Strength and dilatancy behaviors of deep sands in Shanghai with a focus on grain size and shape effect

With rapid development of infrastructures like tunnels and open excavations in Shanghai, investigations on deeper soils have become critically important. Most of the existing laboratory works were focused on the clayey strata up to Layer 6 in Shanghai, i.e. at depth of up to 40 m. In this paper, Layers 7, 9, and 11, which were mostly formed of sandy soils at depth of up to 150 m, were experimentally investigated with respect to physico-mechanical behaviors. The stress–strain behaviors were analyzed by the consolidated drained/undrained (CD/CU) triaxial tests under monotonic loading. One-dimensional (1D) oedometer tests were performed to investigate the consolidation properties of the sandy soils. Specimens were prepared at three different relative densities for each layer. Also, the micro-images and particle size analyzers were used to analyze the shape and size of the sand grains. The influences of grain size, density, and angularity on the stress–strain behaviors and compressibility were also studied. Compared to the other layers, Layer 11 had the smallest mean grain size (D₅₀), highest compressibility, and lowest shear strength. In contrast, Layer 9 had the largest mean grain size, lowest compressibility, and highest shear strength. Layer 7 was of intermediate mean grain size, exhibiting more compressibility and less shear strength than that of Layer 9. Also, the critical state parameters and maximum dilatancy rate of different layers were discussed.     

A novel PEM fuel cell remaining useful life prediction method based on singular spectrum analysis and deep Gaussian processes

Accurate remaining useful life (RUL) prediction of proton exchange membrane fuel cells (PEMFCs) can assess the reliability of fuel cells to determine the occurrence of failures and mitigate their operational risk. However, is it quite challenging to design a high-precision prediction method because the implicit degradation details of PEMFCs are difficult to learn well from the measurement data with high-frequency noise. Recognizing this, a novel RUL prediction method based on singular spectrum analysis (SSA) and deep Gaussian process (DGP) is proposed in this paper. The SSA-based method is firstly employed to preprocess the measurement data, which can strengthen the effective information of PEMFC degradation data at the same time remove the noise and spikes that interfere with degradation prediction. As a deep structural model, DGP has strong feature learning ability which can represent the nonlinear details of degradation data and give more accurate prediction results. At the same time, it serves as a probabilistic model that can provide the confidence interval to enhance reliability of RUL prediction. The effectiveness of the proposed method is evaluated by experimental data of the PEMFCs under steady-state conditions, and the results show that the SSA-DGP method has higher accuracy and reliability than conventional methods.     

A Novel Method of Heart Failure Prediction Based on DPCNNXGBOOST Model

The occurrence of perioperative heart failure will affect the quality of medical services and threaten the safety of patients. Existing methods depend on the judgment of doctors, the results are affected by many factors such as doctors’ knowledge and experience. The accuracy is difficult to guarantee and has a serious lag. In this paper, a mixture prediction model is proposed for perioperative adverse events of heart failure, which combined with the advantages of the Deep Pyramid Convolutional Neural Networks (DPCNN) and Extreme Gradient Boosting (XGBOOST). The DPCNN was used to automatically extract features from patient’s diagnostic texts, and the text features were integrated with the preoperative examination and intraoperative monitoring values of patients, then the XGBOOST algorithm was used to construct the prediction model of heart failure. An experimental comparison was conducted on the model based on the data of patients with heart failure in southwest hospital from 2014 to 2018. The results showed that the DPCNN-XGBOOST model improved the predictive sensitivity of the model by 3% and 31% compared with the text-based DPCNN Model and the numeric-based XGBOOST Model.     

Image and Feature Space Based Domain Adaptation for Vehicle Detection

The application of deep learning in the field of object detection has experienced much progress. However, due to the domain shift problem, applying an off-the-shelf detector to another domain leads to a significant performance drop. A large number of ground truth labels are required when using another domain to train models, demanding a large amount of human and financial resources. In order to avoid excessive resource requirements and performance drop caused by domain shift, this paper proposes a new domain adaptive approach to cross-domain vehicle detection. Our approach improves the cross-domain vehicle detection model from image space and feature space. We employ objectives of the generative adversarial network and cycle consistency loss for image style transfer in image space. For feature space, we align feature distributions between the source domain and the target domain to improve the detection accuracy. Experiments are carried out using the method with two different datasets, proving that this technique effectively improves the accuracy of vehicle detection in the target domain.     

Grain boundary strengthening in ZrB2 by segregation of W: Atomistic simulations with deep learning potential

Interaction between grain boundaries and impurities usually leads to significant altering of material properties. Understanding the composition-structure-property relationship of grain boundaries is a key avenue for tailoring and designing high performance materials. In this work, we studied segregation of W into ZrB₂ grain boundaries by a hybrid method combining Monte Carlo (MC) and molecular dynamics (MD), and examined the effects of segregation on grain boundary strengths by MD tensile testing with a fitted machine learning potential. It is found that W prefers grain boundary sites with local compression strains due to its smaller size compared to Zr. Rich segregation patterns (including monolayer, off-center bilayer, and other complex patterns); segregation induced grain boundary structure reconstruction; and order-disorder like segregation pattern transformation are discovered. Strong segregation tendency of W into ZrB₂ grain boundaries and significant improvements on grain boundary strengths are certified, which guarantees outstanding high temperature performance of ZrB₂-based UHTCs.     

