Rational Return——Significance of and Reflection on Freehand Sketching

The overuse of and heavy dependency on a technology will result in the rigidity of thinking, which is true of computer-aided drawing. Due to the similarities in computer graphics, freehand sketching has attracted the public attention again. In this paper, the necessity and importance of freehand sketching were reaffirmed, and suggestions for teaching reform in terms of learning interest, methods, quantity, and achievements were put forward, so as to help students to comprehensively improve their skills and accomplishment in their learning and practice.

Wu Zuoren's Westward Sketching Journey and Han-Tibetan Cultural Exchange in the 1940s

The nationalization,modernization,and Chinafication of artistic creation are the motivation for pushing forward the further development of art and advocating Chinese culture.During the Comprehensive War of resistance Against Japanese Aggression,the painter Wu Zuoren answered the art community's call of the times and went twice westward to Xizang for sketching;these trips enriched his artistic creation with Tibetan culture and inspired him to some national paintings;he not only changed and diversified his painting style,but also held a series of exhibitions and communication events,which deepened the understanding and love for Tibetan culture among people from home and abroad.Many of his Tibetan paintings,as they spread across China,have gradually become symbols of Chinese culture shared by multiple ethnic groups;these Tibetan paintings not only demonstrate the united image and general status of the Chinese nation but also serve as vivid cases for the study of forging a strong sense of community for the Chinese nation.

关于Kaczmarz的一类加速免伪逆贪婪块方法

块贪婪Kaczmarz方法在解决大规模一致线性系统方面取得了成功应用。然而在每次迭代步骤中,GBK方法都涉及伪逆计算,这不仅复杂化了计算并减慢了收敛速度,且不适合分布式实现。在本文中基于Sketching技术提出了两种免伪逆计算的GBK方法,分别是杠杆得分抽样免伪逆GBK方法和稀疏随机投影免伪逆GBK方法,其算法效率更加高效,收敛速度可以达到指数收敛。为了进一步加快收敛速度,我们还提出了CountSketch免伪逆重力球GBK方法、杠杆得分抽样免伪逆重力球GBK方法和稀疏随机投影免伪逆重力球GBK方法。为了验证新方法的有效性,我们进行了一些数值示例。结果表明,这些新方法在解决大规模一致线性系统方面具有很高的效率和准确性。

基于近似测量算法Sketch的高精度区块链异常检测机制

区块链存在网络动态性强和其管理困难等问题,使得区块链普遍存在DDoS攻击和账户接管等异常现象.现有区块链异常检测方法多从所有区块链账户中提取历史交易信息和交易频率等特征加以分析以甄别异常.然而,随着区块链数据规模的不断扩大,现有方法在提取特征时面临内存消耗高、检测精度低的挑战.为此,提出了一种高检测精度、低内存开销的区块链异常检测机制,该机制采用近似测量算法将区块链异常检测转化为异常交易账户检测,包括区块内异常账户和跨区块异常账户.对于区块内异常账户,即仅存在于单个区块内的异常账户,使用Sketch算法进行账户识别,精度高.而对于存在于多个区块且难以通过单个区块信息检测到的跨区块异常账户,则通过聚合和分析多区块信息进行账户的准确检测.使用包含88847个区块的真实区块链数据评估上述机制.实验结果表明,与现有代表性方法对比,所提出的机制将区块链异常检测的召回率最高提升了6.3倍,F1分数最高提升了4.4倍.因此,提出的高精度区块链异常检测机制对于规范区块链交易行为、维护系统安全性具有意义.

