Epilepsy is one of the most common neurological diseases. Clinically, epileptic seizure detection is usually performed by analyzing electroencephalography (EEG) signals. At present, deep learning models have been widely used for single-channel EEG signal epilepsy detection, but this method is difficult to explain the classification results. Researchers have attempted to solve interpretive problems by combining graph representation of EEG signals with graph neural network models. Recently, the combination of graph representations and graph neural network (GNN) models has been increasingly applied to single-channel epilepsy detection. By this methodology, the raw EEG signal is transformed to its graph representation, and a GNN model is used to learn latent features and classify whether the data indicates an epileptic seizure episode. However, existing methods are faced with two major challenges. First, existing graph representations tend to have high time complexity as they generally require each vertex to traverse all other vertices to construct a graph structure. Some of them also have high space complexity for being dense. Second, while separate graph representations can be derived from a single-channel EEG signal in both time and frequency domains, existing GNN models for epilepsy detection can learn from a single graph representation, which makes it hard to let the information from the two domains complement each other. For addressing these challenges, we propose a Weighted Neighbour Graph (WNG) representation for EEG signals. Reducing the redundant edges of the existing graph, WNG can be both time and space-efficient, and as informative as its less efficient counterparts. We then propose a two-stream graph-based framework to simultaneously learn features from WNG in both time and frequency domain. Extensive experiments demonstrate the effectiveness and efficiency of the proposed methods.

Summary Transforming seismic data to other domains to improve sparsity then suppressing surface wave is the main idea of traditional surface wave suppression. In this paper, we proposed a new surface wave suppression method based on deep neural network, DeSurface. This method can learn a nonlinear function from the input seismic data sparsely represented in the time-frequency domain, which maps the representation to masks and decomposition the input data into effective signals and surface waves. Then, the proposed method is applied to common shot gathers to verify its feasibility, and compared with other surface wave suppression methods, it is proved that the proposed method has the ability to improve the surface wave suppression effect.

The recent years we have seen the rise of graph neural networks for prediction tasks on graphs. One of the dominant architectures is graph attention due to its ability to make predictions using weighted edge features and not only node features. In this paper we analyze, theoretically and empirically, graph attention networks and their ability of correctly labelling nodes in a classic classification task. More specifically, we study the performance of graph attention on the classic contextual stochastic block model (CSBM). In CSBM the nodes and edge features are obtained from a mixture of Gaussians and the edges from a stochastic block model. We consider a general graph attention mechanism that takes random edge features as input to determine the attention coefficients. We study two cases, in the first one, when the edge features are noisy, we prove that the majority of the attention coefficients are up to a constant uniform. This allows us to prove that graph attention with edge features is not better than simple graph convolution for achieving perfect node classification. Second, we prove that when the edge features are clean graph attention can distinguish intra- from inter-edges and this makes graph attention better than classic graph convolution.

Machine learning based classifiers are often a black box when considering the contribution of inputs to the output probability of a label, especially with complex non-linear models such as neural networks. A popular way to explain machine learning model outputs in a model agnostic manner is through the use of Shapley values. For our use case of abuse fighting in digital advertisements, one primary impediment of using Shapley values in explanations was a problem of instability. Specifically, the instability problem manifests as explanations for the same example varying greatly due to random sampling in the algorithm. We found it useful to view this problem explicitly as Monte Carlo integration in the form of averaging the model output while varying only a subset of features in the example to be explained. In turn, this guides the number of samples needed to achieve a stable estimate of individual Shapley values and unlocked the use of Shapley value based explainers for our models as well as classifiers in general, including neural networks.