基于深度概率图模型的鲁棒人脸画像合成

针对基于数据驱动的人脸画像合成算法像素特征缺乏对光照变化和复杂背景的鲁棒性,常合成低质量的画像的问题,文中提出基于深度概率图模型的鲁棒人脸画像合成算法.采用预处理方法调整测试照片的光照亮度和人脸姿态,使之与训练照片一致.采用深度特征代替像素特征进行近邻匹配,采用深度概率图模型对画像重建权重和深度特征权重联合建模,得到合成画像的最佳重构表示.为了提高画像合成速度,提出快速近邻搜索方法.实验验证文中算法的鲁棒性和快速性.     

Novel DDoS Feature Representation Model Combining Deep Belief Network and Canonical Correlation Analysis

Distributed denial of service (DDoS) attacks launch more and more frequently and are more destructive. Feature representation as an important part of DDoS defense technology directly affects the efficiency of defense. Most DDoS feature extraction methods cannot fully utilize the information of the original data, resulting in the extracted features losing useful features. In this paper, a DDoS feature representation method based on deep belief network (DBN) is proposed. We quantify the original data by the size of the network flows, the distribution of IP addresses and ports, and the diversity of packet sizes of different protocols and train the DBN in an unsupervised manner by these quantified values. Two feedforward neural networks (FFNN) are initialized by the trained deep belief network, and one of the feedforward neural networks continues to be trained in a supervised manner. The canonical correlation analysis (CCA) method is used to fuse the features extracted by two feedforward neural networks per layer. Experiments show that compared with other methods, the proposed method can extract better features.     

基于特征选择和深度信念网络的文本情感分类算法

由于人类语言的复杂性，文本情感分类算法大多都存在因为冗余而造成的词汇量过大的问题。深度信念网络（DBN）通过学习输入语料中的有用信息以及它的几个隐藏层来解决这个问题。然而对于大型应用程序来说，DBN是一个耗时且计算代价昂贵的算法。针对这个问题，提出了一种半监督的情感分类算法，即基于特征选择和深度信念网络的文本情感分类算法（FSDBN）。首先使用特征选择方法（文档频率（DF）、信息增益（IG）、卡方统计（CHI）、互信息（MI））过滤掉一些不相关的特征从而使词汇表的复杂性降低；然后将特征选择的结果输入到DBN中，使得DBN的学习阶段更加高效。将所提算法应用到中文以及维吾尔语中，实验结果表明在酒店评论数据集上，FSDBN在准确率方面比DBN提高了1.6%，在训练时间上比DBN缩短一半。     

Stacked autoencoder based deep random vector functional link neural network for classification

Extreme learning machine (ELM), which can be viewed as a variant of Random Vector Functional Link (RVFL) network without the input–output direct connections, has been extensively used to create multi-layer (deep) neural networks. Such networks employ randomization based autoencoders (AE) for unsupervised feature extraction followed by an ELM classifier for final decision making. Each randomization based AE acts as an independent feature extractor and a deep network is obtained by stacking several such AEs. Inspired by the better performance of RVFL over ELM, in this paper, we propose several deep RVFL variants by utilizing the framework of stacked autoencoders. Specifically, we introduce direct connections (feature reuse) from preceding layers to the fore layers of the network as in the original RVFL network. Such connections help to regularize the randomization and also reduce the model complexity. Furthermore, we also introduce denoising criterion, recovering clean inputs from their corrupted versions, in the autoencoders to achieve better higher level representations than the ordinary autoencoders. Extensive experiments on several classification datasets show that our proposed deep networks achieve overall better and faster generalization than the other relevant state-of-the-art deep neural networks.     

基于优势学习的深度Q网络

强化学习问题中，同一状态下不同动作所对应的状态-动作值存在差距过小的现象，Q-Learning算法采用MAX进行动作选择时会出现过估计问题，且结合了Q-Learning的深度Q网络（Deep Q Net）同样存在过估计问题。为了缓解深度Q网络中存在的过估计问题，提出一种基于优势学习的深度Q网络，通过优势学习的方法构造一个更正项，利用目标值网络对更正项进行建模，同时与深度Q网络的评估函数进行求和作为新的评估函数。当选择的动作是最优动作时，更正项为零，不对评估函数的值进行改动，当选择的动作不是最优动作时，更正项的值为负，降低了非最优动作的评估值。和传统的深度Q网络相比，基于优势学习的深度Q网络在Playing Atari 2600的控制问题breakout、seaquest、phoenix、amidar中取得了更高的平均奖赏值，在krull、seaquest中取得了更加稳定的策略。