基于DeblurGAN的运动模糊图像盲复原算法研究

DeblurGAN方法利用条件生成对抗网络解决了端到端的图像去模糊问题,但存在图像边缘细节恢复不足以及鲁棒性不高的问题,针对此问题,提出一种基于DeblurGAN的运动模糊图像盲复原方法。在生成网络中,采用多尺度卷积核神经网络提取特征,并使用级联空洞卷积扩大神经元的感受野;采用自适配归一化方法代替原来生成器中使用的实例归一化方法。其次,引入了梯度图像L1损失,结合对抗损失和感知损失,将其作为图像去模糊的正则约束,使得生成图像的边缘特征更加清晰。实验结果表明,提出方法复原的图像峰值信噪比数值较DeblurGAN算法高出5.4%,结构相似性指标高出1%;在主观上清晰化效果较好,且消除了网格效应。     

改进生成式对抗网络的图像数据集增强算法

针对现有深度学习中图像数据集缺乏的问题,提出了一种基于深度卷积生成式对抗网络(Deep Convolutional Generative Adversarial Network, DCGAN)的图像数据集增强算法。该算法对DCGAN网络进行改进,首先在不过多增加计算量的前提下改进现有的激活函数,增强生成特征的丰富性与多样性;然后通过引入相对判别器有效缓解模式坍塌现象,从而提升模型稳定性;最后在现有生成器结构中引入残差块,获得相对高分辨率的生成图像。实验结果表明,将所提方法应用在MNIST、SAR和医学血细胞数据集上,图像数据增强效果与未改进的DCGAN网络相比显著提升。     

RFID-based 3D human pose tracking: A subject generalization approach

Three-dimensional (3D) human pose tracking has recently attracted more and more attention in the computer vision field. Real-time pose tracking is highly useful in various domains such as video surveillance, somatosensory games, and human-computer interaction. However, vision-based pose tracking techniques usually raise privacy concerns, making human pose tracking without vision data usage an important problem. Thus, we propose using Radio Frequency Identification (RFID) as a pose tracking technique via a low-cost wearable sensing device. Although our prior work illustrated how deep learning could transfer RFID data into real-time human poses, generalization for different subjects remains challenging. This paper proposes a subject-adaptive technique to address this generalization problem. In the proposed system, termed Cycle-Pose, we leverage a cross-skeleton learning structure to improve the adaptability of the deep learning model to different human skeletons. Moreover, our novel cycle kinematic network is proposed for unpaired RFID and labeled pose data from different subjects. The Cycle-Pose system is implemented and evaluated by comparing its prototype with a traditional RFID pose tracking system. The experimental results demonstrate that Cycle-Pose can achieve lower estimation error and better subject generalization than the traditional system.     

基于自注意力对抗的深度子空间聚类

子空间聚类(Subspace clustering)是一种当前较为流行的基于谱聚类的高维数据聚类框架.近年来,由于深度神经网络能够有效地挖掘出数据深层特征,其研究倍受各国学者的关注.深度子空间聚类旨在通过深度网络学习原始数据的低维特征表示,计算出数据集的相似度矩阵,然后利用谱聚类获得数据的最终聚类结果.然而,现实数据存在维度过高、数据结构复杂等问题,如何获得更鲁棒的数据表示,改善聚类性能,仍是一个挑战.因此,本文提出基于自注意力对抗的深度子空间聚类算法(SAADSC).利用自注意力对抗网络在自动编码器的特征学习中施加一个先验分布约束,引导所学习的特征表示更具有鲁棒性,从而提高聚类精度.通过在多个数据集上的实验,结果表明本文算法在精确率(ACC)、标准互信息(NMI)等指标上都优于目前最好的方法.     

基于多对多生成对抗网络的非对称跨域迁移行人再识别

无监督跨域迁移学习是行人再识别中一个非常重要的任务. 给定一个有标注的源域和一个没有标注的目标域, 无监督跨域迁移的关键点在于尽可能地把源域的知识迁移到目标域. 然而, 目前的跨域迁移方法忽略了域内各视角分布的差异性, 导致迁移效果不好. 针对这个缺陷, 本文提出了一个基于多视角的非对称跨域迁移学习的新问题. 为了实现这种非对称跨域迁移, 提出了一种基于多对多生成对抗网络(Many-to-many generative adversarial network, M2M-GAN)的迁移方法. 该方法嵌入了指定的源域视角标记和目标域视角标记作为引导信息, 并增加了视角分类器用于鉴别不同的视角分布, 从而使模型能自动针对不同的源域视角和目标域视角组合采取不同的迁移方式. 在行人再识别基准数据集Market1501、DukeMTMC-reID和MSMT17上, 实验验证了本文的方法能有效提升迁移效果, 达到更高的无监督跨域行人再识别准确率.     