分布式网络中连续时间周期的全局top-K频繁流测量

在分布式网络中,测量top-K频繁流对资源分配、安全监控等应用至关重要。现有的top-K频繁流测量工作存在不适用于测量分布式网络流量或只考虑单时间周期等局限。为此,提出了分布式网络中连续时间周期的全局top-K频繁流测量方案,在分布节点中布置了紧凑的概率数据结构来记录网络流信息,每个时间周期结束后分布节点向中心节点发送必要信息,中心节点汇聚得到从测量开始至当前时间周期的全局top-K频繁流。考虑到每条流可能出现在一个或多个测量节点,使用了不同的方法来减少传输开销。对于每条流只会出现在单一节点的情况,采用传输分段最小值的方法来获得阈值,实验结果表明这种方法减少了全量传输超过50%的传输开销。对于每条流会出现在多个节点的情况,提出了多阶段无误差处理方法和单阶段快速处理方法,分别应对不能容忍误差的场景和实际高速网络流量,相比每个时间周期都使用已有单周期方法,传输开销的实验表现降低了两个数量级。最后还提出了一种利用历史平均增值信息降低通信延迟的方法,实验结果表明该方法有效降低了限制信息的平均相对误差。

基于Count Sketch的预处理贪婪Kaczmarz方法

在贪婪Kaczmarz方法中,通过对系数矩阵进行正交三角分解引入右预处理子能够提高贪婪Kaczmarz方法的收敛速率。但在系数矩阵的行数远大于列数的情况下,正交三角分解的成本过高。为降低预处理的成本,通过引入Count Sketch变换,提出了基于Count Sketch的预处理贪婪Kaczmarz方法,并对新方法进行了收敛性分析。理论分析说明了新方法在系数矩阵条件数较大时比已有方法具有更好的收敛速率。数值实验验证了新方法的有效性。

DualFace:Two-stage drawing guidance for freehand portrait sketching

Special skills are required in portrait painting, such as imagining geometric structures and facial detail for final portrait designs. This makes it a difficult task for users, especially novices without prior artistic training, to draw freehand portraits with high-quality details. In this paper, we propose dualFace, a portrait drawing interface to assist users with different levels of drawing skills to complete recognizable and authentic face sketches. Inspired by traditional artist workflows for portrait drawing, dualFace gives two-stages of drawing assistance to provide global and local visual guidance. The former helps users draw contour lines for portraits (i.e., geometric structure), and the latter helps users draw details of facial parts, which conform to the user-drawn contour lines. In the global guidance stage, the user draws several contour lines, and dualFace then searches for several relevant images from an internal database and displays the suggested face contour lines on the background of the canvas. In the local guidance stage, we synthesize detailed portrait images with a deep generative model from user-drawn contour lines, and then use the synthesized results as detailed drawing guidance. We conducted a user study to verify the effectiveness of dualFace, which confirms that dualFace significantly helps users to produce a detailed portrait sketch.

PeakSketch:检测网络流中的top-k流的无偏和通用草图

通过对现有Sketch结构的研究,提出一种新的Sketch结构:PeakSketch,本文将其应用于三种任务:检测top-k频繁流,检测top-k重变化流,检测top-k持久流,从理论上证明了PeakSketch可以提供无偏估计,并且给出了算法的误差界。实验结果表明,PeakSketch的各项性能优秀,在检测top-k频繁流任务中,PeakSketch的吞吐量显著提升,特别是在分配内存小于200 kB以下时,吞吐量最高提升可以达到50%,准确率最高提升一倍,PeakSketch也展现突出的性能。

Sketch软件在UI设计中的应用研究

随着移动智能设备的普及和各类App的迭代发展,用户对产品的用户界面(User Interface,UI)期望越来越高。为了追求良好的设计效果,越来越多的计算机图像处理软件被运用到UI设计行业。作为一款专门为数字产品的设计而生的软件,目前,Sketch已经基本替代Photoshop,被广泛地应用于UI设计领域。基于此,文章从Sketch软件在UI设计中的实际应用角度出发,介绍了UI设计与Sketch软件的基本内容,分析了Sketch软件在设计应用中的主要优势和一般步骤。

基于多层Sketch的SDN网络流量测量技术研究

针对大流检测、突变流检测和基数估计等的网络流量测量对保障网络安全具有重要意义.但当前相关研究存在实时性不足、测量精度不高等问题.针对上述问题,设计了一种基于多层Sketch(multiple layer sketch, ML Sketch)的网络流量测量模型.首先,该模型采用自主设计的ML Sketch结构,使用分类存储结构提高了流量测量的精度.其次,在SDN(software defined network)环境下利用流量实时回放技术,模拟了流量的动态发生场景.最后,在SDN控制平面实现了对大流、突变流和基数估计类流量的实时动态检测.在UNSW-NB15上的实验结果表明,与传统Sketch结构相比,所设计的ML Sketch结构在F1_Score指标上最高提高4.81%,相关误差最高降低81.12%,验证了该模型的有效性.

