Search

Learning-to-learn, a progressive speedup of learning while solving a series of similar problems, represents a core process of knowledge acquisition that draws attention in both neuroscience and artificial intelligence. To investigate its underlying brain mechanism, we trained a recurrent neural network model on arbitrary sensorimotor mappings known to depend on the prefrontal cortex. The network displayed an exponential time course of accelerated learning. The neural substrate of a schema emerges within a low-dimensional subspace of population activity; its reuse in new problems facilitates learning by limiting connection weight changes. Our work highlights the weight-driven modifications of the vector field, which determines the population trajectory of a recurrent network and behavior. Such plasticity is especially important for preserving and reusing the learnt schema in spite of undesirable changes of the vector field due to the transition to learning a new problem; the accumulated changes across problems account for the learning-to-learn dynamics.

The cloud-assisted mobile edge computing system is a critical architecture to process computation-intensive and delay-sensitive mobile applications in close proximity to mobile users with high resource efficiency. Due to the heterogenous dynamics of task arrivals at edge nodes and the distributed nature of the system, the workloads of edge nodes are prone to be unbalanced, which can cause high task response time and resource cost. This paper solves the dynamic task scheduling problem in cloud-assisted mobile edge computing (including both peer task scheduling among edge nodes and cross-layer task scheduling from edge nodes to the cloud), aiming at minimizing average task response time within resource budget limit. To overcome the challenges of task arrival dynamics, edge node heterogeneity, and computation-communication delay tradeoff, we propose a Water-filling Based Dynamic Task Scheduling (WiDaS) algorithm. WiDaS dynamically tunes the usage of cloud resources based on the Lyapunov optimization method and efficiently schedules mobile tasks among edge nodes (and the cloud) by exploiting the idea of water filling. Extensive simulations are conducted to evaluate WiDaS under a trace-driven traffic pattern and two mathematic traffic patterns. The results demonstrate that WiDaS shows two-fold benefits of efficiency and effectiveness.

Gesture recognition by using Inertial Measurement Unit (IMU) sensors plays an important role in various Internet of Things (IOT) applications, e.g., smart home, intelligent medical system and so on. Traditional technologies usually utilize machine learning algorithms to train different gestures during the offline phase, then recognize the gesture during the online phase. However, such technologies cannot recognize these gestures without prior training. Even for the same gesture, with different gesture amplitude may result in unsuccessful recognition. Also if we change the person to perform the same gesture, the algorithms fails. In order to overcome these drawbacks, we propose an approach, which will be able to track the human body motion in realtime and also recognize complicated gestures. It utilizes the accelerometer information and proposes comprehensive localization algorithms for each deployed sensor attached on the human body. Then, it takes the correlation and limitation among body parts into account to recognize the gesture. Our experiments results show that, the successful recognition rate of our algorithm is 100%. Furthermore, any part of the human body can be well tracked, the tracking accuracy can reach 0.06m.

Xiao, F, Yang, Y, Xiao, L, Xia, Z, Wang, L, Yang, K, and Wang, S. Reduction of T cells and hsa-miR150-5p in female canoeing athletes: Preliminary evidence between exercise training and immune. J Strength Cond Res 36(11): e106-e113, 2022-This article aims to reveal the alteration of immune profile in teenage canoeing athletes, by which applies a clue for regulation of exercise on human immune. Thirty-one teenagers of female canoeing athletes and age-matched subjects participated in this research. Peripheral leukocytes' microRNAs (miRNAs) were analyzed using Agilent human microRNA 2.0 and gene software. The miRNA candidates were quantified by reverse transcription real-time quantitative polymerase chain reaction (RT-qPCR). The percentages of various lymphocytes were tested using flow cytometry. There were 6 miRNAs (hsa-miR150-5p, 31-5p, 3659, 4419a, 650, and 8485) lower in canoeing athletes, and the reduction of miR-150 was identified by RT-qPCR ( p = 0.021). Canoeing athletes had lower percent of CD3 + T cells than the subjects with no exercise training had ( p0.001), but the ratio of CD4 + to CD8 + and the percent of CD4 + T cells and CD8 + T cells showed no significant difference between these 2 groups. T cells and hsa-miR150-5p are sensitive to the long-time heavy exercise training, and the exercise for winning competition regulates the immune system by inhibiting T cells and hsa-miR150-5p.

Traditional searchable encryption schemes based on the Term Frequency-Inverse Document Frequency (TF-IDF) model adopt the presence of keywords to measure the relevance of documents to queries, which ignores the latent semantic meanings that are concealed in the context. Latent Dirichlet Allocation (LDA) topic model can be utilized for modeling the semantics among texts to achieve semantic-aware multi-keyword search. However, the LDA topic model treats queries and documents from the perspective of topics, and the keywords information is ignored. In this paper, we propose a privacy-preserving searchable encryption scheme based on the LDA topic model and the query likelihood model. We extract the feature keywords from the document using the LDA-based Information Gain (IG) and Topic Frequency-Inverse Topic Frequency (TF-ITF) model. With feature keyword extraction and the query likelihood model, our scheme can achieve a more accurate semantic-aware keyword search. A special index tree is used to enhance search efficiency. The secure inner product operation is utilized to implement the privacy-preserving ranked search. The experiments on real-world datasets demonstrate the effectiveness of our scheme.

Network traffic classification is important to intrusion detection and network management. Most of existing methods are based on machine learning techniques and rely on the features extracted manually from flows or packets. However, with the rapid growth of network applications, it is difficult for these approaches to handle new complex applications. In this paper, we design a novel neural network, the Extended Byte Segment Neural Network (EBSNN), to classify network traffic. EBSNN first divides a packet into header segments and payload segments, which are then fed into encoders composed of the recurrent neural networks with the attention mechanism. Based on the outputs, another encoder learns the high-level representation of the whole packet. In particular, side-channel features are learned from header segments to improve the performance. Finally, the label of the packet is obtained by the softmax function. Furthermore, EBSNN can classify network flows by examining the first few packets. Thorough experiments on the real-world datasets show that EBSNN achieves better performance than the state-of-the-art methods in both the application identification task and the website identification task.