Abstract Background Streptococcus pneumoniae meningitis is a destructive central nervous system (CNS) infection with acute and long-term neurological disorders. Previous studies suggest that p75NTR signaling influences cell survival, apoptosis, and proliferation in brain-injured conditions. However, the role of p75NTR signaling in regulating pneumococcal meningitis (PM)-induced neuroinflammation and altered neurogenesis remains largely to be elucidated. Methods p75NTR signaling activation in the pathological process of PM was assessed. During acute PM, a small-molecule p75NTR modulator LM11A-31 or vehicle was intranasally administered for 3 days prior to S. pneumoniae exposure. At 24 h post-infection, clinical severity, histopathology, astrocytes/microglia activation, neuronal apoptosis and necrosis, inflammation-related transcription factors and proinflammatory cytokines/mediators were evaluated. Additionally, p75NTR was knocked down by the adenovirus-mediated short-hairpin RNA (shRNA) to ascertain the role of p75NTR in PM. During long-term PM, the intranasal administration of LM11A-31 or vehicle was continued for 7 days after successfully establishing the PM model. Dynamic changes in inflammation and hippocampal neurogenesis were assessed. Results Our results revealed that both 24 h (acute) and 7, 14, 28 day (long-term) groups of infected rats showed increased p75NTR expression in the brain. During acute PM, modulation of p75NTR through pretreatment of PM model with LM11A-31 significantly alleviated S. pneumoniae-induced clinical severity, histopathological injury and the activation of astrocytes and microglia. LM11A-31 pretreatment also significantly ameliorated neuronal apoptosis and necrosis. Moreover, we found that blocking p75NTR with LM11A-31 decreased the expression of inflammation-related transcription factors (NF-κBp65, C/EBPβ) and proinflammatory cytokines/mediators (IL-1β, TNF-α, IL-6 and iNOS). Furthermore, p75NTR knockdown induced significant changes in histopathology and inflammation-related transcription factors expression. Importantly, long-term LM11A-31 treatment accelerated the resolution of PM-induced inflammation and significantly improved hippocampal neurogenesis. Conclusion Our findings suggest that the p75NTR signaling plays an essential role in the pathogenesis of PM. Targeting p75NTR has beneficial effects on PM rats by alleviating neuroinflammation and promoting hippocampal neurogenesis. Thus, the p75NTR signaling may be a potential therapeutic target to improve the outcome of PM.

Video sharing (e.g., YouTube, Vimeo, Facebook, TikTok) accounts for the majority of internet traffic, and video processing is also foundational to several other key workloads (video conferencing, virtual/augmented reality, cloud gaming, video in Internet-of-Things devices, etc.). The importance of these workloads motivates larger video processing infrastructures and – with the slowing of Moore’s law – specialized hardware accelerators to deliver more computing at higher efficiencies. This paper describes the design and deployment, at scale, of a new accelerator targeted at warehouse-scale video transcoding. We present our hardware design including a new accelerator building block – the video coding unit (VCU) – and discuss key design trade-offs for balanced systems at data center scale and co-designing accelerators with large-scale distributed software systems. We evaluate these accelerators “in the wild" serving live data center jobs, demonstrating 20-33x improved efficiency over our prior well-tuned non-accelerated baseline. Our design also enables effective adaptation to changing bottlenecks and improved failure management, and new workload capabilities not otherwise possible with prior systems. To the best of our knowledge, this is the first work to discuss video acceleration at scale in large warehouse-scale environments.

This paper revisits the problem of rate distortion optimization (RDO) with focus on inter-picture dependence. A joint RDO framework which incorporates the Lagrange multiplier as one of parameters to be optimized is proposed. Simplification strategies are demonstrated for practical applications. To make the problem tractable, we consider an approach where prediction residuals of pictures in a video sequence are assumed to be emitted from a finite set of sources. Consequently the RDO problem is formulated as finding optimal coding parameters for a finite number of sources, regardless of the length of the video sequence. Specifically, in cases where a hierarchical prediction structure is used, prediction residuals of pictures at the same prediction layer are assumed to be emitted from a common source. Following this approach, we propose an iterative algorithm to alternatively optimize the selections of quantization parameters (QPs) and the corresponding Lagrange multipliers. Based on the results of the iterative algorithm, we further propose two practical algorithms to compute QPs and the Lagrange multipliers for the RA(random access) hierarchical video coding: the first practical algorithm uses a fixed formula to compute QPs and the Lagrange multipliers, and the second practical algorithm adaptively adjusts both QPs and the Lagrange multipliers. Experimental results show that these three algorithms, integrated into the HM 16.20 reference software of HEVC, can achieve considerable RD improvements over the standard HM 16.20 encoder, in the common RA test configuration.

E-cash has its merits comparing with other payment modes. However, there are two problems, which are how to achieve practical/complete tracing and how to achieve it in compact E-cash. First, the bank and the TTP (i.e., trusted third party) have different duties and powers in the reality. Therefore, double-spending tracing is bank's task, while unconditional tracing is TTP's task. In addition, it is desirable to provide lost-coin tracing before they are spent by anyone else. Second, compact E-cash is an efficient scheme, but tracing the coins from double-spender without TTP results in poor efficiency. To solve the problems, we present a compact E-cash scheme. For this purpose, we design an embedded structure of knowledge proof based on a new pseudorandom function and improve the computation complexity from O(k) to O(1). Double-spending tracing needs leaking dishonest users' secret knowledge, but preserving the anonymity of honest users needs zero-knowledge property, and our special knowledge proof achieves it with complete proofs. Moreover, the design is also useful for other applications, where both keeping zero-knowledge and leaking information are necessary.

