An empirical evaluation of the comprehensibility of decision table, tree and rule based predictive models

An important objective of data mining is the development of predictive models. Based on a number of observations, a model is constructed that allows the analysts to provide classifications or predictions for new observations. Currently, most research focuses on improving the accuracy or precision of these models and comparatively little research has been undertaken to increase their comprehensibility to the analyst or end-user. This is mainly due to the subjective nature of ‘comprehensibility’, which depends on many factors outside the model, such as the user's experience and his/her prior knowledge. Despite this influence of the observer, some representation formats are generally considered to be more easily interpretable than others. In this paper, an empirical study is presented which investigates the suitability of a number of alternative representation formats for classification when interpretability is a key requirement. The formats under consideration are decision tables, (binary) decision trees, propositional rules, and oblique rules. An end-user experiment was designed to test the accuracy, response time, and answer confidence for a set of problem-solving tasks involving the former representations. Analysis of the results reveals that decision tables perform significantly better on all three criteria, while post-test voting also reveals a clear preference of users for decision tables in terms of ease of use.     

3D model alignment based on minimum projection area

3D model alignment is an important step for applications such as 3D model retrieval and 3D model recognition. In this paper, we propose a novel Minimum Projection Area-based (MPA) alignment method for pose normalization. Our method finds three principal axes to align a model: the first principal axis gives the minimum projection area when we perform an orthographic projection of the model in the direction parallel to this axis, the second axis is perpendicular to the first axis and gives the minimum projection area, and the third axis is the cross product of the first two axes. We devise an optimization method based on Particle Swarm Optimization to efficiently find the axis with minimum projection area. For application in retrieval, we further perform axis ordering and orientation in order to align similar models in similar poses. We have tested MPA on several standard databases which include rigid/non-rigid and open/watertight models. Experimental results demonstrate that MPA has a good performance in finding alignment axes which are parallel to the ideal canonical coordinate frame of models and aligning similar models in similar poses under different conditions such as model variations, noise, and initial poses. In addition, it achieves a better 3D model retrieval performance than several commonly used approaches such as CPCA, NPCA, and PCA.     

On partitioning a graph into two connected subgraphs

Suppose a graph G is given with two vertex-disjoint sets of vertices Z₁ and Z₂. Can we partition the remaining vertices of G such that we obtain two connected vertex-disjoint subgraphs of G that contain Z₁ and Z₂, respectively? This problem is known as the 2-Disjoint Connected Subgraphs problem. It is already NP-complete for the class of n-vertex graphs G=(V,E) in which Z₁ and Z₂ each contain a connected set that dominates all vertices in V?(Z₁∪Z₂). We present an O^∗(1.2051ⁿ) time algorithm that solves it for this graph class. As a consequence, we can also solve this problem in O^∗(1.2051ⁿ) time for the classes of n-vertex P₆-free graphs and split graphs. This is an improvement upon a recent O^∗(1.5790ⁿ) time algorithm for these two classes. Our approach translates the problem to a generalized version of hypergraph 2-coloring and combines inclusion/exclusion with measure and conquer.     

Recent developments in large‐area photomasks for display applications

One of the most critical areas in the manufacturing process for FPD panels or shadow masks for CRTs is lithography. Most existing lithography technologies require high‐quality large‐area photomasks. The requirements on these photomasks include positioning accuracy (registration) and repeatability (overlay), systematic image quality errors (“mura” or display quality), and resolution (minimum feature size). The general trend toward higher resolution and improved performance, e.g., for TFT desktop monitors, has put a strong focus on the specifications for large‐area‐display photomasks. This article intends to give an overview of the dominant issues for large‐area‐display photomasks, and illustrates differences compared with other applications. The article will also present state‐of‐the‐art methods and trends. In particular, the aspects of positioning accuracy over large areas and systematic image‐quality errors will be described. New qualitative and objective methods have been developed as means to capture systematic image‐quality errors. Results indicating that errors below 25 nm can be found early in the manufacturing process is presented, thus allowing inspection for visual effects before the actual display is completed. Positioning accuracy below 400 nm (3 sigma) over 720 × 560 mm have been achieved. These results will in the future be extended up toward 1 × 1 m for generation 4 in TFT‐LCD production.     

Artificial Agent Action in Markets

We summarise our experiences of a number of demonstrators and simulation experiments designed to test the feasibility of using artificial decision making agents in real-time domains, and comment on the significance of our results to autonomous artificial agent action patterns in markets. Our main hypothesis is that the use of norms can extend the capability of artificial decision makers beyond what is obtained from implementing individual utility maximizers in keeping with rational choice theory.     

A Secure Grid Medical Data Manager Interfaced to the gLite Middleware

The medical community is producing and manipulating a tremendous volume of digital data for which computerized archiving, processing and analysis is needed. Grid infrastructures are promising for dealing with challenges arising in computerized medicine but the manipulation of medical data on such infrastructures faces both the problem of interconnecting medical information systems to Grid middlewares and of preserving patients’ privacy in a wide and distributed multi-user system. These constraints are often limiting the use of Grids for manipulating sensitive medical data. This paper describes our design of a medical data management system taking advantage of the advanced gLite data management services, developed in the context of the EGEE project, to fulfill the stringent needs of the medical community. It ensures medical data protection through strict data access control, anonymization and encryption. The multi-level access control provides the flexibility needed for implementing complex medical use-cases. Data anonymization prevents the exposure of most sensitive data to unauthorized users, and data encryption guarantees data protection even when it is stored at remote sites. Moreover, the developed prototype provides a Grid storage resource manager (SRM) interface to standard medical DICOM servers thereby enabling transparent access to medical data without interfering with medical practice.     

Development of a Modelica Compiler Using JastAdd

This paper describes experiences from implementing key parts of a compiler for Modelica, an object-oriented language supporting declarative modeling and simulation of complex physical systems. Our implementation uses the attribute-grammar based tool JastAdd. In particular, we discuss the implementation of Modelica name analysis which is highly context-dependent, type analysis which is based on structural subtyping, and flattening which is a fundamental part of the Modelica compilation process.of so called modifications, Modelica.     

An Industrial Case Study on Distributed Prioritisation in Market-Driven Requirements Engineering for Packaged Software

When developing packaged software, which is sold ‘off-the-shelf’ on a worldwide marketplace, it is essential to collect needs and opportunities from different market segments and use this information in the prioritisation of requirements for the next software release. This paper presents an industrial case study where a distributed prioritisation process is proposed, observed and evaluated. The stakeholders in the requirements prioritisation process include marketing offices distributed around the world. A major objective of the distributed prioritisation is to gather and highlight the differences and similarities in the requirement priorities of the different market segments. The evaluation through questionnaires shows that the stakeholders found the process useful. The paper also presents novel approaches to visualise the priority distribution among stakeholders, together with measures on disagreement and satisfaction. Product management found the proposed charts valuable as decision support when selecting requirements for the next release, as they revealed unforeseen differences among stakeholder priorities. Conclusions on stakeholder tactics are provided and issues of further research are identified, including ways of addressing identified challenges.     

Cloud brokering mechanisms for optimized placement of virtual machines across multiple providers

In the past few years, we have witnessed the proliferation of a heterogeneous ecosystem of cloud providers, each one with a different infrastructure offer and pricing policy. We explore this heterogeneity in a novel cloud brokering approach that optimizes placement of virtual infrastructures across multiple clouds and also abstracts the deployment and management of infrastructure components in these clouds. The feasibility of our approach is evaluated in a high throughput computing cluster case study. Experimental results confirm that multi-cloud deployment provides better performance and lower costs compared to the usage of a single cloud only.