SE-MCNN-CTC的中文语音识别声学模型

针对国内外缺少对振动轮噪声预估的问题,以某型振动轮为研究对象,首先基于动力学有限元理论对振动轮进行频率响应分析,其次采用声学边界元技术对振动轮辐射噪声进行了数值模拟,并通过实验验证了仿真结果的准确性,然后比较了垂直振动与圆周振动两种不同激振形式对辐射噪声的影响,得出垂直振动辐射噪声低的结论,最后对驾驶室声腔模态进行了仿真,与振动轮激振频率相近发生共振。通过调整激振频率,降低了司机耳旁噪声。所得研究成果可为振动轮辐射噪声的预估与改进提供一种切实可行的参考依据。     

ALGA在岸基导弹航路规划中的应用研究

针对基本遗传算法(GA)易局部收敛的缺陷,设计了基于模式搜索的自学习算子,提出一种基于模式搜索的自学习遗传算法(ALGA)。通过仿真测试函数将ALGA与基本遗传算法、自适应遗传算法(AGA)进行比较,显示改进的ALGA提高了算法的综合搜索能力。将改进的ALGA运用到岸基导弹航路规划中,并进行仿真实验,仿真结果验证了改进算法的有效性。     

经典电路实验翻转课堂教学模式的设计

以经典电路实验课程《RC过渡电路性能的验证》为例，给出了其翻转课堂教学模式的设计。课前以网络平台为载体推送学习资料，课上测试合格的学生才被允许进行实验，实验前给出启发式问题，实验后组织学生分组讨论，教师可针对学生的学习情况进行个性化指导，实现了自主学习+沉浸性操作+个性化教育的学习模式。教学实践情况表明，翻转课堂有效解决了实验学时不足的难题，效果良好。     

我国青少年科学教育的历史与展望

本文回顾了我国科学教育的历史;论述了科学教育是基础教育阶段的核心课程以及在国际上受到的重视;分析了我国科学教育的现状,并提出加强科学教育的建议.     

Diagnosis of learner dropout based on learning styles for online distance learning

The amount of data generated by computer systems in Online Distance Learning (ODL) contains rich information. One example of this information we define as the Learner Learning Trail (LLT), which is the sequence of interactions between the students and the virtual environment. Another example is the Learner Learning Style (LLS), which is associated with the student behavior and choices during the learning process. This information can be used to identify learner behavior and learning style. We perceived, after the study of related literature, that the research field of learner diagnosis for ODL does not apply the conjoint use of LLT and LLS. In this article, we propose a model capable of integrating data generated from the behavior of students in ODL with cognitive aspects of them, such as their Learning Styles, by crossing LLT with LLS. We also propose the CPAD method (Collect, Preprocessing, Analysis, Diagnosis), which is implemented by collecting the raw data regarding learning activities, preprocessing the data into structured time sequences, analyzing the sequences regarding the learning styles and using this analysis to diagnose the learner behavior. We selected the dropout to investigate, once the dropout rate in ODL is a real problem in universities around the world. In addition, the dropout is a student decision which can be associated with previous students behaviors. We performed a study with 202 learners to evaluate if learning styles are capable of explaining aspects of the student behavior. The results suggest that Sequential/Global learning style dimension is more capable of explaining the dropout than the other dimensions. Also, we performed four classification experiments to verify how the dimensions of Felder-Silverman Learning Style Model influence the learner diagnosis. We perceived that the Sequential/Global dimension could provide a higher accuracy average with lower variation independently of the diagnosis technique.     

Hybrid internet traffic classification technique

Accurate and real-time classification of network traffic is significant to network operation and management such as QoS differentiation, traffic shaping and security surveillance. However, with many newly emerged P2P applications using dynamic port numbers, masquerading techniques, and payload encryption to avoid detection, traditional classification approaches turn to be ineffective. In this paper, we present a layered hybrid system to classify current Internet traffic, motivated by variety of network activities and their requirements of traffic classification. The proposed method could achieve fast and accurate traffic classification with low overheads and robustness to accommodate both known and unknown/encrypted applications. Furthermore, it is feasible to be used in the context of real-time traffic classification. Our experimental results show the distinct advantages of the proposed classification system, compared with the one-step Machine Learning (ML) approach. Communication author: Li Jun, born in 1971, female, Ph.D. candidate, Associate Professor. Nanjing University of Posts and Telecommunications, Nanjing 210003, China.     

