State pattern

Abstract: Object and pattern detection is a classical computer vision problem with many potential applications, ranging from automatic target recognition to image-based industrial inspection tasks in assembly lines. While there have been some successful object and pattern detection systems in the past, most such systems handle only specific rigid objects or patterns that can be accurately described by fixed geometric models or pictorial templates. This thesis presents a learning based approach for detecting classes of objects and patterns with variable image appearance but highly predictable image boundaries. Some examples of such object and pattern classes include human faces, aerial views of structured terrain features like volcanoes, localized material defect signatures in industrial parts, certain tissue anomalies in medical images, and instances of a given digit or character, which may be written or printed in many different styles. The thesis consists of two parts. In part one, we introduce our object and pattern detection approach using a concrete human face detection example. The approach first builds a distribution-based model of the target pattern class in an appropriate feature space to describe the target's variable image appearance. It then learns from examples a similarity measure for matching new patterns against the distribution-based target model. We also discuss some pertinent learning issues, including ideas on virtual example generation and example selection. The approach makes few assumptions about the target pattern class and should therefore be fairly general, as long as the target class has predictable image boundaries. We show that this is indeed the case by demonstrating the technique on two other pattern detection/recognition problems. Because our object and pattern detection approach is very much learning-based, how well a system eventually performs depends heavily on the quality of training examples it receives. The second part of this thesis looks at how one can select high quality examples for function approximation learning tasks. Active learning is an area of research that investigates how a learner can intelligently select future training examples to get better approximation results with less data. We propose an active learning formulation for function approximation, and show for three specific approximation function classes, that the active example selection strategy learns its target with fewer data samples than random sampling. Finally, we simplify the original active learning formulation, and show how it leads to a tractable example selection paradigm, suitable for use in many object and pattern detection problems. (Copies available exclusively from MIT Libraries, Rm. 14-0551, Cambridge, MA 02139-4307. Ph. 617-253-5668; Fax 617-253-1690.)

Abstract: This paper develops a theory and algoritbrns for an application problem arising in software maintenance. The application is to track down duplication in a large software system. We want to find not only exact matches between sections of code, but parametrized matches, where a parametrized match between two sections of code means that one section can be transformed into the other by replacing the parameter names (e.g. identifiers and constants) of one section by the parameter names of the other via a one-to-one function. This paper formalizes this problem in terms of parametrized strings and parametrized pattern matching and detirtes a new data structure (parametrized sujjfi.x tree) suitable for parametrized pattern matching. It gives efficient algorithms for constructing this data structure, efficient algorithms for parametrized pattern matchmg, and an efficient algorithm for timing all maximal parametrized matches over a threshold length in a parametrized string. The algorithms for constructing parametrized suffix trees and for reporting duplication over a threshold length have been implemented. Tests on C code indicate that these algorithms should perform well in the application.

Abstract: Software design patterns are the core abstractions from successful recurring problem solutions in software design. Composite design patterns are the core abstractions from successful recurring frameworks. A composite design pattern is a pattern that is best described as the composition of further patterns the integration of which shows a synergy that makes the composition more than just the sum of its parts. This paper presents examples of composite patterns, discusses a role-based analysis and composition technique, and demonstrates that composite patterns extend the pattern idea from single problem solutions to object-oriented frameworks.

Abstract: There has been substantial recent interest in captured design expertise expressed as design patterns. Prevalent descriptions of these design patterns suffer from two demerits. Firstly, they capture specific instances of pattern deployment, rather than the essential pattern itself, thus the spirit of the pattern is often lost in the superfluous details of the specific instances described. Secondly, existing pattern descriptions rely upon relatively informal diagrammatic notations supplemented with natural language annotations. This can result in imprecision and ambiguity. This paper addresses these problems by separating the specification of patterns into three models (role, type, and class). The most abstract (role-centric) model presents patterns in their purest form, capturing their essential spirit without deleterious detail. A role-model is refined by a type-model (adding usually-domain-specific constraints), which is further refined by a class-model (forming a concrete deployment). We utilise recent advances in visual modelling notation to achieve greater precision without resorting to obtuse mathematical symbols. A set-oriented view of state, operations, and instances is adopted, permitting their abstract presentation in models via this visual notation. This paper utilises these ideas in the unambiguous specification of a selection of prominent design patterns. The expectation is that precise visual pattern specification will firstly enable clear communication between domain experts and pattern writers (and ultimately pattern users), and secondly enable CASE tool support for design patterns, permitting the designer (pattern user) to operate at a higher level of abstraction without ambiguity.