The reliability function of variable-rate Slepian-Wolf coding is linked to the reliability function of channel coding with constant composition codes, through which computable lower and upper bounds are derived. The bounds coincide at rates close to the Slepian-Wolf limit, yielding a complete characterization of the reliability function in that rate region. It is shown that variable-rate Slepian-Wolf codes can significantly outperform fixed-rate Slepian-Wolf codes in terms of rate-error tradeoff. Variable-rate Slepian-Wolf coding with rate below the Slepian-Wolf limit is also analyzed. In sharp contrast with fixed-rate Slepian-Wolf codes for which the correct decoding probability decays to zero exponentially fast if the rate is below the Slepian-Wolf limit, the correct decoding probability of variable-rate Slepian-Wolf codes can be bounded away from zero.

Streptococcus pneumoniae meningitis is a serious inflammatory disease of the central nervous system (CNS) and is associated with high morbidity and mortality rates. The inflammatory processes initiated by recognition of bacterial components contribute to apoptosis in the hippocampal dentate gyrus. Brain-derived neurotrophic factor (BDNF) has long been recommended for the treatment of CNS diseases due to its powerful neuro-survival properties, as well as its recently reported anti-inflammatory and anti-apoptotic effects in vitro and in vivo.

目的 探讨迷走神经刺激术(VNS)治疗儿童难治性癫痫的临床疗效.方法 回顾性分析5例接受VNS治疗的难治性癫痫患儿临床资料,术后2周开机并根据患儿个体癫痫发作情况逐步调整刺激参数,通过患者来院或电话随访至术后1年.结果 该组患儿术前癫痫发作频率为(45.6±55.3)次/个月,术后6个月癫痫发作频率降为(38.2±47.3)次/个月,发作频率平均减少29.67％;患儿术后1年癫痫发作频率降为(25.2±31.8)次/个月,发作频率平均减少48％,与术前相比,患儿癫痫发作频率减少,且发作程度减轻、生活质量提升.结论 VNS手术创伤小、耐受性良好,术后可减少患儿癫痫发作频率,提高生活质量,对于不适合开颅的儿童难治性癫痫来说是一种安全、有效的治疗方法.

Objective To explore the effect of human leukocyte antigen B* genotype and age on serum homocysteine (Hcy) levels in children with seizures or epilepsy. Methods Fifteen children with seizures or epilepsy in whom HLA-B*15:02 genotype was detected during October 2015 to June 2016 were included. The plasma Hcy concentration in children with different genotypes was compared. The association of Hcy concentration and age was performed by linear-regression analysis. Results The mean concentration of Hcy was 8.38±4.23 μmol/L in children not carrying HLA-B*15:02 gene, which was obviously higher than that in children carrying HLA-B*15:02 gene 13.03±0.97 μmol/L (P<0.05). The Hcy concentration increased with the age (r2 =0.29, P<0.05). Conclusions Elder children with seizures or epilepsy carrying HLA-B*15:02 gene tend to have higher Hcy concentration and increased potential risk of disease. HLA-B*15:02 gene type and age can predict the changes of Hcy concentration in children with convulsions.

弱监督关系抽取利用已有关系实体对从文本集中自动获取训练数据,有效解决了训练数据不足的问题.针对弱监督训练数据存在噪声、特征不足和不平衡,导致关系抽取性能不高的问题,文中提出NF-Tri-training(Tri-training with Noise Filtering)弱监督关系抽取算法.它利用欠采样解决样本不平衡问题,基于Tri-training从未标注数据中迭代学习新的样本,提高分类器的泛化能力,采用数据编辑技术识别并移除初始训练数据和每次迭代产生的错标样本.在互动百科采集数据集上实验结果表明NF-Tri-training算法能够有效提升关系分类器的性能.

Bi-geometric transparent composite models (BGTCM) are used to model distributions of transform coefficients in HEVC (High efficiency video coding). Both Kullback-Leibler divergence and χ 2 test show that, for both original and quantized transform coefficients in HEVC, BGTCMs provide better modelling performance than popular Laplacian and Cauchy models. Based on BGTCMs, a rate control algorithm is proposed for HEVC. Experimental results using the HEVC reference software show that the proposed algorithm achieves better performance in constant-bit-rate control than previous rate control algorithms based on Laplacian models.