Bone Marrow Mesenchymal Stem Cells (BMSCs), multidirectional cells with self-renewal capacity, can differentiate into many cell types and play essential roles in tissue healing and regenerative medicine. Cell experiments and in vivo research in animal models have shown that BMSCs can repair degenerative discs by promoting cell proliferation and expressing Extracellular Matrix (ECM) components, such as type II collagen and protein-polysaccharides. Delaying or reversing the Intervertebral Disc Degeneration (IDD) process at an etiological level may be an effective strategy. However, despite increasingly in-depth research, some deficiencies in cell transplantation timing and strategy remain, preventing the clinical application of cell transplantation. Exosomes exhibit the characteristics of the mother cells from which they are secreted and can inhibit Nucleus Pulposus Cell (NPC) apoptosis and delay IDD through intercellular communication. Furthermore, the use of exosomes effectively avoids problems associated with cell transplantation, such as immune rejection. This manuscript introduces almost all of the BMSCs and exosomes derived from BMSCs (BMSCs-Exos) described in the IDD literature. Many challenges regarding the use of cell transplantation and therapeutic exosome intervention for IDD remain to be overcome.

Computational experiments are an important method for carrying out the quantitative analysis of complex systems and play a major role in mapping the real world to the virtual world. However, the flexibility of computational experiments leads to arbitrary modeling processes and unconvincing results, which greatly hinder the large-scale application of this method. In this context, the verification of computational models has become an urgent problem in this field. Currently, model evaluation is still in its infancy and the existing evaluation methods are not mature enough. Thus, we took epidemic models as the research object and proposed a capability maturity evaluation framework for computational models of artificial society. The framework differs from previous assessment methods that focus on the validity of results, but instead provides a comprehensive evaluation from two perspectives: 1) evaluation of the model itself--by comparing the expectation with the final implementation, we can obtain whether the model meets the expectation and 2) comparison between different models--by evaluating the implementation process of each model and comparing the results, we can identify more mature models. The implementation of the model is evaluated from input, process, and output. Further, specific analyses and evaluations are conducted for several representative COVID-19 models to verify the validity of this evaluation framework. The results of the case study show that the proposed evaluation framework can help decision-makers identify more mature and referential models, and point out the directions where modelers can improve their models. IEEE

Relation classification (RC) task is one of fundamental tasks of information extraction, aiming to detect the relation information between entity pairs in unstructured natural language text and generate structured data in the form of entity-relation triple. Although distant supervision methods can effectively alleviate the problem of lack of training data in supervised learning, they also introduce noise into the data, and still cannot fundamentally solve the long-tail distribution problem of the training instances. In order to enable the neural network to learn new knowledge through few instances like humans, this work focuses on few-shot relation classification (FSRC), where a classifier should generalize to new classes that have not been seen in the training set, given only a number of samples for each class. To make full use of the existing information and get a better feature representation for each instance, we propose to encode each class prototype in an adaptive way from two aspects. First, based on the prototypical networks, we propose an adaptive mixture mechanism to add label words to the representation of the class prototype, which, to the best of our knowledge, is the first attempt to integrate the label information into features of the support samples of each class so as to get more interactive class prototypes. Second, to more reasonably measure the distances between samples of each category, we introduce a loss function for joint representation learning to encode each support instance in an adaptive manner. Extensive experiments have been conducted on FewRel under different few-shot (FS) settings, and the results show that the proposed adaptive prototypical networks with label words and joint representation learning has not only achieved significant improvements in accuracy, but also increased the generalization ability of few-shot RC models.

OBJECTIVE To clarify whether systemic lymph node dissection (LND) influences the safety of surgery and the survival of patients with locally advanced esophageal squamous cell carcinoma (ESCC) after neoadjuvant chemoradiotherapy (nCRT). SUMMARY BACKGROUND DATA Prognostic impact of systemic lymphadenectomy during surgery after nCRT for ESCC is still uncertain and requires clarification. METHODS This is a secondary analysis of NEOCRTEC5010 trial which compared nCRT followed by surgery versus surgery alone for locally advanced ESCC. Relationship between number of LND and perioperative, recurrence and survival outcomes were analyzed in the nCRT group. RESULTS Three-year overall survival was significantly better in the nCRT group than the S group (75.2% vs 61.5%; P=0.011). In the nCRT group, greater number of LND was associated with significantly better overall survival (HR, 0.358; P < 0.001) and disease-free survival (HR, 0.415; P=0.001), but without any negative impact on postoperative complications. Less LND (< 20 vs ≥ 20) was significantly associated with increased local recurrence (18.8% vs 5.2%, P=0.004) and total recurrence rates (41.2% vs 25.8%, P=0.027). Compared to patients with persistent nodal disease, significantly better survival was seen in patients with complete response and with LND ≥ 20, but not in those with LND < 20. CONCLUSIONS Systemic lymph node dissection does not increase surgical risks after nCRT in ESCC patients. And it is associated with better survival and local disease control. Therefore, systemic lymphadenectomy should still be considered as an integrated part of surgery after nCRT for ESCC.

Video stabilization is the process of improving the video quality by removing annoying fluctuant motion caused by camera jittering. A key issue of a successful solution is the temporal adaptability to motion and the overall robustness with respect to different motion types. However, most previous methods usually produce non-motion adaptive stabilized videos. In other words, under-smoothing in slow motion segments and over-smoothing in rapid motion segments will be produced for complex shaky videos. To overcome these drawbacks, we propose a novel video stabilization approach using a motion morphological component (MMC decomposition. Specifically, the observed motion is decomposed into three MMCs: low-frequency smoothed (LFS motion, high-frequency compensatory (HFC motion, and shaky motion. LFS motion helps to largely stabilize videos, and HFC motion helps to recover missing motion to deal with over-smoothing. Subsequently, we present an MMC-based model to retrieve the desired smoothed motion, in which weighted nuclear norm and autoregression priors are used for LFS motion, while a sparsity prior is adopted for HFC motion. In addition, we design an adaptive weight setting scheme to detect rapid motions and to calculate the optimal weights. Finally, we develop a stabilization algorithm under the Alternating Direction Method of Multipliers (ADMM framework. Experimental results demonstrate that our method can achieve high-quality results compared with that of other state-of-the-art stabilization methods in terms of robustness and efficiency, both quantitatively and qualitatively.

