Rodrigo T. N. Cardoso

Multi-Objective Evolutionary Optimization of Biological Pest Control with Impulsive Dynamics in Soybean Crops

The biological pest control in agriculture, an environment-friendly practice, maintains the density of pests below an economic injury level by releasing a suitable quantity of their natural enemies. This work proposes a multi-objective numerical solution to biological pest control for soybean crops, considering both the cost of application of the control action and the cost of economic damages. The system model is nonlinear with impulsive control dynamics, in order to cope more effectively with the actual control action to be applied, which should be performed in a finite number of discrete time instants. The dynamic optimization problem is solved using the NSGA-II, a fast and trustworthy multi-objective genetic algorithm. The results suggest a dual pest control policy, in which the relative price of control action versus the associated additional harvest yield determines the usage of either a low control action strategy or a higher one.

Multiobjective dynamic optimization of vaccination campaigns using convex quadratic approximation local search

The planning of vaccination campaigns has the purpose of minimizing both the number of infected individuals in a time horizon and the cost to implement the control policy. This planning task is stated here as a multiobjective dynamic optimization problem of impulsive control design, in which the number of campaigns, the time interval between them and the number of vaccinated individuals in each campaign are the decision variables. The SIR (Susceptible-Infected-Recovered) differential equation model is employed for representing the epidemics. Due to the high dimension of the decision variable space, the usual evolutionary computation algorithms are not suitable for finding the efficient solutions. A hybrid optimization machinery composed by the canonical NSGA-II coupled with a local search procedure based on Convex Quadratic Approximation (CQA) models of the objective functions is used for performing the optimization task. The final results show that optimal vaccination campaigns with different trade-offs can be designed using the proposed scheme.

A Multi-Quantile Approach for Open-Loop Stochastic Dynamic Programming Problems*

Abstract This paper proposes a multi-quantile approach for solving open-loop continuous-variable discrete-time stochastic dynamic programming problems in systems with non-standard probability distribution functions. Instead of using the expected value of the objective function for building the optimization criterion, the decision maker performs a choice on the decision variables over the objective function value quantiles. The proposed procedure relies on a Monte Carlo simulation of the unknown process input outcomes, associated with an open-loop multiobjective optimization. The optimal control comes from a trade-off analysis that considers, for instance, the risk associated with each policy versus its yield.

Multi-objective Design with a Stochastic Validation of Vaccination Campaigns

Abstract The planning of vaccination campaigns should minimize two factors: the number of infected individuals in a time horizon and the cost to implement the control. This problem is stated here as a non-linear dynamic programming optimisation with impulsive control. The traditional SIR (Susceptible-Infected-Recovered) differential equation model is employed for representing the system, and the dynamic programming problem is solved in open-loop, leading to a static non-linear multi-objective optimisation problem. The NSGA-II, which is a standard multi-objective genetic algorithm, is employed as the optimisation machinery. A stochastic dynamic model of the epidemics is employed in order to validate the vaccination strategy, helping in the choice of the specific strategy to be implemented. The final result shows a set of interval of confidence for each optimal policy strategy.

Pareto Optimal Solutions for Stochastic Dynamic Programming Problems via Monte Carlo Simulation

A heuristic algorithm is proposed for a class of stochastic discrete-time continuous-variable dynamic programming problems submitted to non-Gaussian disturbances. Instead of using the expected values of the objective function, the randomness nature of the decision variables is kept along the process, while Pareto fronts weighted by all quantiles of the objective function are determined. Thus, decision makers are able to choose any quantile they wish. This new idea is carried out by using Monte Carlo simulations embedded in an approximate algorithm proposed to deterministic dynamic programming problems. The new method is tested in instances of the classical inventory control problem. The results obtained attest for the efficiency and efficacy of the algorithm in solving these important stochastic optimization problems.

