Free Parallel Data Mining Bin Li New York University, 1998 Research Advisor: Professor Dennis Shasha Data mining is the emerging field of applying statistical and artificial intelligence techniques to the problem of finding novel, useful, and non-trivial patterns from large databases. This thesis presents a framework for easily and efficiently parallelizing data mining algorithms. We propose an acyclic directed graph structure, exploration dag (E- dag), to characterize the computation model of data mining algorithms in classification rule mining, association rule mining, and combinatorial pattern discovery. An E-dag can be constructively formed in parallel from specifications of a data mining problem, then a parallel E-dag traversal is performed on the fly to efficiently solve the problem. The effectiveness of the E-dag framework is demonstrated in biological pattern discovery applications. We also explore data parallelism in data mining applications. The cross-validation and the...

Carrying out empirical studies to understand the problems of software engineering is widely held to be of importance. A view less widely held within the software engineering community is that experiments should be replicated externally (i.e. by different researchers) to both verify and validate the original results and so build up a sound body of knowledge (as opposed to drawing conclusions from a single study). This paper presents the results of an external replication of an experiment which tested the benefits to maintenance of using modular code against non-modular (monolithic) code. The results of the original experiment [8, 9] showed that a modular program could be maintained significantly faster than an equivalent monolithic version of the same program under the condition that modularity has been used to implement information hiding which localizes changes required by a modification. The results of our replication, however, were strikingly different from those of the original and...

The classification algorithm CLEF combines a version of a linear machine known as a \Phi- machine with a non-linear function approximator that constructs its own features. The algorithm finds non-linear decision boundaries by constructing features that are needed to learn the necessary discriminant functions. The CLEF algorithm is proven to separate all consistently labelled training instances, even when they are not linearly separable in the input variables. The algorithm is illustrated on a variety of tasks, showing an improvement over C4.5, a state-of-art decision tree learning algorithm. 1 Introduction The task of classification is to find an approximate definition for an unknown function f : X ! fc 1 ; ::c R g; R 2 based on a set of training examples of the form hx i ; f(x i )i. The components of an instance vector x i can take values from discrete or continuous domains. It is also possible that the values of one or more components are missing or imprecisely recorded for cert...

This paper describes an Expert System for apple orchard integrated control. The system consists of two parts. In the first one, help to the agricultural producer is provided in classifying the producer observations. In the second part an interactive abductive module suggests explanations of those observations. Classification has been implemented with the aid of discrimination net techniques. The abductive module exploits results from the parsimonious covering theory by Peng and Reggia. The system has been implemented in KEE with an extensive use of the graphical tools provided by the language. In a future work it will be extended with a module responsible for treatment suggestion. 1 Introduction The general reevaluation of ecological-environmental problems has increased interest in those forms of agriculture characterized by reduced environmental impact. In fact, the traditional model of agricultural development, based on massive introduction of innovative technologies, while allowi...

This paper describes a knowledge-based system able to favor the full application of integrated protection criteria in apple orchards. Integrated protection is definable as the reasoned application of agronomic methods products as well as chemicals, to allow optimal productive factors, while respecting the farm worker, the environment, and the consumer. The system consists of two parts. In the first one, help to the agricultural producer is provided in classifying the findings. In the second part an interactive abductive module suggests explanations of those findings. The system has been implemented in KEE with an extensive use of the graphical tools provided by the language. Keywords: Agriculture, Integrated Pest Management, Diagnosis, Discrimination Trees, Parsimonious Covering Theory. 1 Introduction The general reevaluation of ecological-environmental problems has risen the interest in those forms of agriculture characterized by reduced environmental impact. In fact, the traditional...

The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies or The military forces of the future will use multi-agent robotic workforces for reconnaissance and surveillance, logistics and support, communications infrastructure, forward-deployed offensive operations, and as tactical decoys to conceal maneuver by manned assets. Towards this end, there is a clear and definite need for optimal, or pareto-optimal, multi-robot control strategies in the synthesis, design, implementation, and fielding of autonomous and semi-autonomous teams of combat robots for military systems. Proven coordination methods are essential to enable interactions with in dynamic and hostile environments, synchronized maneuvers, sensible and robust rules of engagement (ROE), and reliable field behavior. This paper surveys the state-of-the art in autonomous multi-robot work systems and investigates the relative strengths and weaknesses of each of approach with respect to military applications. Further, it outlines guidance for future research