We develop WiDE, a WiFi-distance estimation based group profiling system using LightGBM. Given the uploaded WiFi information by users, WiDE can automatically learn powerful hidden features from the proposed features for between-person distance estimation, and infer group membership with the estimated distance. For each group, WiDE classifies the mobility level, and recognizes the group structure by applying the multi-dimensional scaling technique on the matrix of distance between pairwise persons within the same group. We first validate the performance of between-person distance esti- mation via extensive experiments in a three-floor campus building and a shopping center, showing that WiDE system outperforms other machine learning based approaches for between-person distance estimation, with the average absolute error (AAE) of 0.69m and 1.14m for the campus building and shopping center, respectively, and the corridor identification accuracy for the campus building is over 99%. In addition, the experiments in the shopping center show that our approach can accurately detect groups, classify group mobility into fine-grained level and recognize the group structure.

Firms’ knowledge base is one of the key antecedents of firm innovation. While extant literatures identify diversification and depth of firms’ knowledge base as the important inputs of firm innovation, coherence of firms’ knowledge base reflecting the relatedness of knowledge base is largely neglected. This article goes a further step and points out that coherence of firms’ knowledge base will co-affect with diversification or depth of firms’ knowledge base to foster firm innovation, because it improves the efficiency of knowledge combination and integration. Moreover, degree centralization of industrial knowledge networks, as a characteristic of industrial knowledge base, moderates the abovementioned relationships by offering technological combination opportunities in the industry and helping firms break cognitive inertia. Based on patent data of 76 global firms from three-dimensional printing, information and communication technology, wind energy, and lithium battery industries during the period of 1993–2013, the analysis results show that coherence of firms’ knowledge base positively moderates the relationship between diversification of firms’ knowledge base and firm innovation. Furthermore, degree centralization of industrial knowledge networks, together with coherence of firms’ knowledge base, strengthens the positive relationship between knowledge diversification and firm innovation. Theoretical and practical suggestions about firms’ knowledge management were discussed.

Recent transformer-based models, especially patch-based methods, have shown huge potentiality in vision tasks. However, the split fixed-size patches divide the input features into the same size patches, which ignores the fact that vision elements are often various and thus may destroy the semantic information. Also, the vanilla patch-based transformer cannot guarantee the information communication between patches, which will prevent the extraction of attention information with a global view. To circumvent those problems, we propose the Efficient Attention Pyramid Transformer (EAPT) in this paper. More specifically, we first propose Deformable Attention, which learns an offset for each position in patches. Therefore, even with split fixed-size patches, our method can still obtain non-fixed attention information that can cover various vision elements. Then, we design the Encode-Decode Communication module (En-DeC module), which can obtain communication information between all patches to get more complete global attention information. Finally, we also proposed a position encoding specifically for vision transformers, which can be used for patches of any dimensions and any lengths. Extensive experiments on the vision tasks of image classification, object detection, semantic segmentation demonstrate the effectiveness of our proposed model. Furthermore, we also conduct rigorous ablation studies to evaluate the key components of the proposed structure.

Pulmonary arterial hypertension (PAH) is a chronic pulmonary vascular disease characterized by increased pulmonary arterial pressure and pulmonary arterioles remodeling. Some studies have discovered the relationship between sympathetic nerves (SNs) and pathogenesis of PAH. This review is aimed to illustrate the location and components of SNs in the pulmonary artery, along with different methods and effects of pulmonary artery denervation (PADN). Studies have shown that the SNs distributed mainly around the main pulmonary artery (MPA) and pulmonary artery (PA) bifurcation. And the SNs could be destroyed by three ways: the chemical way, the surgical way and the catheter-based way. PADN can significantly decrease pulmonary arterial pressure rapidly, improve hemodynamic varieties, and then palliate PAH. PADN has been recognized as a prospective and effective therapy for PAH patients, especially for those with medication-refractory PAH. However, further enlarged clinical studies are needed to confirm accurate distribution of SNs in the pulmonary artery and the efficacy of PADN.

Background: Rhabdomyolysis (RM) associated acute kidney injury (AKI) is the most common systemic complication of RM. The present study aimed to assess the clinical characteristics and risk factors for intensive care unit (ICU) transfer for patients with RM-associated AKI.Methods: We included all patients who were age ≥ 18 years old with a diagnosis of RM from September 2012 to October 2018 and divided them into RM-associated AKI group and RM without AKI group. The primary outcome was transferring to ICU treatment. Regression analysis was performed to identify factors associated with ICU treatment and recovery of renal function.Results: Among the 149 patients with RM, 68 (45.6%) developed AKI. The percentage of patients with AKI who transferred to ICU was higher than patients without AKI (33.8% versus 12.3%, P < 0.002). Additionally, patients with AKI had higher percentage of undergoing dialysis (19.1% versus 2.5%, P < 0.01), all-cause mortality (13.2% versus 1.2%, P < 0.01), cost of hospitalization [10.8 1,000 yuan, IQR (5.5, 3.5) versus 5.9 1,000 yuan, IQR 5.9 (3.6, 9.9), P = 0.03], as well as longer length of hospital stay [8.0 (5.0, 14.0)] versus [6.0 (4.0, 11.0)], P = 0.02). Moreover, most patients with AKI achieved complete recovery (77.9%) at discharge. After adjusting for potential risk factors, RM-associated AKI remained an independent risk factor for ICU transfer (OR = 3.0, 95% CI, 1.11–8.3, P = 0.03). However, ICU transfer was not associated with recovery of renal function (OR = 0.88, 95% CI, 0.22–3.57, P = 0.856).Conclusion: RM leaded to AKI in most patients. RM-associated AKI could cause worse clinical outcome and predict ICU transfer for patients with RM.

Lithium plating in working batteries has attracted wide attention in the exploration of safe energy storage. Establishing an effective and rapid early-warning method is strongly considered but quite challenging since lithium plating behavior is determined by diverse factors. In this contribution, we present a non-destructive electrochemical detection method based on transient state analysis and three-electrode cell configuration. Through dividing the iR drop value by the current density, the as-obtained impedance quantity (Ri) can serve as a descriptor to describe the change of electrochemical reaction impedance on the graphite anode. The onset of lithium plating can be identified from the sharp drop of Ri. Once the dendritic plated lithium occurs, the extra electrochemical reactions at the lithium interfaces leads to growing active area and reduced surface resistance of the anode. We proposed a protocol to operate the batteries under the limited capacity, which renders the cell with 98.2% capacity retention after 1000 cycles without lithium plating. The early-warning method has also been validated in in-situ optical microscopy batteries and practical pouch cells, providing a general but effective method for online lithium plating detection towards safe batteries.

