Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

This volume contains the papers accepted for presentation at the Fourth International Symposium on Programming Language Implementation and Logic Programming (PLILP '92) held in Leuven, Belgium, August 26-28, 1992. Theaim of the symposium was to explore new declarative concepts, methods and techniques relevant for implementation of all kinds of programming languages, whether algorithmic or declarative. The symposium gathered researchers from the fields of algorithmic programming languages as well as logic, functional, and object-oriented programming. The topics discussed in the volume include: - Declarative implementation techniques; - Implementation of declarative concepts and modules; - Compiler specification and construction; - Program transformation; - Programming environments for declarative languages; - Amalgamation of logic programming with algorithmic, functional, or object-oriented techniques; - Executable specification.

681.3*D31 --- 681.3*F33 --- 681.3*F41 --- 681.3*F42 --- 681.3*I23 --- Formal definitions and theory: semantics; syntax (Programming languages)--See also {681.3*D21}; {681.3*F31}; {681.3*F32}; {681.3*F42}; {681.3*F43} --- Studies of program constructs: control primitives; functional constructs; program and recursion schemes; type structure (Logics and meanings of programs)-- See also {681.3*D32}; {681.3*D33} --- Mathematical logic: computability theory; computational logic; lambda calculus; logic programming; mechanical theorem proving; model theory; proof theory;recursive function theory--See also {681.3*F11}; {681.3*I22}; {681.3*I23} --- Grammars and other rewriting systems: decision problems; grammar types; parallel rewriting systems; parsing; thue systems (Mathematical logic and formal languages)--See also {681.3*D31} --- Deduction and theorem proving: answer/reason extraction; reasoning; resolution; metatheory; mathematical induction; logic programming (Artificial intelligence) --- 681.3*I23 Deduction and theorem proving: answer/reason extraction; reasoning; resolution; metatheory; mathematical induction; logic programming (Artificial intelligence) --- 681.3*F42 Grammars and other rewriting systems: decision problems; grammar types; parallel rewriting systems; parsing; thue systems (Mathematical logic and formal languages)--See also {681.3*D31} --- 681.3*F41 Mathematical logic: computability theory; computational logic; lambda calculus; logic programming; mechanical theorem proving; model theory; proof theory;recursive function theory--See also {681.3*F11}; {681.3*I22}; {681.3*I23} --- 681.3*F33 Studies of program constructs: control primitives; functional constructs; program and recursion schemes; type structure (Logics and meanings of programs)-- See also {681.3*D32}; {681.3*D33} --- 681.3*D31 Formal definitions and theory: semantics; syntax (Programming languages)--See also {681.3*D21}; {681.3*F31}; {681.3*F32}; {681.3*F42}; {681.3*F43} --- Programming languages (Electronic computers) --- Congresses --- Logic programming --- Computer science. --- Information theory. --- Logic design. --- Artificial intelligence. --- Programming Techniques. --- Theory of Computation. --- Programming Languages, Compilers, Interpreters. --- Logics and Meanings of Programs. --- Mathematical Logic and Formal Languages. --- Artificial Intelligence. --- AI (Artificial intelligence) --- Artificial thinking --- Electronic brains --- Intellectronics --- Intelligence, Artificial --- Intelligent machines --- Machine intelligence --- Thinking, Artificial --- Bionics --- Cognitive science --- Digital computer simulation --- Electronic data processing --- Logic machines --- Machine theory --- Self-organizing systems --- Simulation methods --- Fifth generation computers --- Neural computers --- Design, Logic --- Design of logic systems --- Digital electronics --- Electronic circuit design --- Logic circuits --- Switching theory --- Communication theory --- Communication --- Cybernetics --- Informatics --- Science

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Over the last decades, the amount of available data has grown rapidly. Data mining is the field studying how to discover interesting knowledge from data. When the data points are drawn identically and independently (i.i.d.), classical data mining techniques work well in practice. In particular, the employed statistical theories make sure that, if the sample size is large enough, then the mined knowledge is very likely to have a good generalization ability. However, the structure of available data becomes more and more complex, and the assumption that data points are drawn i.i.d. is violated. This usually makes the data less informative. We begin with defining support measures which gauge the frequency of a given pattern in a given dataset. For transactional datasets, one can simply count the occurrences of the pattern in the dataset. However, if the dataset is a large network, occurrences of a subgraph pattern may overlap, which makes the support measure definition less straightforward. For example, if we ignore overlaps and directly count the occurrences, we lose the important property of anti-monotonicity, which allows us to effectively prune in search spaces, An important class of anti-monotonic support measures relies on overlap graphs. Unfortunately, but all earlier overlap graph based support measures are expensive to compute. In this thesis, we introduce the concept of overlap hypergraphs, and propose an overlap hypergraph based support measure which is anti-monotonic and can be computed efficiently. In order to understand this support measure from a statistical point of view, we consider the problem of statistics on networked examples. We propose a model for representing networked examples as hyperedges in a hypergraph and reasonable assumptions to replace the classical i.i.d. assumption. Based on this model, we design an estimator on networked examples with the minimum variance and generalize Chernoff-Hoeffding inequalities to weighted networked examples. We can observe a close relationship between the effective sample size and the graph support measure we proposed. Our statistical concentration results can be used to adapt learning principles to learning from networked examples. We illustrate this by proving generalization error bounds for the empirical risk minimization principle.

Theses

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Logic Programming --- Programmation logique