基于无人机影像的电网绝缘子自爆识别

电网因其在电能传输方面的关键性作用,在我国民生项目建设领域一直扮演着至关重要的角色。电网杆塔上的绝缘子一旦发生自爆(也称“缺陷”),绝缘子会自动剥落,输电线路就会产生安全隐患,严重时会降低输电线路的运行寿命,甚至会引发供电中断,发生大范围的停电事故,造成巨大的财产损失。目前,主流的巡检方法为人工巡检,该方法不仅耗时耗力,而且也存在一定主观出错率,已不适用于目前电路巡检的实际情况。本设计采用YOLO V5网络模型,对无人机航拍影像中绝缘子串及绝缘子自爆进行自动识别。首先通过平移、翻转、裁剪等,对航拍绝缘子影像数据集进行数据增广,并对增广后的数据集在LabelImg中进行标注,然后利用YOLO V5网络模型对绝缘子串及绝缘子自爆进行识别,最后采用PyQt5框架在PyCharm中设计了绝缘子自爆识别的系统界面,对模型进行调用,实现了绝缘子串及绝缘子自爆识别。本设计采用从网络上下载、国家电网提供、数据增广所得到的500张无人机航拍影像作为数据集,对所得数据集进行人工标注,再使用YOLO V5网络模型进行训练和测试,结果表明YOLO V5网络模型对绝缘子串具有较高的识别精度,最高识别精度为90.2%,对绝缘子自爆的最高检测精度为80.8%。这说明了YOLO V5网络模型在绝缘子串识别方面有较好的表现,但是由于训练集中绝缘子自爆的样本影像数量有限,所以该网络模型对绝缘子的自爆识别存在一定局限性,本实验能够部分代替人力实现电网绝缘子智能巡检,提高了检测效率。     

图像分类问题的几种深度学习策略

针对图像分类学习不够深入的问题,提出图像分类问题的几种深度学习策略研究。通过分析当前主流的主动深度学习图像、多标签图像和多尺度网络图像三种深度学习方法的工作原理和存在的优势与不足,探讨图像分类问题的优化学习策略。随后采用图像分类问题的几种深度学习策略实验的方式加以对比,实验结果表明,参数共享的深度学习图像分类方法不仅提高了预测速度,而且还能确保模型的准确性。     

基于深度学习的高频雷达射频干扰自动识别与抑制

高频地波雷达的探测性能极易受到射频干扰的影响,当前射频干扰抑制的研究主要是通过人工识别来逐一处理,鲜见实时自动识别与抑制射频干扰的研究。随着深度学习在雷达图像处理方面应用的展开,本文尝试将其引入高频雷达射频干扰抑制中,利用YOLO (You Only Look Once)模型来识别雷达距离多普勒谱图中的射频干扰,继而用高阶奇异值分解(Higher Order Singular Value Decomposition, HOSVD)方法对其进行抑制。仿真和实测数据处理结果表明,此YOLO-HOSVD联合算法实现了对高频雷达射频干扰的自动识别与抑制,单场数据处理时间不超过1.8 s。该方法可以应用于高频地波雷达常规海态观测。     

基于历史梯度平均方差缩减的协同参数更新方法

随机梯度下降算法(SGD)随机使用一个样本估计梯度,造成较大的方差,使机器学习模型收敛减慢且训练不稳定。该文提出一种基于方差缩减的分布式SGD,命名为DisSAGD。该方法采用历史梯度平均方差缩减来更新机器学习模型中的参数,不需要完全梯度计算或额外存储,而是通过使用异步通信协议来共享跨节点的参数。为了解决全局参数分发存在的“更新滞后”问题,该文采用具有加速因子的学习速率和自适应采样策略：一方面当参数偏离最优值时,增大加速因子,加快收敛速度;另一方面,当一个工作节点比其他工作节点快时,为下一次迭代采样更多样本,使工作节点有更多时间来计算局部梯度。实验表明：DisSAGD显著减少了循环迭代的等待时间,加速了算法的收敛,其收敛速度比对照方法更快,在分布式集群中可以获得近似线性的加速。     

Single image shadow detection via uncertainty analysis and GCN-based refinement strategy

Learning-based shadow detection methods have achieved an impressive performance, while these works still struggle on complex scenes, especially ambiguous soft shadows. To tackle this issue, this work proposes an efficient shadow detection network (ESDNet) and then applies uncertainty analysis and graph convolutional networks for detection refinement. Specifically, we first aggregate global information from high-level features and harvest shadow details in low-level features for obtaining an initial prediction. Secondly, we analyze the uncertainty of our ESDNet for an input shadow image and then take its intensity, expectation, and entropy into account to formulate a semi-supervised graph learning problem. Finally, we solve this problem by training a graph convolution network to obtain the refined detection result for every training image. To evaluate our method, we conduct extensive experiments on several benchmark datasets, i.e., SBU, UCF, ISTD, and even on soft shadow scenes. Experimental results demonstrate that our strategy can improve shadow detection performance by suppressing the uncertainties of false positive and false negative regions, achieving state-of-the-art results.     