BACKGROUND Transcutaneous auricular vagus nerve stimulation (taVNS) may modulate cardiac autonomic function. However, the response rate of the traditional tonic paradigm is low, and the results remain inconsistent. A recent pilot study presented a novel burst paradigm to activate the cardiac parasympathetic system, which might offer a new approach to treat cardiac autonomic function. The present study reassessed the effect of burst taVNS on modulating heart rate variability and explored the difference between burst and traditional tonic paradigms. MATERIALS AND METHODS Forty-two young adults were recruited for this study. Each participant underwent three types of taVNS with sham (30 sec of stimulation), tonic (25 Hz, 500 μsec), and burst (five pulses at 500 Hz every 200 msec) paradigms, respectively, with simultaneous electrocardiogram recording. One-way analysis of variance, multivariate analysis of variance, and linear regression were used for analysis. Multiple testing was performed using Bonferroni correction. RESULTS Both burst and tonic paradigms induced a significant decrease in heart rate, which continued until poststimulation, and increased cardiac parasympathetic activity. Moreover, two parasympathetic system indicators showed significant increase only in burst taVNS. The response rates during burst (35.7%) and tonic (38.1%) stimulations were both higher than that during sham stimulation (11.9%). The response to taVNS showed parameter specificity with few nonresponders to the tonic paradigm responding to the burst paradigm. The overall response rate increased from 38.1% in tonic taVNS to 54.8% in taVNS using both burst and tonic paradigms. For both burst and tonic responders, baseline cardiac parasympathetic activity was found to be significantly negatively correlated with changes during stimulation. CONCLUSION The burst parameter could be used as an alternative strategy for regulating cardiac parasympathetic function by taVNS, which has the potential to be used as a complementary paradigm to traditional tonic taVNS for promoting clinical treatment efficacy.

The clinical success of the percutaneous coronary intervention (PCI) is highly dependent on endovascular manipulation skills and dexterous manipulation strategies of interventionalists. However, the analysis of endovascular manipulations and related discussion for technical skill assessment are limited. In this study, a multilayer and multimodal-fusion architecture is proposed to recognize six typical endovascular manipulations. The synchronously acquired multimodal motion signals from ten subjects are used as the inputs of the architecture independently. Six classification-based and two rule-based fusion algorithms are evaluated for performance comparisons. The recognition metrics under the determined architecture are further used to assess technical skills. The experimental results indicate that the proposed architecture can achieve the overall accuracy of 96.41%, much higher than that of a single-layer recognition architecture (92.85%). In addition, the multimodal fusion brings significant performance improvement in comparison with single-modal schemes. Furthermore, the K -means-based skill assessment can obtain an accuracy of 95% to cluster the attempts made by different skill-level groups. These hopeful results indicate the great possibility of the architecture to facilitate clinical skill assessment and skill learning.

We show that distortion in the size distribution of banks around regulatory thresholds can be used to identify costs of bank regulation. We build a structural model in which banks can strategically bunch their assets below regulatory thresholds to avoid regulations. The resultant distortion in the size distribution of banks reveals the magnitude of regulatory costs. Using U.S. bank data, we estimate the regulatory costs imposed by the Dodd-Frank Act. Although the estimated regulatory costs are substantial, they are significantly lower than banks’ self-reported estimates. Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web site next to the link to the final published paper online.

Recently, multi-task correlation filters has drawn much attention in the object tracking field, which utilizes the multi-task learning (MTL) approach to explore the interdependencies among deep features for object tracking. However, the existing multi-task correlation filters based method fails to consider the relations between the correlation filters. To address this problem, a novel correlation filters based visual tracking method is proposed in this paper, with the integration of multi-task convolution operators and object detection. In our method, convolution and correction filters are jointly learnt through using the MTL technique, with the purpose of exploring not only the interdependencies of deep features but also the internal relevance of the convolution filters. In addition, object detection is introduced into our algorithm to handle the problem of object missing to ensure a better performance of our tracking method. Experiments on five benchmark datasets demonstrate that the proposed visual tracking method outperforms existing state-of-the-art approaches.

Recent learning-based intrinsic image decomposition methods have achieved remarkable progress. However, they usually require massive ground truth intrinsic images for supervised learning, which limits their applicability on real-world images since obtaining ground truth intrinsic decomposition for natural images is very challenging. In this paper, we present an unsupervised framework that is able to learn the decomposition effectively from a single natural image by training solely with the image itself. Our approach is built upon the observations that the reflectance of a natural image typically has high internal self-similarity of patches, and a convolutional generation network tends to boost the self-similarity of an image when trained for image reconstruction. Based on the observations, an unsupervised intrinsic decomposition network (UIDNet) consisting of two fully convolutional encoder-decoder sub-networks, i.e., reflectance prediction network (RPN) and shading prediction network (SPN), is devised to decompose an image into reflectance and shading by promoting the internal self-similarity of the reflectance component, in a way that jointly trains RPN and SPN to reproduce the given image. A novel loss function is also designed to make effective the training for intrinsic decomposition. Experimental results on three benchmark real-world datasets demonstrate the superiority of the proposed method.

Understanding and predicting types of bugs are of practical importance for developers to improve the testing efficiency and take appropriate steps to address bugs in software releases. However, due to the complex conditions under which faults manifest and the complexity of the classification rules, the automatic classification of Mandelbugs is a difficult task. In this article, we present a deep semantic information-based Mandelbug classification method that combines a semantic model with a deep learning classifier and makes use of both labeled and unlabeled bug reports. By training the bug report semantic model on millions of bug reports, each word in the text of a bug report is represented as a word embedding that preserves the semantic relationship among the words. Then, a convolutional neural network model is designed to capture the high-level features of bug reports to obtain a more accurate classification. Moreover, the effects of the semantic model size and domain on the classification results are investigated, and the quality of word embeddings is evaluated by analyzing several important parameters.