IntervalSketch:面向数据流的间隔项近似统计方法

流式数据库在数据库中的占比逐渐增加,在流式数据库的数据流中提取所需信息是一项重要任务。文中研究了数据流的间隔项,并将其应用到了网络场景中。其中间隔项指在数据流中以固定时间间隔到达的元素对,这是第一项在数据流中定义和统计间隔项的工作。为了高效统计间隔项的top-K,提出了IntervalSketch。IntervalSketch首先基于模拟退火对数据流分块以加快统计速度,其次利用Sketch进行间隔项的存储,最后通过特征分组存储策略降低Sketch存储间隔项的空间开销,提升了统计间隔项的精度。IntervalSketch在两个真实数据集上进行了大量对比实验,实验结果表明,在同样内存的情况下,IntervalSketch明显优于基线方案,其中处理时间为基线方案的1/3~1/2,平均绝对误差、平均相对误差约为基线方案的1/3。

RBFRadar:基于可编程数据平面检测价值突发流

在各种网络流量中,突发是一种常见且重要的流量模式。突发会增大网络时延并影响应用性能,因此对突发流的检测、分析和缓解对于提升网络性能和鲁棒性是有意义的。然而,当前基于逐次突发的检测方案存在显著的带宽开销和高用户负担问题。文中通过观察并分析多个场景下的突发流量特征,提出了价值突发流(Remarkable Burst Flow,RBF)检测,在降低带宽开销的同时,减少了传统突发检测中的密集手工劳动和专家经验要求,减轻了网络管理者的负担。RBFRadar是基于Sketch数据结构的框架,支持可编程数据平面上的RBF检测,在一段时间内观察流级别的突发性。该框架仅产生有限的内存占用和低时间复杂性,其原型可在PISA架构上实现。实验结果表明,在检测RBF的准确性方面,RBFRadar的F1分数是现有方案的5.6~23.4倍;在带宽开销方面,与基于逐次突发的检测方案相比,RBFRadar可降低84.62%~98.84%的带宽开销。

自动生成词典:我们接近了吗?

就在十余年前,多篇论文评介了计算技术在词典编纂中的应用研发情况(尤见于Rundell&Kilgarriff 2011)。这些论文展示了词典编纂过程如何在某种程度上实现了自动化,并对完全自动化道路上可能取得的更多进展进行了预测。文章首先简述2011年的前沿技术,然后梳理迄今所取得的进展。对早期论文所做出的预期进行了回顾。在被称作“后期编辑词典编纂”的模式中,人类词典编纂者的角色是后期编辑,即对自动生成并转入词典编写系统的词典初稿进行评估优化。但这些已取得的进展目前皆受到怀疑,因为ChatGPT等大型语言模型似乎有望绕过眼下所使用的技术。文章通过ChatGPT生成词典文本的诸多实验,探讨了这些人工智能工具取代目前词典编纂前沿技术的可能性。

Short Video Craze at a Nursing Home

Short videos shot at a nursing home in the northern Chinese city of Tianjin are aimed at enriching the inner world of its elderly residents and maintaining their optimistic life view.Agroup of senior citizens face the camera and deliver lines from a scripted sketch ending with the homespun homily,“There are some parents who always compare their children with other people’s children,but they never think to compare themselves with other parents.”The occasional bloopers of these senior players make fellow actors fall about laughing.But rather than reproving them,the director joins in the merriment.This is not a stage rehearsal or movie set scenario,but a video shooting taking place at a nursing home in the northern Chinese city of Tianjin.Since last April,Jingya Nursing Home,in an effort to enrich the lives of its elderly residents,has taken the new approach of shooting short videos.These have since attracted nearly a million followers and scored over six million likes on Douyin,the Chinese version of TikTok.