Multimodal face aging framework via learning disentangled representation

Existing face aging (FA) approaches usually concentrate on a universal aging pattern, and produce restricted aging faces from one-to-one mapping. However, the diversity of living environments impact individuals differently in their oldness. To simulate various aging effects, we propose a multimodal FA framework based on face disentanglement technique of age-specific and age-irrelevant information. A Variational Autoencoder (VAE)-based encoder is designed to represent the distribution of the age-specific attributes. To capture the age-irrelevant features, a cycle-consistency loss of unpaired faces is utilized among various age spans. The extensive experimental results demonstrate that the sampled age-specific codes along with an age-irrelevant feature make the multimodal FA diverse and realistic.     

Generating Time-Series Data Using Generative Adversarial Networks for Mobility Demand Prediction

The increasing penetration rate of electric kickboard vehicles has been popularized and promoted primarily because of its clean and efficient features. Electric kickboards are gradually growing in popularity in tourist and education-centric localities. In the upcoming arrival of electric kickboard vehicles, deploying a customer rental service is essential. Due to its free-floating nature, the shared electric kickboard is a common and practical means of transportation. Relocation plans for shared electric kickboards are required to increase the quality of service, and forecasting demand for their use in a specific region is crucial. Predicting demand accurately with small data is troublesome. Extensive data is necessary for training machine learning algorithms for effective prediction. Data generation is a method for expanding the amount of data that will be further accessible for training. In this work, we proposed a model that takes time-series customers’ electric kickboard demand data as input, pre-processes it, and generates synthetic data according to the original data distribution using generative adversarial networks (GAN). The electric kickboard mobility demand prediction error was reduced when we combined synthetic data with the original data. We proposed Tabular-GAN-Modified-WGAN-GP for generating synthetic data for better prediction results. We modified The Wasserstein GAN-gradient penalty (GP) with the RMSprop optimizer and then employed Spectral Normalization (SN) to improve training stability and faster convergence. Finally, we applied a regression-based blending ensemble technique that can help us to improve performance of demand prediction. We used various evaluation criteria and visual representations to compare our proposed model’s performance. Synthetic data generated by our suggested GAN model is also evaluated. The TGAN-Modified-WGAN-GP model mitigates the overfitting and mode collapse problem, and it also converges faster than previous GAN models for synthetic data creation. The presented model’s performance is compared to existing ensemble and baseline models. The experimental findings imply that combining synthetic and actual data can significantly reduce prediction error rates in the mean absolute percentage error (MAPE) of 4.476 and increase prediction accuracy.     

Design by framework completion

An object-oriented framework in essence defines an architecture for a family of applications or subsystems in a given domain. Every application in the family obeys these architectural restrictions. Such frameworks are typically delivered as collections of inter-dependent abstract classes, together with their concrete subclasses. The abstract classes and their interdependencies implicitly realize the architecture. Developing a new application reusing classes of a framework requires a thorough understanding of the framework architecture.We introduce an approach called Design by Framework Completion, in which an exemplar (an executable visual model for a minimal instantiation of the architecture) is used for documenting frameworks. We propose exploration of exemplars as a means for learning the architecture, following which new applications can be built by replacing selected pieces of the exemplar. For the piece to be replaced, the inheritance lattice around its class provides the space of alternatives, one of these classes may be suitably adapted (say, by sub-classing) to create the new replacement.Design by Framework Completion proposes a paradigm shift when designing in presence of reusable components: It enables a much simpler top-down approach for creating applications, as opposed to the prevalent search for components and assemble them bottom-up strategy. We believe that this paradigm shift is essential because components can only be fitted together if they all obey the same architectural rules that govern the framework.     

优化程序设计课堂教学的探讨

针对程序设计课堂教学中的主要问题,笔者以充分发挥学生的主观能动性、培养学生的创造性为主要目标,以学生为中心,教师为主导,提出优化程序设计课堂教学的一系列方法,对优化程序设计课程教学具有重要的作用。本文对整个优化过程进行了详细阐述。     

数据结构中绪论部分的教学方法研究与实践

针对数据结构课程中传统教学方法存在的问题,以提高学生的学习兴趣和增强自信心为出发点,从绪论的内容选择、讲授方法、应用举例等方面进行了一系列的探讨,提出一种兴趣-实例驱动的教学方法。实践表明,该教学方式有效地调动了学生的学习主动性,为数据结构的学习开创了一个良好开端。