Few-shot learning, aiming to learn novel concepts from one or a few labeled examples, is an interesting and very challenging problem with many practical advantages. Existing few-shot methods usually utilize data of the same classes to train the feature embedding module and in a row, which is unable to learn adapting to new tasks. Besides, traditional few-shot models fail to take advantage of the valuable relations of the support-query pairs, leading to performance degradation. In this article, we propose a transductive relation-propagation graph neural network (GNN) with a decoupling training strategy (TRPN-D) to explicitly model and propagate such relations across support-query pairs, and empower the few-shot module the ability of transferring past knowledge to new tasks via the decoupling training. Our few-shot module, namely TRPN, treats the relation of each support-query pair as a graph node, named relational node, and resorts to the known relations between support samples, including both intraclass commonality and interclass uniqueness. Through relation propagation, the model could generate the discriminative relation embeddings for support-query pairs. To the best of our knowledge, this is the first work that decouples the training of the embedding network and the few-shot graph module with different tasks, which might offer a new way to solve the few-shot learning problem. Extensive experiments conducted on several benchmark datasets demonstrate that our method can significantly outperform a variety of state-of-the-art few-shot learning methods.

Searchable Symmetric Encryption (SSE) is a key tool for secure data processing. To date, most of the SSEs were studied alone, while an SSE supporting update operations over encrypted data remained a challenging problem due to various statistical attacks and multi-user environments. In this paper, we propose SeUpdate, the first SSE scheme that simultaneously achieves keyword search and controlled update over encrypted data, with flexible read (search) and write (update) access control policies among multiple users. In SeUpdate, users do not need to share secret keys and a single query enables one to efficiently search all his authorized data. We formally define a security model, and prove our scheme have both forward and backward security. We note that the write permission of an SSE is realized for the first time. We further extend the basic scheme with dynamic access policy update and support of a large number of files. We also implement SeUpdate and some related work. The theoretical and experimental analyses demonstrate our scheme and its extension are practical and efficient.

Weakly supervised semantic segmentation is receiving great attention due to its low human annotation cost. In this paper, we aim to tackle bounding box supervised semantic segmentation, i.e., training accurate semantic segmentation models using bounding box annotations as supervision. To this end, we propose Affinity Attention Graph Neural Network ($A^2$GNN). Following previous practices, we first generate pseudo semantic-aware seeds, which are then formed into semantic graphs based on our newly proposed affinity Convolutional Neural Network (CNN). Then the built graphs are input to our $A^2$GNN, in which an affinity attention layer is designed to acquire the short- and long- distance information from soft graph edges to accurately propagate semantic labels from the confident seeds to the unlabeled pixels. However, to guarantee the precision of the seeds, we only adopt a limited number of confident pixel seed labels for $A^2$GNN, which may lead to insufficient supervision for training. To alleviate this issue, we further introduce a new loss function and a consistency-checking mechanism to leverage the bounding box constraint, so that more reliable guidance can be included for the model optimization. Experiments show that our approach achieves new state-of-the-art performances on Pascal VOC 2012 datasets (val: 76.5\%, test: 75.2\%). More importantly, our approach can be readily applied to bounding box supervised instance segmentation task or other weakly supervised semantic segmentation tasks, with state-of-the-art or comparable performance among almot all weakly supervised tasks on PASCAL VOC or COCO dataset. Our source code will be available at https://github.com/zbf1991/A2GNN., Accepted by IEEE Transactions on Pattern Analysis and Machine Intelligence (TAPMI 2021)

Introduction Evidence of programmed death-1 inhibitors in nasopharyngeal carcinoma has been accumulated. However, previous clinical studies were basically small sample size. Objective This study aimed to summarize existing studies to comprehensively compare programmed death-1 inhibitors in nasopharyngeal carcinoma with or without chemotherapy. Methods Different databases were searched for full-text publications with a programmed death-1 inhibitor with or without chemotherapy. No study-to-study heterogeneity was detected, and fixed-effect models were applied to synthesize data. Results Seven studies were included. The mean progression-free survival duration of programmed death-1 inhibitors treatment was 4.66 months. The 6 month progression-free survival rate was 50%, however, the12 month progression-free survival rate fell to 27%. Comparing with programmed death-1 inhibitor monotherapy, the objective response rate was higher in combination therapy (pooled RR = 2.90, 95% CI: 2.07-4.08). The partial response rate was higher in patients receiving programmed death-1 in association with chemotherapy (pooled RR = 3.09, 95% CI: 2.15-4.46), In contrast, the progressive disease rate was lower in combination therapy group (pooled RR = 0.06, 95% CI: 0.01-0.31). Stable disease condition was comparable (pooled RR = 0.90, 95% CI: 0.50-1.64) with or without chemotherapy. Programmed death-1 single use or combined with chemotherapy did not influence the total adverse events occurrence (pooled RR = 0.99, 95% CI: 0.93-1.05). However, combination therapy could increase the risk of serious adverse events such as anemia, thrombocytopenia, and neutropenia. Conclusion The present study summarized the efficacy and safety of programmed death-1 inhibitors in nasopharyngeal carcinoma. Combination therapy showed higher anti-tumor activity except for higher risk of myelosuppression. Resumo Introdução Há um acúmulo de evidências sobre os inibidores de morte celular programada‐1 no carcinoma nasofaríngeo. Porém, os estudos clínicos anteriores foram quase todos feitos com amostras de tamanho pequeno. Objetivo Resumir os estudos existentes para comparar de forma abrangente os inibidores de morte celular programada‐1 em carcinoma nasofaríngeo com ou sem quimioterapia. Método A pesquisa foi feita em diferentes bases de dados em busca de publicações de texto completo que usaram um inibidor de morte celular programada‐1 com ou sem quimioterapia. Nenhuma heterogeneidade entre os estudos foi detectada e modelos de efeito fixo foram aplicados para sintetizar os dados. Resultados Sete estudos foram incluídos. A duração média da sobrevida livre de progressão no tratamento com inibidores de morte celular programada‐1 foi de 4,66 meses. A taxa de sobrevida livre de progressão em seis meses foi de 50%; no entanto, a taxa de sobrevida livre de progressão em 12 meses caiu para 27%. Em comparação com a monoterapia com inibidor de morte celular programada‐1, a taxa de resposta objetiva (taxa de resposta combinada = 2,90, IC de 95%: 2,07-4,08). A taxa de resposta parcial foi maior em pacientes que receberam morte celular programada‐1 em combinação com quimioterapia (taxa de resposta combinada = 3,09, IC 95%: 2,15‐4,46). Ao contrário, a taxa de progressão da doença foi menor no grupo com terapia combinada (taxa de resposta combinada = 0,06, IC de 95%: 0,01-0,31). A condição de estabilidade da doença com ou sem quimioterapia foi comparável (taxa de resposta combinada = 0,90, IC de 95%: 0,50-1,64). O uso isolado de morte celular programada‐1 ou combinado com quimioterapia não influenciou a ocorrência de eventos adversos totais (taxa de resposta combinada = 0,99, IC de 95%: 0,93-1,05). No entanto, a terapia combinada pode aumentar o risco de eventos adversos graves, como anemia, trombocitopenia e neutropenia. Conclusão O presente estudo fez um resumo da eficácia e segurança dos inibidores de morte celular programada‐1 em carcinoma nasofaríngeo. A terapia combinada mostrou uma atividade antitumoral maior, excetuando‐se o risco maior de mielossupressão.