Meta-transfer-adjustment learning for few-shot learning

Deep neural network models with strong feature extraction capacity are prone to overfitting and fail to adapt quickly to new tasks with few samples. Gradient-based meta-learning approaches can minimize overfitting and adapt to new tasks fast, but they frequently use shallow neural networks with limited feature extraction capacity. We present a simple and effective approach called Meta-Transfer-Adjustment learning (MTA) in this paper, which enables deep neural networks with powerful feature extraction capabilities to be applied to few-shot scenarios while avoiding overfitting and gaining the capacity for quickly adapting to new tasks via training on numerous tasks. Our presented approach is classified into two major parts, the Feature Adjustment (FA) module, and the Task Adjustment (TA) module. The feature adjustment module (FA) helps the model to make better use of the deep network to improve feature extraction, while the task adjustment module (TA) is utilized for further improve the model’s fast response and generalization capabilities. The proposed model delivers good classification results on the benchmark small sample datasets MiniImageNet and Fewshot-CIFAR100, as proved experimentally.     

Semantic guided knowledge graph for large-scale zero-shot learning

Zero-shot learning has received growing attention, which aims to improve generalization to unseen concepts. The key challenge in zero-shot tasks is to precisely model the relationship between seen and unseen classes. Most existing zero-shot learning methods capture inter-class relationships via a shared embedding space, leading to inadequate use of relationships and poor performance. Recently, knowledge graph-based methods have emerged as a new trend of zero-shot learning. These methods use a knowledge graph to accurately model the inter-class relationships. However, the currently dominant method for zero-shot learning directly extracts the fixed connection from off-the-shelf WordNet, which will inherit the inherent noise in WordNet. In this paper, we propose a novel method that adopts class-level semantic information as a guidance to construct a new semantic guided knowledge graph (SG-KG), which can correct the errors in the existing knowledge graph and accurately model the inter-class relationships. Specifically, our method includes two main steps: noise suppression and semantic enhancement. Noise suppression is used to eliminate noise edges in the knowledge graph, and semantic enhancement is used to connect two classes with strong relations. To promote high efficient information propagation among classes, we develop a novel multi-granularity fusion network (MGFN) that integrates discriminative information from multiple GCN branches. Extensive experiments on the large-scale ImageNet-21K dataset and AWA2 dataset demonstrate that our method consistently surpasses existing methods and achieves a new state-of-the-art result.     

Unsupervised learning of multi-task deep variational model

We propose a general deep variational model (reduced version, full version as well as the extension) via a comprehensive fusion approach in this paper. It is able to realize various image tasks in a completely unsupervised way without learning from samples. Technically, it can properly incorporate the CNN based deep image prior (DIP) architecture into the classic variational image processing models. The minimization problem solving strategy is transformed from iteratively minimizing the sub-problem for each variable to automatically minimizing the loss function by learning the generator network parameters. The proposed deep variational (DV) model contributes to the high order image edition and applications such as image restoration, inpainting, decomposition and texture segmentation. Experiments conducted have demonstrated significant advantages of the proposed deep variational model in comparison with several powerful techniques including variational methods and deep learning approaches.     

Learning to align organizational design and data

The relationship between organizational learning and organizational design is explored. In particular, we examine the information processing aspects of organizational learning as they apply to a two-valued decision making task and the relation of such aspects to organizational structure. Our primary contribution is to extend Carley's (1992) model of this process. The original model assumes that all data input into the decision making processes are of equal importance or weight in determining the correct overall organizational decision. The extension described here allows for the more natural situation of non-uniform weights of evidence. Further extensions to the model are also discussed. Such organizational learning performance measures provide an interesting framework for analyzing the recent trend towards flatter organizational structures. This research suggests that flatter structures are not always better, but rather that data environment, ultimate performance goals, and relative need for speed in learning can be used to form a contingency model for choosing organizational structure.     

构建网络学习资源提高民族高校有机化学课程教学质量

High-quality online course materials were reconstructed through feedback from undergraduate students and online data analysis regarding constructed learning materials on the Chao Xing Learning Platform. They were used for studying organic chemistry courses online and offline in university for nationalities. This model, based on student-centered teaching, can effectively transform students from passively accepting knowledge to actively learning and internalizing knowledge. This strategy can also enhance the students' learning initiative and the effects of learning, and hopefully be helpful to the universities in multi-ethnic areas in developing fundamental disciplined construction of organic chemistry and other courses.