We show that all algorithms that search for an extremum of a cost function perform exactly the same when averaged over all possible cost functions In particular if algorithm A outperforms algorithm B on some cost functions then loosely speaking there must exist exactly as many other functions where B outperforms A Starting from this we analyze a number of the other a priori characteristics of t...

We propose an algorithm called query by com mittee in which a committee of students is trained on the same data set The next query is chosen according to the principle of maximal disagreement The algorithm is studied for two toy models the high low game and perceptron learning of another perceptron As the number of queries goes to in nity the committee algo rithm yields asymptotically nite info...

Inductive learning in First-Order Logic (FOL) is a hard task due to both the pro­ hibitive size of the search space and the computational cost of evaluating hypotheses. This paper introduces an evolutionary algo­ rithm for concept learning in (a fragment of) FOL. The algorithm evolves a population of Horn clauses by repeated selection, mutation and optimization of more fit clauses. Its main nov...

Quadratic stochastic progr ms (QSP) with recourse ca be formulate s nonlinear co ve progr mming probl ms. By attaching a Larange ultiplier vector to the nonlinear convex program, we rewrite the problem as a ystem of nonsmooth equations. We co sid a Newton-like method for solving he system and establish global c nverg nce a d l c superlinear co vergence of the method. Several methods f r special...

We propose an algorithm for solving the maximumweighted stable set problem on claw-free graphsthat runs in O(n3)−time, drastically improving theprevious best known complexity bound. This algo-rithm is based on a novel decomposition theoremfor claw-free graphs, which is also introduced in thepresent paper. Despite being weaker than the wellknown structure result for claw-...

This article discusses a rule-based fuzzy model for the identi cation of nonlinear MISO (multiple input, single output) systems. The dis cussed method of fuzzy modeling consists of two parts: structure modeling, i.e. determing the num ber of rules and input variables involved respec tively, and parameter optimization, i.e. optimizing the location and form of the curves which describe the fuz...

Caching is a standard solution to the problem of insuf cient bandwidth caused by the rapid increase of in formation circulation across the Internet Cache con sistency mechanisms are a crucial component of each cache scheme in uencing the cache usefulness and re liability This paper presents a model for optimizing Internet cache content by the use of a genetic algo rithm and examines the model b...

Distributed video coding (DVC) reverses the traditional coding paradigm of complex encoders allied with basic decoding, to one where the computational cost is largely incurred by the decoder. This enables low-cost, resource-poor sensors to be used at the transmitter in various applications including multi-sensor surveillance. A key constraint governing DVC performance is the quality of side inf...

In this paper, an algorithm based on penalty cost function for synthesizing flat-top patterns is proposed. A descent algorithm (DA) as its optimizing approach is proposed in this paper as well. Apparently, whole algorithm efficiency totally depends on the DA. Unlike traditional descent method, the DA defines step length by solving a inequality, instead of Wolf or Armijo-type search rule, stimul...

Interference and multipath is one of the current issues in a wireless communication system, with complicated scenarios of environment especially in urban areas with a high number of users. Introducing smart antenna systems at the base station can contribute to reducing interference and improve quality of service. This paper proposes and explores the application of artificial immune system and n...