Traditional data sharing systems are facing new challenges when implementing access control with more and more complex data sharing requirements. Flexibility of user revocation in completely decentralized environments needs to be taken into account. In this paper, we propose a Key-Policy Attribute-Based Encryption scheme with Multiple Authorities and Flexible Revocation (MAFR-KP-ABE) to achieve the features of decentralized authorization and flexible revocation. We prove the security of our MAFR-KP-ABE scheme in the standard model and provide the comparison with relevant schemes to demonstrate its efficiency. Then we propose a fine-grained access control system based on MAFR-KP-ABE scheme and blockchain that matches the need of paid data sharing services with several security properties enhanced. Security analysis and system implementation are given subsequently to demonstrate our system efficient and secure.

The intentional use of viruses for cancer therapy dates back over a century. As viruses are inherently immunogenic and naturally optimized delivery vehicles, repurposing viruses for drug delivery, tumor antigen presentation, or selective replication in cancer cells represents a simple and elegant approach to cancer treatment. While early virotherapy was fraught with harsh side effects and low response rates, virus-based therapies have recently seen a resurgence due to newfound abilities to engineer and tune oncolytic viruses, virus-like particles, and virus-mimicking nanoparticles for improved safety and efficacy. However, despite their great potential, very few virus-based therapies have made it through clinical trials. In this review, we present an overview of virus-inspired approaches for cancer therapy, discuss engineering strategies to enhance their mechanisms of action, and highlight their application for overcoming the challenges of traditional cancer therapies.

This article deals with the problem of secure distributed adaptive platooning control of automated vehicles over vehicular ad-hoc networks (VANETs) in the presence of intermittent denial-of-service (DoS) attacks. The platoon, which is wirelessly connected via directed vehicle-to-vehicle (V2V) communication, is composed of a group of following vehicles subject to unknown heterogeneous nonlinearities and external disturbance inputs, and a leading vehicle subject to unknown nonlinearity and external disturbance as well as an unknown control input. Under such a platoon setting, this article aims to accomplish secure distributed platoon formation tracking with the desired longitudinal spacing and the same velocities and accelerations guided by the leader regardless of the simultaneous presence of nonlinearities, uncertainties, and DoS attacks. First, a new logical data packet processor is developed on each vehicle to identify the intermittent DoS attacks via verifying the time-stamps of the received data packets. Then, a scalable distributed neural-network-based adaptive control design approach is proposed to achieve secure platooning control. It is proved that under the established design procedure, the vehicle state estimation errors and platoon tracking errors can be regulated to reside in small neighborhoods around zero. Finally, comparative simulation studies are provided to substantiate the effectiveness and merits of the proposed control design approach on maintaining the desired platooning performance and attack tolerance.

An interesting phenomenon was found that fluorophore was introduced into ionic liquid (IL) to magnify fluorescence signal via the thermally activated delayed fluorescence process; thus up to 15 fold enhancements were achieved. Hence, we reported a succinct enhanced luminescence strategy to reduce single-triplet energy split by the tunability of ILs. This strategy could be extended to more kinds of ILs by the virtue of a preliminary DFT calculation screening. Moreover, the optical feature of IL sensor made it a promising candidate as a gas fluorescence sensor.

Multi-FPGA architecture has gained great interests in accelerating large-scale graph processing with great potential on high throughput and energy efficiency. As a beneficial complement, work stealing functions effectively to balance the computational workload on different FPGAs dynamically. Unfortunately, existing graph partitioning schemes originally designed in distributed settings potentially mismatch the work stealing-enabled multi-FPGA situations, where the computation is balanced while the communication overhead is unprecedentedly significant over otherwise. In this paper, we present a 2-dimension balanced graph partitioning for work stealing assisted graph system on multi-FPGAs, which can reduce communication overhead while preserving the optimal performance of work stealing. Our approach is novel by 1) exploring the tradeoff between load balance dimension and communication dimension in work-stealing-enabled graph processing system for the optimal performance, and 2) optimizing the memory access sequences to improve the granularity of graph partitioning for high-throughput graph analytics. Our experimental results show that our approach achieves 1.63∼2.56x speedups compared with state-of-the-art FPGA-based systems.

Despite the success of stochastic variance-reduced gradient (SVRG) algorithms in solving large-scale problems, their stochastic gradient complexity often scales linearly with data size and is expensive for huge data. Accordingly, we propose a hybrid stochastic-deterministic minibatch proximal gradient~(HSDMPG) algorithm for strongly convex problems with linear prediction structure, e.g.~least squares and logistic/softmax regression. HSDMPG~enjoys improved computational complexity that is data-size-independent for large-scale problems. It iteratively samples an evolving~minibatch of individual losses to estimate the original problem, and efficiently minimizes the sampled smaller-sized subproblems. For strongly convex loss of n components, HSDMPG~attains an $\epsilon$ -optimization-error within $\mathcal{O} \left(\kappa\log^{\zeta+1}\left(\frac{1}{\epsilon}\right)\frac{1}{\epsilon}\bigwedge n\log^{\zeta}\left(\frac{1}{\epsilon}\right)\right)$ stochastic gradient evaluations, where $\kappa$ is condition number, $\zeta=1$ for quadratic loss and $\zeta=2$ for generic loss. For large-scale problems, our complexity outperforms those of SVRG-type algorithms with/without dependence on data size. Particularly, when $\epsilon=\mathcal{O}(1/\sqrt{n})$ which matches the intrinsic excess error of a learning model and is sufficient for generalization, our complexity for quadratic and generic losses is respectively $\mathcal{O} (n^{0.5}\log^{2}(n))$ and $\mathcal{O} (n^{0.5}\log^{3}(n))$ , which for the first time achieves optimal generalization in less than a single pass over data. Besides, we extend HSDMPG~to online strongly convex problems and prove its higher efficiency over the prior algorithms. Numerical results demonstrate the computational advantages of~HSDMPG.

