Alex Fukunaga

Success-history based parameter adaptation for Differential Evolution

Differential Evolution is a simple, but effective approach for numerical optimization. Since the search efficiency of DE depends significantly on its control parameter settings, there has been much recent work on developing self-adaptive mechanisms for DE. We propose a new, parameter adaptation technique for DE which uses a historical memory of successful control parameter settings to guide the selection of future control parameter values. The proposed method is evaluated by comparison on 28 problems from the CEC2013 benchmark set, as well as CEC2005 benchmarks and the set of 13 classical benchmark problems. The experimental results show that a DE using our success-history based parameter adaptation method is competitive with the state-of-the-art DE algorithms.

Improving the search performance of SHADE using linear population size reduction

SHADE is an adaptive DE which incorporates success-history based parameter adaptation and one of the state-of-the-art DE algorithms. This paper proposes L-SHADE, which further extends SHADE with Linear Population Size Reduction (LPSR), which continually decreases the population size according to a linear function. We evaluated the performance of L-SHADE on CEC2014 benchmarks and compared its search performance with state-of-the-art DE algorithms, as well as the state-of-the-art restart CMA-ES variants. The experimental results show that L-SHADE is quite competitive with state-of-the-art evolutionary algorithms.

Automated discovery of local search heuristics for satisfiability testing

The development of successful metaheuristic algorithms such as local search for a difficult problem such as satisfiability testing (SAT) is a challenging task. We investigate an evolutionary approach to automating the discovery of new local search heuristics for SAT. We show that several well-known SAT local search algorithms such as Walksat and Novelty are composite heuristics that are derived from novel combinations of a set of building blocks. Based on this observation, we developed CLASS, a genetic programming system that uses a simple composition operator to automatically discover SAT local search heuristics. New heuristics discovered by CLASS are shown to be competitive with the best Walksat variants, including Novelty. Evolutionary algorithms have previously been applied to directly evolve a solution for a particular SAT instance. We show that the heuristics discovered by CLASS are also competitive with these previous, direct evolutionary approaches for SAT. We also analyze the local search behavior of the learned heuristics using the depth, mobility, and coverage metrics proposed by Schuurmans and Southey.

Evaluating the performance of SHADE on CEC 2013 benchmark problems

This paper evaluates the performance of Success-History based Adaptive DE (SHADE) on the benchmark set for the CEC2013 Competition on Real-Parameter Single Objective Optimization. SHADE is an adaptive differential algorithm which uses a history-based parameter adaptation scheme. Experimental results on 28 problems from the CEC2013 benchmarks for 10, 30, and 50 dimensions are presented, including measurements of algorithmic complexity. In addition, we investigate the parameter adaptation behavior of SHADE on these instances.

Towards an application framework for automated planning and scheduling

A number of successful applications of automated planning and scheduling applications to spacecraft operations have recently been reported in the literature. However, these applications have been one-of-a-kind applications that required a substantial amount of development effort. In this paper, we describe ASPEN (Automated Planning/Scheduling Environment), a modular, reconfigurable application framework which is capable of supporting a wide variety of planning and scheduling applications. We describe the architecture of ASPEN, as well as a number of current spacecraft control/operations applications in progress.

Using ASPEN to automate EO-1 activity planning

This paper describes the application of an automated planning and scheduling system to the NASA Earth Orbiting 1 (EO-1) mission. The planning system, ASPEN, is used to autonomously schedule the daily activities of the satellite. The satellite and operations constraints are encoded within a software model used by the planner. This paper includes a description of the planning system and the associated modeling language. We then discuss how we encoded the EO-1 spacecraft operations with the modeling language. We conclude with a description of the end-to-end planning system as we envision it for EO-1.

Automatic detection of dust devils and clouds on Mars

The acquisition of science data in space applications is shifting from teleoperated data collection to an automated onboard analysis, resulting in improved data quality, as well as improved usage of limited resources such as onboard memory, CPU, and communications bandwidth. Science instruments onboard a modern deep-space spacecraft can acquire much more data that can be downloaded to Earth, given the limited communication bandwidth. Onboard data analysis offers a means of compressing the huge amounts of data collected and downloading only the most valuable subset of the collected data. In this paper, we describe algorithms for detecting dust devils and clouds onboard Mars rovers, and summarize the results. These algorithms achieve the accuracy required by planetary scientists, as well as the runtime, CPU, memory, and bandwidth constraints set by the engineering mission parameters. The detectors have been uploaded to the Mars Exploration Rovers, and currently are operational. These detectors are the first onboard science analysis processes on Mars.

Bin completion algorithms for multicontainer packing, knapsack, and covering problems

Many combinatorial optimization problems such as the bin packing and multiple knapsack problems involve assigning a set of discrete objects to multiple containers. These problems can be used to model task and resource allocation problems in multi-agent systems and distributed systms, and can also be found as subproblems of scheduling problems. We propose bin completion, a branch-and-bound strategy for one-dimensional, multicontainer packing problems. Bin completion combines a bin-oriented search space with a powerful dominance criterion that enables us to prune much of the space. The performance of the basic bin completion framework can be enhanced by using a number of extensions, including nogood-based pruning techniques that allow further exploitation of the dominance criterion. Bin completion is applied to four problems: multiple knapsack, bin covering, min-cost covering, and bin packing. We show that our bin completion algorithms yield new, state-of-the-art results for the multiple knapsack, bin covering, and min-cost covering problems, outperforming previous algorithms by several orders of magnitude with respect to runtime on some classes of hard, random problem instances. For the bin packing problem, we demonstrate significant improvements compared to most previous results, but show that bin completion is not competitive with current state-of-the-art cutting-stock based approaches.

Restart Scheduling for Genetic Algorithms

In order to escape from local optima, it is standard practice to periodically restart heuristic optimization algorithms such as genetic algorithm according to some restart criteria/policy. This paper addresses the issue of finding a good restart strategy in the context of resource-bounded optimization scenarios, in which the goal is to generate the best possible solution given a fixed amount of time. We propose the use of a restart scheduling strategy which generates a static restart strategy with optimal expected utility, based on a database of past performance of the algorithm on a class of problem instances. We show that the performance of static restart schedules generated by the approach can be competitive to that of a commonly used dynamic restart strategy based on detection of lack of progress.

Automating the process of optimization in spacecraft design

Spacecraft design optimization is a difficult problem, due to the complexity of optimization cost surfaces and the human expertise in optimization that is necessary in order to achieve good results. In this paper, we propose the use of a set of generic, metaheuristic optimization algorithms (e.g., genetic algorithms, simulated annealing), which is configured for a particular optimization problem by an adaptive problem solver based on artificial intelligence and machine learning techniques. We describe work in progress on OASIS, a system for adaptive problem solving based on these principles.