highest accuracy of 99.95% and obtained higher accuracy (up to 6%) than conventional MTL method and other related approaches. [Display omitted] [ABSTRACT FROM AUTHOR]

In recent years, with the rapid development of machine learning, automatic emotion recognition based on electroencephalogram (EEG) signals has received increasing attention. However, owing to the great variance of EEG signals sampled from different subjects, EEG-based emotion recognition experiences the individual difference problem across subjects, which significantly hinders recognition performance. In this study, we presented a method for EEG-based emotion recognition using a combination of a multi-scale residual network (MSRN) and meta-transfer learning (MTL) strategy. The MSRN was used to represent connectivity features of EEG signals in a multi-scale manner, which utilized different receptive fields of convolution neural networks to capture the interactions of different brain regions. The MTL strategy fully used the merits of meta-learning and transfer learning to significantly reduce the gap in individual differences between various subjects. The proposed method can not only further explore the relationship between connectivity features and emotional states but also alleviate the problem of individual differences across subjects. The average cross-subject accuracies of the proposed method were 71.29% and 71.92% for the valence and arousal tasks on the DEAP dataset, respectively. It achieved an accuracy of 87.05% for the binary classification task on the SEED dataset. The results show that the framework has a positive effect on the cross-subject EEG emotion recognition task., (Copyright © 2022 Elsevier Ltd. All rights reserved.)

• Individual diversity causes poor performance of the current gesture tracking systems when they were directly applied to new users. • An online meta-transfer learning method to learning the individual characters with low data collection cost. • A data augmentation method that leverages the redundant information to generate virtual instances at the premises of the accurate detection result of recursive Otsu segmentation. • A datum-based data alignment strategy that breaks the limitation of available classifiers for recognition without distort the instance. Individual diversity poses a cross-user performance variance challenge that stumbles the practicality, especially for the wireless gesture tracking systems. Since the difficulty of annotating low-semantic wireless data limits constructing a big dataset, the recognizer should quickly adjust to different individuals via small datasets. To this end, we present TransTrack, an accurate wireless indoor gesture tracking system that can adjust to different users quickly. The key insight is that each unlabeled gesture contains learnable individual features that can help the gesture tracking model learning how to adapt to different users. Specifically, TransTrack uses recursive Otsu segmentation to separate gesture-induced signals with the background noise inspired by image segmentation. It then augments training data to learn the transferable features by leveraging the redundant information. A datum-based alignment method is proposed to unlock the limitation of classifier selection without distortion. Finally, TransTrack proposes an online meta-transfer learning method that collects unlabeled data transparently to train the tracking model for different tasks. Extensive experiments show that TransTrack can quickly adapt to different users and conditions. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNNs) have achieved success in optical flow estimation using labeled datasets, but they fail to build an internal representation to fast adapt to the specific task. On the other hand, limited by the lack of ground truth, existing CNNs-based methods suffer from high noise sensitivity and inferior generalization performance. We integrate the meta-learning technique with optical flow estimation, which can learn internal features to search optimal initial state parameters of the network. Meanwhile, we devise an enhanced network termed MetaFlow to further improve performance. MetaFlow extracts per-pixel features, builds correlation volumes for all pairs of pixels, and iteratively updates optical flow through optical flow predictor using meta-learning. In addition, we propose a meta-transfer pretraining approach to obtain initial network weights, which can efficiently avoid network overfitting. Empirical experiments on MPI Sintel and KITTI benchmarks have shown that the proposed MetaFlow achieves the state-of-the-art results and performs outstanding in challenging scenarios such as textureless regions and abrupt motions. [ABSTRACT FROM AUTHOR]

Screening baggage against potential threats has become one of the prime aviation security concerns all over the world, where manual detection of prohibited items is a time-consuming and hectic process. Many researchers have developed autonomous systems to recognize baggage threats using security X-ray scans. However, all of these frameworks are vulnerable against screening cluttered and concealed contraband items. Furthermore, to the best of our knowledge, no framework possesses the capacity to recognize baggage threats across multiple scanner specifications without an explicit retraining process. To overcome this, we present a novel meta-transfer learning-driven tensor-shot detector that decomposes the candidate scan into dual-energy tensors and employs a meta-one-shot classification backbone to recognize and localize the cluttered baggage threats. In addition, the proposed detection framework can be well-generalized to multiple scanner specifications due to its capacity to generate object proposals from the unified tensor maps rather than diversified raw scans. We have rigorously evaluated the proposed tensor-shot detector on the publicly available SIXray and GDXray datasets (containing a cumulative of 1,067,381 grayscale and colored baggage X-ray scans). On the SIXray dataset, the proposed framework achieved a mean average precision (mAP) of 0.6457, and on the GDXray dataset, it achieved the precision and F1 score of 0.9441 and 0.9598, respectively. Furthermore, it outperforms state-of-the-art frameworks by 8.03% in terms of mAP, 1.49% in terms of precision, and 0.573% in terms of F1 on the SIXray and GDXray dataset, respectively. [ABSTRACT FROM AUTHOR]