Since real-world objects and their interactions are often multi-modal and multi-typed, heterogeneous networks have been widely used as a more powerful, realistic, and generic superclass of traditional homogeneous networks (graphs). Meanwhile, representation learning (\aka~embedding) has recently been intensively studied and shown effective for various network mining and analytical tasks. In this work, we aim to provide a unified framework to deeply summarize and evaluate existing research on heterogeneous network embedding (HNE), which includes but goes beyond a normal survey. Since there has already been a broad body of HNE algorithms, as the first contribution of this work, we provide a generic paradigm for the systematic categorization and analysis over the merits of various existing HNE algorithms. Moreover, existing HNE algorithms, though mostly claimed generic, are often evaluated on different datasets. Understandable due to the application favor of HNE, such indirect comparisons largely hinder the proper attribution of improved task performance towards effective data preprocessing and novel technical design, especially considering the various ways possible to construct a heterogeneous network from real-world application data. Therefore, as the second contribution, we create four benchmark datasets with various properties regarding scale, structure, attribute/label availability, and \etc.~from different sources, towards handy and fair evaluations of HNE algorithms. As the third contribution, we carefully refactor and amend the implementations and create friendly interfaces for 13 popular HNE algorithms, and provide all-around comparisons among them over multiple tasks and experimental settings., Accepted by IEEE TKDE. All code and data available at https://github.com/yangji9181/HNE

Background: Upregulation of mitochondrial E3 ubiquitin ligase 1 (Mul1) contributes to brain injury in ischemic stroke due to disturbance of mitochondrial dynamics, and bioinformatics analysis predicts that Mul1 is a potential target of Dipsacoside B. Objective: The aim of the study was to explore whether Dipsacoside B can exert a beneficial effect on brain injury in the ischemic stroke rat via targeting Mul1. Methods: The SD rat brains or PC12 cells were subjected to 2 h-ischemia or 8 h-hypoxia plus 24 h-reperfusion or 24 h-reoxygenation to establish the ischemic stroke rat model in vivo or in vitro, which were treated with Dipsacoside B at different dosages. The brain or PC12 cell injury, relevant protein levels and mitochondrial functions were measured by methods of biochemistry, flow cytometry or Western blot. Results: The neurological dysfunction and brain injury (such as infarction and apoptosis) observed in the ischemic stroke rats were accompanied by increases in Mul1 and dynamin-related protein 1 (Drp1) levels along with decreases in mitofusin 2 (Mfn2) level and ATP production. These effects were attenuated by Dipsacoside B. Consistently, cell injury (necroptosis and apoptosis) occurred in the PC12 cells exposed to hypoxia concomitant with the upregulation of Mul1 and Drp1 along with downregulation of Mfn2 and mitochondrial functions (such as increases in reactive oxygen species production and mitochondrial fission and decreases in mitochondrial membrane potential and ATP production).These phenomena were reversed in the presence of Dipsacoside B. Conclusion: Dipsacoside B can protect the rat brain against ischemic injury via inhibition of Mul1 due to the improvement of mitochondrial function.

This paper investigates the stochastic optimization problem with a focus on developing scalable parallel algorithms for deep learning tasks. Our solution involves a reformation of the objective function for stochastic optimization in neural network models, along with a novel parallel strategy, coined weighted aggregating stochastic gradient descent (WASGD). Following a theoretical analysis on the characteristics of the new objective function, WASGD introduces a decentralized weighted aggregating scheme based on the performance of local workers. Without any center variable, the new method automatically assesses the importance of local workers and accepts them according to their contributions. Furthermore, we have developed an enhanced version of the method, WASGD+, by (1) considering a designed sample order and (2) applying a more advanced weight evaluating function. To validate the new method, we benchmark our schemes against several popular algorithms including the state-of-the-art techniques (e.g., elastic averaging SGD) in training deep neural networks for classification tasks. Comprehensive experiments have been conducted on four classic datasets, including the CIFAR-100, CIFAR-10, Fashion-MNIST, and MNIST. The subsequent results suggest the superiority of the WASGD scheme in accelerating the training of deep architecture. Better still, the enhanced version, WASGD+, has been shown to be a significant improvement over its basic version.

For multimodal representation learning, traditional black-box approaches often fall short of extracting interpretable multilayer hidden structures, which contribute to visualize the connections between different modalities at multiple semantic levels. To extract interpretable multimodal latent representations and visualize the hierarchial semantic relationships between different modalities, based on deep topic models, we develop a novel multimodal Poisson gamma belief network (mPGBN) that tightly couples the observations of different modalities via imposing sparse connections between their modality-specific hidden layers. To alleviate the time-consuming Gibbs sampler adopted by traditional topic models in the testing stage, we construct a Weibull-based variational inference network (encoder) to directly map the observations to their latent representations, and further combine it with the mPGBN (decoder), resulting in a novel multimodal Weibull variational autoencoder (MWVAE), which is fast in out-of-sample prediction and can handle large-scale multimodal datasets. Qualitative evaluations on bimodal data consisting of image-text pairs show that the developed MWVAE can successfully extract expressive multimodal latent representations for downstream tasks like missing modality imputation and multimodal retrieval. Further extensive quantitative results demonstrate that both MWVAE and its supervised extension sMWVAE achieve state-of-the-art performance on various multimodal benchmarks.

