Elva Díaz

Comparison of Two Evolvable Systems in the Automated Analog Circuit Synthesis

This paper presents the comparison of two different algorithms: a Univariate Marginal Distribution Algorithm for Analog Circuits (UMDA-AC) and a Genetic Algorithm for Analog Circuits (GA-AC). These algorithms are compared in performing the synthesis of topology and sizing of an analog low pass filter. Modeling of circuits is made by means of a linear representation technique with a variable length chromosome. Evaluation of circuits’ functionality is carried out by Simulation Program with Integrated Circuits Emphasis (Spice), since one of the objectives is to keep as low as possible the amount of non Spice-Simulable circuits while keep elements’ values within preferred ones. Experiments show the effectiveness of a set of evolvable mechanisms in both algorithms, and while GA-AC and its three genetic operators are more able to keep low the rate of non Spice-Simulable circuits; UMDA-AC performs less evaluations by means of its estimated distribution

Probabilistic Graphical Markov Model Learning: An Adaptive Strategy

In this paper an adaptive strategy to learn graphical Markov models is proposed to construct two algorithms. A statistical model complexity index (SMCI ) is defined and used to classify models in complexity classes, sparse, medium and dense. The first step of both algorithms is to fit a tree using the Chow and Liu algorithm. The second step begins calculating SMCI and using it to evaluate an index (EMUBI ) to predict the edges to add to the model. The first algorithm adds the predicted edges and stop, and the second, decides to add an edge when the fitting improves. The two algorithms are compared by an experimental design using models of different complexity classes. The samples to test the models are generated by a random sampler (MSRS). For the sparse class both algorithms obtain always the correct model. For the other two classes, efficiency of the algorithms is sensible to complexity.

Planning and Allocation Tasks in a Multicomputer System as a Multi-objective Problem

In this article we address the task planning and assignment problem in a multicomputer system using architectural 2D mesh. The problem of planning and allocation of tasks to a group of computers consists of several sub-problems that can be made to correspond to functions to optimize.The proposed solution to this problem is; first: establish the identification of distinct parts that are involved, such as; maximizing processor usage, minimize task wait time in the queue and avoid indefinite task delay (starvation). Second: a planning algorithm and an allocation algorithm are implemented through the search engine within the queue, the first algorithm makes a previous planning to the allocation to identify the task lists that fit in the mesh, and the second is a sole variant distribution algorithm to identify the best allocations in the processor mesh through a dynamic quadratic allocation. Finally, our final results are presented; they allow us to see that a previous allocation in the queue and a search engine allocation of the tasks best positions in the available (free) sub meshes, are determining factors for bettering the longevity of the processors and optimize answer time in a multicomputer system.

Evolutionary Multi-Objective Algorithms

The versatility that genetic algorithm (GA) has proved to have for solving different problems, has make it the first choice of researchers to deal with new challenges. Currently, GAs are the most well known evolutionary algorithms, because their intuitive principle of operation and their relatively simple implementation; besides they have the ability to reflect the philosophy of evolutionary computation in an easy and quick way.

Markov Structure Random Sampler (MSRS) Algorithm from Unrestricted Discrete Graphic Markov Models

In this paper a new model random sampler algorithm, only based on the interaction structure of the model is presented. It means that the vector values of the parameters of the distribution are not needed to perform the sample generation. The algorithm is tested generating nine structure models of 10, 12, and 14 variables, and conditional independence restrictions with structures, sparse, mean and dense. Eight random samples are generated from each structure model, for a total of 72 random samples. To validate the results an external criterion is used. Every sample is given to the model selection algorithm implemented in MIM software, which obtains the structure of the departure model for 93% of the samples. In all cases the generation time of a sample was not greater than 4 minutes. The mean run time grows with the density of the models. The MSRS algorithm converges in at most 4 iterations.

Discrete Graphic Markov Model Selection by Multiobjective Genetic Algorithm (GMS-MGA)

The problem of graphic Markov model selection (GMS) is considered as a multiobjective one, where the objectives are: (1) best fitting of the model to the sample data and, (2) least possible number of edges, and the fitting criteria function is the Kullback-Leibler. The multiobjective strategy is to obtain an approximate Pareto front using a multi-starting multiobjective genetic algorithm (MGA). To test the performance of the algorithm, 48 samples of 6 different model complexities are generated using a Markov model random sampler (MMRS) and used as benchmarks. The performance is assessed through the times that the Pareto front contains the true model. As results the algorithm obtains the true models in 93% of the cases, the complexity of the model made a difference in the performance of the algorithm. The mean time of execution is least or equal to 2 minutes for 10 binary variables in a PC

Improving Information Retrieval Through a Global Term Weighting Scheme

The output of an information retrieval system is an ordered list of documents corresponding to the user query, represented by an input list of terms. This output relies on the estimated similarity between each document and the query. This similarity depends in turn on the weighting scheme used for the terms of the document index. Term weighting then plays a big role in the estimation of the aforementioned similarity. This paper proposes a new term weighting approach for information retrieval based on the marginal frequencies. Consisting of the global count of term frequencies over the corpus of documents, while conventional term weighting schemes such as the normalized term frequency takes into account the term frequencies for particular documents. The presented experiment shows the advantages and disadvantages of the proposed retrieval scheme. Performance measures such as precision and recall and F-Score are used over classical benchmarks such as CACM to validate the experimental results.The output of an information retrieval system is an ordered list of documents corresponding to the user query, represented by an input list of terms. This output relies on the estimated similarity between each document and the query. This similarity depends in turn on the weighting scheme used for the terms of the document index. Term weighting then plays a big role in the estimation of the aforementioned similarity. This paper proposes a new term weighting approach for information retrieval based on the marginal frequencies. Consisting of the global count of term frequencies over the corpus of documents, while conventional term weighting schemes such as the normalized term frequency takes into account the term frequencies for particular documents. The presented experiment shows the advantages and disadvantages of the proposed retrieval scheme. Performance measures such as precision and recall and F-Score are used over classical benchmarks such as CACM to validate the experimental results.