A Procedure for Trisecting an Acute Angle (Method 2)

This paper presents an alternate graphical procedure (Method 2), to that presented in earlier publications entitled, “A Procedure for Trisecting an Acute Angle” and “A Key to Solving the Angle Trisection Problem”. The procedure, when applied to the 30˚ and 60˚ angles that have been “proven” to be nottrisectable and the 45˚ benchmark angle that is known to be trisectable, in each case produced a construction having an identical angular relationship with Archimedes’ Construction, as in Section 2 on THEORY of this paper, where the required trisection angle was found to be one-third of its respective angle (i.e. DE’MA = 1/3 DE’CG). For example, the trisection angle for the 30˚, 45˚ and 60˚ angles were 10.00000˚, 15.00000˚, and 20.00000˚, respectively, and Section 5 on PROOF in this paper. Therefore, based on this identical angular relationship and the numerical results (i.e. to five decimal places), which represent the highest degree of accuracy and precision attainable by The Geometer’s Sketch Pad software, one can only conclude that not only the geometric requirements for arriving at an exact trisection of the 30˚ and 60˚ angle (which have been “proven” to be not-trisectable) have been met, but also, the construction is valid for any arbitrary acute angle, despite theoretical proofs to the contrary by Wantzel, Dudley, and others.

Efficient policy evaluation by matrix sketching

In the reinforcement learning,policy evaluation aims to predict long-term values of a state under a certain policy.Since high-dimensional representations become more and more common in the reinforcement learning,how to reduce the computational cost becomes a significant problem to the policy evaluation.Many recent works focus on adopting matrix sketching methods to accelerate least-square temporal difference(TD)algorithms and quasi-Newton temporal difference algorithms.Among these sketching methods,the truncated incremental SVD shows better performance because it is stable and efficient.However,the convergence properties of the incremental SVD is still open.In this paper,we first show that the conventional incremental SVD algorithms could have enormous approximation errors in the worst case.Then we propose a variant of incremental SVD with better theoretical guarantees by shrinking the singular values periodically.Moreover,we employ our improved incremental SVD to accelerate least-square TD and quasi-Newton TD algorithms.The experimental results verify the correctness and effectiveness of our methods.

绘画知识

1. Sketch素描.Sketching is a monochrome drawing that uses lines to depict the light and shade of objects with materials such as charcoal, pencils, or pens. Monochrome watercolor and monochrome oil paintings can also be considered sketches;traditional Chinese fineline paintings and ink wash paintings can also be called sketches. Usually, sketching refers to pencil and charcoal drawings.

基于高职应用型人才培养的教学研究——以SketchUp草图大师课程为例

该文以SketchUp草图大师课程为例,阐述了教学改革中高职应用型人才培养的必要性和可行性,分析了现阶段高职院校课程教学中存在的问题,并针对存在的问题提出了具体的解决措施。在此基础上,从学生能力培养、课堂教学质量提升以及教学团队建设等方面探索了课程改革效果,并总结课程改革的实施过程。

室内效果图设计中光影的应用

随着人工智能的快速发展,人们对室内效果图的品质要求也不断提高,光影的应用是体现效果图真实场景的关键,不管从3ds max还是从SketchUP等软件中的光影应用成为我们制作高质量室内效果图的基础。本文主要从高职室内效果图制作课程中针对光影的应用进行浅析,用更有效更快捷的方法设计室内效果图的光影。

基于激光扫描的建筑物三维建模与设计研究

城市数字化建设飞速发展,临街老旧建筑的三维真实信息是建筑物改建、扩建工程设计和施工的依据,因此基于实测数据建立老旧建筑真实三维模型对于优化设计、总体规划、造价计算等具有重要意义。传统测量手段只能获取目标物特征点数据,基于少量特征点建模难度较大。随着科技发展,三维激光扫描技术可以实现自动化、高效率获取目标建筑海量点云数据的功能,以工程案例形式将点云数据采集、应用Sketch Up和Mars软件建模及优化设计的整体技术流程进行详细介绍,研究结果可为老旧街区保护和更新改造提供新思路,具有一定的借鉴和推广价值。