DFD-SS: Document Forgery Detection using Spectral – Spatial Features for Hyperspectral Images

In the present era of machines and edge-cutting technologies, still document frauds persist. They are done intuitively by using almost identical inks, that it becomes challenging to detect them—this demands an approach that efficiently investigates the document and leaves it intact. Hyperspectral imaging is one such a type of approach that captures the images from hundreds to thousands of spectral bands and analyzes the images through their spectral and spatial features, which is not possible by conventional imaging. Deep learning is an edge-cutting technology known for solving critical problems in various domains. Utilizing supervised learning imposes constraints on its usage in real scenarios, as the inks used in forgery are not known prior. Therefore, it is beneficial to use unsupervised learning. An unsupervised feature extraction through a Convolutional Autoencoder (CAE) followed by Logistic Regression (LR) for classification is proposed (CAE-LR). Feature extraction is evolved around spectral bands, spatial patches, and spectral-spatial patches. We inspected the impact of spectral, spatial, and spectral-spatial features by mixing inks in equal and unequal proportion using CAE-LR on the UWA writing ink hyperspectral images dataset for blue and black inks. Hyperspectral images are captured at multiple correlated spectral bands, resulting in information redundancy handled by restoring certain principal components. The proposed approach is compared with eight state-of-art approaches used by the researchers. The results depicted that by using the combination of spectral and spatial patches, the classification accuracy enhanced by 4.85% for black inks and 0.13% for blue inks compared to state-of-art results. In the present scenario, the primary area concern is to identify and detect the almost similar inks used in document forgery, are efficiently managed by the proposed approach.     

基于宽度学习的注塑产品质量预测方法

在注塑成型工业中,产品质量自动监测一直是注塑工业智能化发展的核心问题。高品质和大规模的产品质量数据采集成本高昂,导致数据样本量少、不同类别样本数据不平衡,为注塑产品质量预测提出了更高的挑战。为此,该文提出一种基于宽度学习方法的注塑产品质量预测模型,以产品的3维尺寸为预测目标,在普通的宽度学习系统(BLS)中加入最小p范数来改进得到模型p范数宽度学习系统(pN-BLS),解决小样本和不平衡数据的问题,提高模型对离群点的检测性能。在第4届工业大数据竞赛任务2《注塑成型工艺的虚拟量测和调机优化》数据集中,将192个参数特征与预测目标进行相关分析,提取相关性高的基础特征17个,衍生特征4个和调机参数2个作为模型的输入。将16600条数据平均分为训练集和测试集各8300条,与支持向量机 (SVM)、最近邻算法 (KNN)、多层感知机 (MLP)和BLS进行对比实验,实验结果显示pN-BLS具有更快速和更准确的预测效果。在实际缺陷检测应用中,pN-BLS能更准确地预测异常数据,具有更高的鲁棒性。     

无蜂窝大规模MIMO中基于深度强化学习的无人机辅助通信与资源调度

无蜂窝大规模多入多出(MIMO)网络中分布式接入点(AP)同时服务多个用户,可以实现较大区域内虚拟MIMO的大容量传输;而无人机辅助通信能够为该目标区域热点或边缘用户提供覆盖增强。为了降低反馈链路负载,并有效提升无人机辅助通信的频谱利用率,该文研究了基于AP功率分配、无人机服务区选择和接入用户选择的联合调度;首先将AP功率分配和无人机服务区选择问题联合建模为双动作马尔可夫决策过程 (DAMDP),提出了基于Q-learning和卷积神经网络(CNN)的深度强化学习(DRL)算法;然后将用户调度构造为一个0-1优化问题,并分解成子问题来求解。仿真结果表明,该文提出的基于DRL的资源调度方案与现有方案相比,可以有效提升无蜂窝大规模MIMO网络中频谱利用率。     

运营商网络中基于深度强化学习的服务功能链迁移机制

为改善运营商网络提供的移动服务体验,该文研究服务功能链(SFC)的在线迁移问题。首先基于马尔可夫决策过程(MDP)对服务功能链中的多个虚拟网络功能(VNF)在运营商网络中的驻留位置迁移进行模型化分析。通过将强化学习和深度神经网络相结合提出一种基于双深度Q网络(double DQN)的服务功能链迁移机制,该迁移方法能在连续时间下进行服务功能链的在线迁移决策并避免求解过程中的过度估计。实验结果表明,该文所提出的策略相比于固定部署算法和贪心算法在端到端时延和网络系统收益等方面优势明显,有助于运营商改善服务体验和资源的使用效率。