Automated Planning Integrated with Linear Programming Applied in the Container Loading Problem

Abstract In the transporting of loads, the container loading problems have had much repercussion among logistical studies. It can be proved that through using such device there is a reduction on losses, theft, damaging of goods and general costs. The loading problems are typical optimization problems whose objective is maximizing the used volume or minimize the costs parameters. On the other hand, automated planning research, a sub area of artificial intelligence, intends to assist the industrial processes for systematically specifying a set of actions that allows to achieve a predefined goal. In this paradigm the industrial plant is modeled with a series of diagrams respecting the UML standard, for obtaining the problem domain and the initial and final states. It is noteworthy that in the actions specifying process a metric can be added for minimizing purposes, but according to each automated planners characteristics the insertion of such metrics is relatively bounded. In an attempt to accrete both areas of knowledge, it is intended to use one of the widely studied techniques available in the loading problem literature, in particular the linear programming for modeling and the Simplex method for solving, in a way to obtain a specific final state to be utilized in the automated planning solving process. The main proposal of this integration is to create a fully automated system, capable of carrying out the entire loading process through the information referred to the products, to the industrial plant and the transporting devices.

Using convex quadratic approximation as a local search operator in evolutionary multiobjective algorithms

Local search techniques based on Convex Quadratic Approximation (CQA) of functions are studied here, in order to speed up the convergence and the quality of solutions in evolutionary multiobjective algorithms. The hybrid methods studied here pick up points from the nondominated population and determine a CQA for each objective function. Since the CQA of the functions and the respective weighted sums are convex, fast deterministic methods can be used in order to generate approximated Pareto-optimal solutions from the approximated functions. A new scheme is proposed in this paper, using a CQA model that represents a lower bound for the function points, which can be solved via linear programming. This scheme and also another one using the methodology of linear matrix inequality (LMI) for CQA are coupled with a canonical implementation of the NSGA-II. Comparison tests are performed, using Monte Carlo simulations, considering the S-metric with an equivalent final number of evaluated objective functions and the algorithm execution time. The results indicate that the proposed scheme is promising.

An Open-Loop Invariant-Set Approach for Multiobjective Dynamic Programming Problems*

Abstract This paper considers an open-loop continuous-variable dynamic programming approach, inspired in the model predictive control technique, which is shown to be computationally attractive, compared with the traditional dynamic programming algorithm, and more flexible than the closed-loop optimal control, working well with arbitrary constraints and objective functions. The extension of the method to the multiobjective case is shown to be straightforward. An example of dynamic cattle herd management policy is considered.

Decision-making for financial trading: A fusion approach of machine learning and portfolio selection

Abstract Forecasting stock returns is an exacting prospect in the context of financial time series. This study proposes a unique decision-making model for day trading investments on the stock market. In this regard, the model was developed using a fusion approach of a classifier based on machine learning, with the support vector machine (SVM) method, and the mean-variance (MV) method for portfolio selection. The models experimental evaluation was based on assets from the Sao Paulo Stock Exchange Index (Ibovespa). Monthly rolling windows were used to choose the best-performing parameter sets (the in-sample phase) and testing (the out-of-sample phase). The monthly windows were composed of daily rolling windows, with new training of the classifying algorithm and portfolio optimization. A total of 81 parameter arrangements were formulated. To compare the proposed models performance, two other models were suggested: (i) SVM + 1/N, which maintained the process of classifying the trends of the assets that reached a certain target of gain and which invested equally in all assets that had positive signals in their classifications, and (ii) Random + MV, which also maintained the selection of those assets with a tendency to reach a certain target of gain, but where the selection was randomly defined. Then, the portfolios composition was determined using the MV method. Together, the alternative models registered 36 parameter variations. In addition to these two models, the results were also compared with the Ibovespas performance. The experiments were formulated using historical data for 3716 trading days for the out-of-sample analysis. Simulations were conducted without including transaction costs and also with the inclusion of a proportion of such costs. We specifically analyzed the effect of brokerage costs on buying and selling stocks on the Brazilian market. This study also evaluated the classifiers performance, portfolios’ cardinality, and models’ returns and risks. The proposed main model showed significant results, although demand for trading value can be a limiting factor for its implementation. Nonetheless, this study extends the theoretical application of machine learning and offers a potentially practical approach to anticipating stock prices.

Otimização mono-objetivo no controle do mosquito Aedes aegypti por meio de um modelo de duas populações com influência da pluviosidade

O mosquito Aedes aegypti ´r o principal transmissor das arboviroses dengue, chikungunya, zika e febre amarela que vem acometendo a populacao brasileira ao longo dos anos. [...]