Flagging cartel participants with deep learning based on convolutional neural networks

Adding to the literature on the data-driven detection of bid-rigging cartels, we propose a novel approach based on deep learning (a subfield of artificial intelligence) that flags cartel participants based on their pairwise bidding interactions with other firms. More concisely, we combine a so-called convolutional neural network for image recognition with graphs that in a pairwise manner plot the normalized bids of some reference firm against the normalized bids of any other firms participating in the same tenders as the reference firm. Based on Japanese and Swiss procurement data, we construct such graphs for both collusive and competitive episodes (i.e when a bid-rigging cartel is or is not active) and we use a subset of graphs to train the neural network such that it learns distinguishing collusive from competitive bidding patterns. With the remaining graphs, we test the neural network’s out-of-sample performance in correctly classifying collusive and competitive bidding interactions. We obtain a very decent average accuracy of around 95% or slightly higher when either applying the method within Japanese, Swiss, or mixed data (in which Swiss and Japanese graphs are pooled). When using data from one country for training to test the trained model’s performance in the other country (i.e. transnationally), predictive performance decreases (likely due to institutional differences in procurement procedures across countries), but often remains satisfactorily high. All in all, the generally quite high accuracy of the convolutional neural network despite being trained in a rather small sample of a few 100 graphs points to a large potential of deep learning approaches for flagging and fighting bid-rigging cartels.     

Sensitivity results considering rescheduling by AND/OR graphs

Closely connected with scheduling are also rescheduling considerations which can obviously be addressed by sensitivity analysis. We discuss this question by using AND/OR graphs and show how sensitivity analysis can be performed in this context. We show that AND/OR graphs can be used to model job shop problems and alternative process scheduling, hence sensitivity analysis can be performed in these contexts as well.     

Monitoring R&D project costs against pre-specified targets

The monitoring and interpretation of costs incurred during a project against pre-specified targets is the most common task which R&D managers at all levels have to undertake. Graphical interpretation of such cost and target figures can provide useful feedback as to financial performance.
Spreadsheets provide an ideal environment for the manipulation of figures and production of graphs and, although not widely used, the programming facilities incorporated in such packages also allow for menu driven systems to be developed which can be operated without a knowledge of the spreadsheet's command structure thus greatly widening their area of application.
This article describes some of the elements of such a system and details its ability to automatically produce four specified graphs taken from accountancy, statistical process control and forecasting which have been shown to offer a valuable insight into cost control generally and, it is proposed, within R&D projects in particular.     

N_2—局部连通图的圈扩张与点泛圈

文[2]证明了每个连通,N_2——局部连通,无爪、又δ≥2,且不含同构于G_1或 G_2的导出子图的图具有圈扩张性。本文在同样前题下,在含导出于图 G_1、G_2时得具有[2]中结论的一类图。     

Empowering project scope decisions: introducing R&D content graphs

There have been several studies confirming the strategic and operational importance of choosing a proper project scope. There is a lack of empirically tested managerial tools to assist in this matter, and studies with longitudinally analysed R&D projects with respect to content are scarce. This article discusses the management of advanced engineering in the automotive industry. The study presents a structured approach where individual project leaders together with top management could prepare project material, discuss, visualize, and evaluate the content around a common tool. This tool, i.e. the R&D content graphs, facilitates project content selection, both during the start‐up and during the execution phase of the project. The studied project has been studied for nearly two years using an abductive case‐based research design. Since the project has been longitudinally studied, it has also been possible to test the graphs' predicting capabilities. The graphs are based on a visualization of the project content along five identified dimensions. The results reveal that the tools bring structure to project scope discussions, have some predictive value, and can also function as a vertical as well as a horizontal communication tool.     