Although demonstrating great success, previous multi-view unsupervised feature selection (MV-UFS) methods often construct a view-specific similarity graph and characterize the local structure of data within each single view. In such a way, the cross-view information could be ignored. In addition, they usually assume that different feature views are projected from a latent feature space while the diversity of different views cannot be fully captured. In this work, we resent a MV-UFS model via cross-view local structure preserved diversity and consensus learning, referred to as CvLP-DCL briefly. In order to exploit both the shared and distinguishing information across different views, we project each view into a label space, which consists of a consensus part and a view-specific part. Therefore, we regularize the fact that different views represent same samples. Meanwhile, a cross-view similarity graph learning term with matrix-induced regularization is embedded to preserve the local structure of data in the label space. By imposing the $l_{2,1}$ -norm on the feature projection matrices for constraining row sparsity, discriminative features can be selected from different views. An efficient algorithm is designed to solve the resultant optimization problem and extensive experiments on six publicly datasets are conducted to validate the effectiveness of the proposed CvLP-DCL.

Spatially localized deformation components are very useful for shape analysis and synthesis in 3D geometry processing. Several methods have recently been developed, with an aim to extract intuitive and interpretable deformation components. However, these techniques suffer from fundamental limitations especially for meshes with noise or large-scale nonlinear deformations, and may not always be able to identify important deformation components. In this paper we propose a mesh-based variational autoencoder architecture that is able to cope with meshes with irregular connectivity and nonlinear deformations, assuming that the analyzed dataset contains meshes with the same vertex connectivity, which is common for deformation analysis. To help localize deformations, we introduce sparse regularization in this framework, along with spectral graph convolutional operations. Through modifying the regularization formulation and allowing dynamic change of sparsity ranges, we improve the visual quality and reconstruction ability of the extracted deformation components. Our system also provides a nonlinear approach to reconstruction of meshes using the extracted basis, which is more effective than the current linear combination approach. As an important application of localized deformation components and a novel approach on its own, we further develop a neural shape editing method, achieving shape editing and deformation component extraction in a unified framework, and ensuring plausibility of the edited shapes. Extensive experiments show that our method outperforms state-of-the-art methods in both qualitative and quantitative evaluations. We also demonstrate the effectiveness of our method for neural shape editing.

This study numerically simulated and investigated the flow field characteristics of a typical dual-pulse solid rocket motor with a soft pulse separation device through thermal insulation ablation under high-temperature dual-pulse erosion. The ablation rate of ethylene -propylene-diene monomer (EPDM) insulator was measured after the experiment. Experimental results were analyzed through scanning electron microscopy and microcomputed tomography. The ablation mechanism of the EPDM insulator under the operation conditions of a dual-pulse solid rocket motor was evaluated by analyzing the results. The results reveal that the internal flow field of the motor with a soft pulse separation device is uniform. The original charred layer existing on the EPDM insulator surface in the first pulse combustor is the decisive factor affecting the final ablation rate of the dual-pulse motor during the second pulse operation, and the ablation characteristic region is easily formed with the exfoliation of the charred layer. The ablation rate difference of the insulator increases with gas velocity.

Recently, reconfigurable intelligent surface (RIS) has attracted continuous attention in wireless communications. Because the normal incidence condition of electromagnetic waves cannot always be satisfied in practical applications, the angular insensitivity in RISs has become an important issue. The angular sensitivity in RISs may lead to the failure of RIS-assisted wireless communication networks that rely on the reciprocity of wireless channels. In this work, an angle-insensitive 3-bit RIS meta-atom is proposed. By introducing metallic vias in the substrate of programmable meta-atoms, the interference induced by multiple reflections in the multilayered structure is restrained, and the decoupling between adjacent meta-atoms makes the metasurface insensitive to the incident angle. Hence the angular insensitivity can be achieved. An RIS is fabricated using the proposed angleinsensitive programmable meta-atom with metallic vias, which is capable of maintaining stable phase and amplitude responses at oblique incidences. Both simulated and measured results show stable angle performance. A maximum of 3150 phase range and eight digital coding states with 450 stable interval are obtained for wide incidence angles from 00 to 600. Finally, numerical simulations and experimental results of anomalous reflections are performed to verify the angular reciprocity of the proposed RIS, which is important in wireless communications.

With the widespread application of artificial intelligence and the Internet of Things, the intellectualization of the IIoT (Industrial Internet of Things, IIoT) has received more and more attention. There is quite much demand from machine learning services in such scenarios. Therefore, it is urgent to find an adaptive solution for the machine learning service reasoning system rather than rely on manually configuring hardware or pulling DNN (Deep Neural Network, DNN) models. Compared with the system using only edge or only cloud, the collaborative cloud-edge architecture is a better solution in the IIoT scenario. Thus, this paper presents a collaborative cloud-edge cognitive service framework for DNN model service configuration, which is deployed to industrial scenarios and makes full use of the available resources of edge servers and cloud. Meanwhile, a self-adaptive model selection mechanism is established and the reinforcement learning algorithm to realize the optimal model selection strategy.

Recently, functional homotopy (FH) architecture, defined as robust functional connectivity (FC) between homotopic regions, has been frequently reported to be altered in MDD patients (MDDs) but with divergent locations.In this study, we obtained resting-state functional magnetic resonance imaging (R-fMRI) data from 1004 MDDs (mean age, 33.88 years; age range, 18-60 years) and 898 matched healthy controls (HCs) from an aggregated dataset from 20 centers in China. We focused on interhemispheric function integration in MDDs and its correlation with clinical characteristics using voxel-mirrored homotopic connectivity (VMHC) devised to inquire about FH patterns.As compared with HCs, MDDs showed decreased VMHC in visual, motor, somatosensory, limbic, angular gyrus, and cerebellum, particularly in posterior cingulate gyrus/precuneus (PCC/PCu) (false discovery rate [FDR] q 0.002, z = -7.07). Further analysis observed that the reduction in SMG and insula was more prominent with age, of which SMG reflected such age-related change in males instead of females. Besides, the reduction in MTG was found to be a male-special abnormal pattern in MDDs. VMHC alterations were markedly related to episode type and illness severity. The higher Hamilton Depression Rating Scale score, the more apparent VMHC reduction in the primary visual cortex. First-episode MDDs revealed stronger VMHC reduction in PCu relative to recurrent MDDs.We confirmed a significant VMHC reduction in MDDs in broad areas, especially in PCC/PCu. This reduction was affected by gender, age, episode type, and illness severity. These findings suggest that the depressive brain tends to disconnect information exchange across hemispheres.