Victor Parque

On succinct representation of directed graphs

Directed graphs encode meaningful dependencies among objects ubiquitously. This paper introduces new and simple representations for labeled directed graphs with the properties of being succinct (space is information-theoretically optimal); in which we avoid exploiting a-priori knowledge on digraph regularity such as triangularity, separability, planarity, symmetry and sparsity. Our results have direct implications to model directed graphs by using single integer numbers effectively, which is significant to enable canonical (generation of graph instances is unique) and efficient (coding and decoding take polynomial time) encodings for learning and optimization algorithms. To the best of our knowledge, the proposed representations are the first known in the literature.

On k-subset sum using enumerative encoding

Being a significant construct in a wide range of combinatorial problems, the k-subset sum problem (k-SSP) computes k-element subsets, out of an n-element set, satisfying a user-defined aggregation value. In this paper, we formulate the k-subset sum problem as a search (optimization) problem over the space of integers associated with combination elements. And by using rigorous computational experiments using the search space over more than 1014 integer numbers, we show that our approach is effective and efficient: it is feasible to find any combination with a user-defined sum within 104 function evaluations by using a gradient-free optimization algorithm. Our scheme opens the door to further advance the understanding of combinatorial problems by improved/tailored gradient-free optimization algorithms based on enumerative encoding. Also, our approach realizes the practical building block for combinatorial problems in planning and operations research using k-SSP concepts.

Learning the Optimal Product Design Through History

The search for novel and high-performing product designs is a ubiquitous problem in science and engineering: aided by advances in optimization methods the conventional approaches usually optimize a (multi) objective function using simulations followed by experiments.

Route bundling in polygonal domains using Differential Evolution

Route bundling implies compounding multiple routes in a way that anchoring points at intermediate locations minimize a global distance metric to obtain a tree-like structure where the roots of the tree (anchoring points) serve as coordinating locus for the joint transport of information, goods and people. Route bundling is a relevant conceptual construct in a number of path-planning scenarios where the resources and means of transport are scarce/expensive, or where the environments are inherently hard to navigate due to limited space. In this paper we propose a method for searching optimal route bundles based on a self-adaptive class of Differential Evolution using a convex representation. Rigorous computational experiments in scenarios with and without convex obstacles show the feasibility and efficiency of our approach.

Bundling n-Stars in Polygonal Maps

This paper aims at computing minimal-length tree layouts given an n-star graph in a polygonal map. This problem is strongly related to the edge bundling problem, which consists of compounding the edges of an input graph to obtain topologically compact graph layouts being free of clutter and easy to visualize. Computational experiments using a diverse set of polygonal maps and number of edges in the input graph shows the feasibility, efficiency and robustness of our approach.

A method to learn high-performing and novel product layouts and its application to vehicle design

In this paper we aim at tackling the problem of searching for novel and high-performing product designs. Generally speaking, the conventional schemes usually optimize a (multi) objective function on a dynamic model/simulation, then perform a number of representative real-world experiments to validate and test the accuracy of the some product performance metric. However, in a number of scenarios involving complex product configuration, e.g. optimum vehicle design and large-scale spacecraft layout design, the conventional schemes using simulations and experiments are restrictive, inaccurate and expensive.In this paper, in order to guide/complement the conventional schemes, we propose a new approach to search for novel and high-performing product designs by optimizing not only a proposed novelty metric, but also a performance function which is learned from historical data. Rigorous computational experiments using more than twenty thousand vehicle models over the last thirty years and a relevant set of well-known gradient-free optimization algorithms shows the feasibility and usefulness to obtain novel and high performing vehicle layouts under tight and relaxed search scenarios.The promising results of the proposed method opens new possibilities to build unique and high-performing systems in a wider set of design engineering problems.

Genetic network programming with changing structures for a novel stock selection model

Stock selection involves the continuous quest for the margin of safety, or a favorable difference between the stock price and its intrinsic value. Although this variable might not be quantified with exact precision, it may be approximated through the underlying relationships in financial markets and the real economy. We propose Genetic Network Programming with changing structures(GNP-cs), a novel evolutionary based algorithm to approximate these relationships through graph networks, and build asset selection models to identify the prospective stocks in the context of changing environments. GNP-cs uses functionally distributed systems to monitor the change of the economic environment and execute the strategy for stock selection adaptively. The comparison shows that the proposed scheme outperforms the standard stock selection styles using the stocks listed in the Russell 3000 Index. This paper suggests that the use of evolutionary computing techniques is an excellent tool to tackle the stock selection problem, whose advantages imply the usefulness to manage the risk and safeguard investments.

Asset selection in global financial markets using Genetic Network Programming

Asset selection is a challenging task in the complex global financial system, whose nature has highlighted the need to rethink conventional practices. The attractive and non-toxic assets must be kept on the eye so that our financial systems sustain building blocks in our economic systems. This paper presents an asset selection framework using Genetic Network Programming(GNP). GNP handles evolvable graph structures that prevent the size expansion for dynamic and complex environments, which in turn make it suitable for dealing with decision processes effectively under uncertainty such as partially observable Markov decision processes. Simulations using stocks, bonds and currencies from relevant financial markets in USA, Europe and Asia show the competitive advantages of the proposed method against relevant selection strategies in the finance literature.

Computing path bundles in bipartite networks

Path bundling, a class of path planning problem, consists of compounding multiple routes to minimize a global distance metric. Naturally, a tree-like structure is obtained as a result wherein roots play the role of coordinating the joint transport of information, goods, and people. In this paper we tackle the path bundling problem in bipartite networks by using gradient-free optimization and a convex representation. Then, by using 7,500 computational experiments in diverse scenarios with and without obstacles, implying 7.5 billion shortest path computations, show the feasibility and efficiency of the mesh adaptive search.

Path planning on hierarchical bundles with differential evolution

Computing hierarchical routing networks in polygonal maps is significant to realize the efficient coordination of agents, robots and systems in general; and the fact of considering obstacles in the map, makes the computation of efficient networks a relevant need for cluttered environments. In this paper, we present an approach to compute the minimal-length hierarchical topologies in polygonal maps by Differential Evolution and Route Bundling Concepts. Our computational experiments in scenarios considering convex and non-convex configuration of polygonal maps show the feasibility of the proposed approach.