椭圆上特殊点的准确作图原理简析

画法几何中用八点法作椭圆时,一般是根据圆的外切四边形画出平行四边形,从而确定两对特殊的共轭直径,所画椭圆的8个特殊点中总有4个点可以比较直观地确定,而其余4个点只能近似地在相应的共轭直径上求出。本文应用仿射对应等方法,阐述了准确定出椭圆周必定经过的特殊点的位置,并提出其作图的理论根据,同时提供了多种准确快速地确定这些特殊点的方法。     

3—连通、3—正则图中的圈

本文介绍了3—连通、3—正则图中包含给定点集的大子集的圈。     

Measurement of bias in project durations

A note by Milliken (1973) suggests the usage of programme trend charts will help managers estimate the total duration of the project slippage. Whilst his sample graphs are informative they are of little use in evaluating two aspects of the problem: (a) The uncertainty inherent in the project–known to the Shop Floor, but unknown to the management. (b) The expected completion date–until sufficient data has been collected to yield reliable regression extrapolations. It is postulated that the assessment of suitably encoded subjective probability distributions be used to satisfy points (a) and (b) above. These distributions, when evaluated by a scoring rule, will also give the management, and the assessor, a true statement of the quality of the assessments. Furthermore, the underlying theory of subjective probability scoring rules suggests that assessors should become ‘better’ when being scored than when given no incentive, other than the usual behavioural guidance of most hierarchic organizations.     

最优伍配及其应用

本文给出了图的最优伍配的基本概念并探讨了它的性质、算法及应用。     

长城机在cdBASEⅢ状态下直接作图

本文介绍了利用 GWBIOS3.00提供的系统图形驱动文件 GRD.SYS,在cdBASEⅢ状态下直接作图的方法及常用命令格式。     

Role of network structure and network effects in diffusion of innovations

Why does diffusion of innovation sometimes propagate throughout the whole population, and why at other times does it halt in its interim process? The current paper provides a potential answer to this question by developing a simple computational model of social networks. The proposed computational approach incorporating small-world graphs enables the authors to find that diffusion of innovation is more likely to fail in a random network than in a highly clustered network of consumers. A marketing implication is that the choice of initial target groups and their network structures matter in influencing whether an innovation makes full or partial penetration, in markets where network effects plays a role.     

一种任意无向图的R边连通扩充算法

本文提出了一个最小地扩充任意无向连通图为 R 边连通图的有效算法RMA。该算法采用了“先满足必要条件,再满足充要条件”的指导思想。对于一个任意无向连通图,首先将图中各点扩充到它所要求的最小度,然后检查它是否满足充要条件,如果不满足,则将该图分解,再根据最优程则,增加扩充边。将图合并,最后进行可行删除,解雇增广点,得到一个最小 R 边连通图。     

Local network externalities and market segmentation

This paper models interaction between groups of agents by means of a graph where each node represents a group of agents and an arc represents bilateral interaction. It departs from the standard Katz–Shapiro framework by assuming that network benefits are restricted only amongst groups of linked agents. It shows that even if rival firms engage in Bertrand competition, this form of network externalities permits strong market segmentation in which firms divide up the market and earn positive profits. The analysis also shows that some graphs or network structures do not permit such segmentation, while for others, there are easy to interpret conditions under which market segmentation obtains in equilibrium.     

无线数据传输技术在石油行业中的应用及展望

概括总结了在石油行业中使用的无线数据传输技术,主要可分为常规无线数据传输和井下无线随钻传输。提出了两点展望:利用短距离无线采集与远程无线数传相结合,可实现钻采现场参数实时监测;研发声波无线数据传输新型钻杆接头,能克服能量衰减问题,提高声波传输距离。     

Analysing clusters of skills in R&D — core competencies, metaphors, visualization, and the role of IT

Though it is now universally accepted that companies should try to align their R&D activities with their business objectives, achieving this alignment is notoriously difficult in practice. The rise of the core competence framework has been very helpful in creating, and legitimizing, a language in which issues of technical competence and R&D strengths can be followed through to their consequences for competitive advantage. Companies are starting to express their R&D priorities explicitly in terms of core competencies.
Without effective IT support, core competence concepts are often applied arbitrarily. This has led to accusations that core competence theory can become yet another battlefield upon which companies play out their internal political battles. Computer-based techniques can help counteract this danger by enabling large volumes of relatively objective data to be collected, then making it possible to analyse and draw out patterns from this data, and finally enabling the data to be represented effectively.
It is in this last area of data representation that information technology is now of particular benefit. In order to make the core competence approach sufficiently robust as a basis for decision making, it is necessary to collect and process large volumes of data. However, this data is normally difficult to represent in such a way that managers can assimilate it. In our recent experience, we have come to realize the particular importance of effective representations and metaphors, and have started to shift our own emphasis towards these areas in addition to analysis per se .
The paper shows how core competence approaches can support R&D management decision making by exploring the roles of data collection, analysis and representation. Information technology is an integral part of these approaches, and we draw out some generalized lessons for the successful use of IT in decision support.     

区块链辅助航天数据分发的可行性分析及方案设计

为了快速准确、安全可信地分发海量航天数据,基于区块链技术的可溯源和不可篡改等特性,提出了一种新的航天数据分发方案.首先,在总结传统航天数据分发服务不足的基础上,分析了区块链技术在航天数据分发领域的优势以及应用过程中可能遇到的重点问题,并给出了可能的解决方法;其次,从技术架构、节点部署、模块划分3个角度设计了基于区块链的...     

Delphi的决策组件在三层结构中的应用

Delphi中的DecisionCube控件能够以多维结构来组织数据,并且为DecisionGrid等显示组件提供多维数据,但是在三层体系结构中,它不能自动建立数据集中的字段与维度、度量的映射。本文通过定义一个获取分析主题的接口函数,以及一个字段与维的映射函数,解决了这个问题,应用此技术可以低成本、快速地开发三层结构的决策支持系统。     

煤层气开发投资项目不确定性综合评估方法探讨

在评价煤层气开发项目经济效益的过程中,常采用敏感性分析来评估项目的不确定性,但敏感性分析无法综合评估项目的不确定性。本文开展此项研究以期建立简便、可操作的综合不确定性分析方法。为此,利用模拟方案生成数据,对数据进行统计分析,寻找不确定因素与效益指标之间的函数关系。研究发现,当不确定因素在一定范围内波动时,多因素变动对经济效益的综合影响可以近似为单因素影响的线性叠加,即可利用单因素敏感曲线的拟合斜率来构建综合评价公式。线性叠加会产生计算误差,误差随因素变动范围的扩大而增加,不确定因素在±25% 范围内波动时,90% 以上的数据落在误差区间为±10% 的范围之内。文中建立的方法可以对方案进行综合不确定性评价,优选方案。     

Indicators of discontinuous technological change: an exploratory study of two discontinuities in the machine tool industry

The objective of this paper is to explore the extent to which a set of indicators of technological and industrial change can act as warning signals for technical change. A particular kind of technical change can give a new substitute such price/performance attributes that it is taken into the reach of mass market segments. Two processes of discontinuous technical change in the machine tool industry are analysed using patent data, bibliometrics, data on new entrants, relative price changes and diffusion data. In the first case (the transition from conventional to CNC machine tools) relative prices and new entrants were the first indicators to react whilst patents and bibliometrics increased in activity in parallel with the large scale diffusion of CNC machine tools. In the second case (the transition from CNC machine tools to flexible manufacturing systems) new entrants and publishing preceded the large scale diffusion by some years. The different patterns between the two cases and between these and what can be found in the literature, suggest that more work needs to be done to understand the conditions under which changes in each of the indicators can be used as a warning signal. The paper is concluded by a brief discussion which may form the basis for some further work in that direction.     

可靠性数据处理方法的探讨

数据的收集和分析处理工作是一切可靠性工作的基础,不进行数据的收集和分析处理,就不可能定量地表示可靠性。可靠性数据可以分为规则寿命数据和不规则寿命数据。通常可靠性数据多为不规则寿命数据。通过对不规则寿命数据进行分段处理,应用可靠性理论,推导出这种常见的且非常重要的不规则寿命数据的处理方法,使得对不规则厅靠性寿命数据处理方法的理解更加深入,公式的意义更加明确,公式的应用更加方便。