As in multiobjective optimization, multimodal optimization generates solution sets that must be measured in order to compare different optimization algorithms. We discuss similarities and differences in the requiremen...
详细信息
ISBN:
(纸本)9783319011271
As in multiobjective optimization, multimodal optimization generates solution sets that must be measured in order to compare different optimization algorithms. We discuss similarities and differences in the requirements for measures in both domains and suggest a property-based taxonomy. The process of measuring actually consists of two subsequent steps, a subset selection that only considers 'suitable' points (or just takes all available points of a solution set) and the actual measuring. Known quality indicators often rely on problem knowledge (objective values and/or locations of optima and basins) which makes them unsuitable for real-world applications. Hence, we propose a new subset selection heuristic without such demands, which thereby enables measuring solution sets of single-objective problems, provided a distance metric exists.
Recently, we presented a new practical method for upward crossing mini- mization [8], which clearly outperformed existing approaches for drawing hier- archical graphs in that respect. The outcome of this method is an ...
详细信息
Recently, we presented a new practical method for upward crossing mini- mization [8], which clearly outperformed existing approaches for drawing hier- archical graphs in that respect. The outcome of this method is an upward planar representation (UPR), a planarly embedded graph in which crossings are repre- sented by dummy vertices. However, straight-forward approaches for drawing such UPRs lead to quite unsatisfactory results. In this paper, we present a new algorithm for drawing UPRs that greatly improves the layout quality, leading to good hierarchal drawings with few crossings. We analyze its performance on well-known benchmark graphs and compare it with alternative approaches.
Two related problems with TORCS car racing competition controllers are approached here. At first, we demonstrate how to handle the 10% artificial sensor noise that made proper track segment recognition quite difficult...
详细信息
Two related problems with TORCS car racing competition controllers are approached here. At first, we demonstrate how to handle the 10% artificial sensor noise that made proper track segment recognition quite difficult for some controllers in 2010 (when the noise was introduced). This is successfully dealt with by a combination of averaging and regression. The presented solution copes well with the natural antagonism between accuracy and the produced time lag, meaning that controllers are enabled to use full sensor information despite noise. Secondly, we suggest a solution for the problem of selecting a minimal set of controller parameter configurations suitable for several different tracks by applying principles of multi-objective optimization. While a full multi-objective approach is unfeasible here, we investigate the conflict potential between objectives (in this case tracks) in order to remove the ones that are less problematic. Naturally, the second problem is much more interesting but can only be tackled if the first one is resolved.
In less-than-truckload terminals arriving trucks have to be allocated to a gate and to a time slot for unloading. The allocation to a specific gate results in different transportation volumes for the forklift trucks i...
详细信息
In less-than-truckload terminals arriving trucks have to be allocated to a gate and to a time slot for unloading. The allocation to a specific gate results in different transportation volumes for the forklift trucks inside of the terminal, depending on the destinations of the truck’s loads. While minimizing these transports the time for trucks waiting to be ordered to a gate also has to be minimized. For the first time this problem has been tackled as a 2-objective optimization problem and was solved by an (1+1)-evolution strategy. We developed a model which is derived from real freight forwarder’s data and represents a small company’s terminal on an average workday.
Popular games often have a high-quality graphic design but quite simple-minded non player characters (NPC). Recently, Computational Intelligence (CI) methods have been discovered as suitable methods to revive NPC, mak...
详细信息
Popular games often have a high-quality graphic design but quite simple-minded non player characters (NPC). Recently, Computational Intelligence (CI) methods have been discovered as suitable methods to revive NPC, making games more interesting, challenging, and funny. We present a fairly large study of human players on the simple arcade game Pac-Man, controlling the ghosts behaviors by simple strategies, neural networks or evolutionary algorithms. The playerpsilas fun is of course a subjective experience, but we presume that it is related to the psychological flow concept. We deal with the question whether flow is a more reliable measure than asking human players directly for the fun experienced during the game. In order to detect flow, we introduce a measure based on the interaction time fraction between the human-controlled Pac-Man and the ghosts, and compare the outcome to the results of a fun measure suggested by Yannakakis and Hallam [1].
Probably Approximately Correct (i.e., PAC) learning is a core concept of sample complexity theory, and efficient PAC learnability is often seen as a natural counterpart to the class P in classical computational comple...
详细信息
Parallel evolutionary algorithms have been used for solving multiobjective optimization problems. The aim is to find or approximate the Pareto optimal set in a reasonable time. In this work, we present a new approach ...
详细信息
ISBN:
(数字)9781728169293
ISBN:
(纸本)9781728169309
Parallel evolutionary algorithms have been used for solving multiobjective optimization problems. The aim is to find or approximate the Pareto optimal set in a reasonable time. In this work, we present a new approach that divides the objective search-space into different partitions and assigns each processor its corresponding partition. Each processor will try to find the set of solutions for its partition only. The sub-Pareto fronts will be combined later and the parallelisation approach is based on a mutli-start approach by having independent algorithm on every processor with its own starting points. Experimental results on well known test cases showed that the proposed method outperformed several state-of-the-art evolutionary algorithms regarding convergence to the true Pareto front and gave very competitive results when considering the hypervolume metric. Also, superlinear speedup results were achieved for all test functions.
Assembling suitable groups of fighting units to combat incoming enemy groups is a tactical necessity in real-time strategy (RTS) games. Furthermore it heavily influences future strategic decisions like unit building. ...
详细信息
Assembling suitable groups of fighting units to combat incoming enemy groups is a tactical necessity in real-time strategy (RTS) games. Furthermore it heavily influences future strategic decisions like unit building. Here, we demonstrate how to efficiently (offline) solve the problem of finding matches for the current enemy group(s) based on self-organizing maps (SOMs), powered by a simple evolutionary algorithm. The concept is implemented and thoroughly experimentally investigated in the RTS game Glest. We show that the offline learning is reliable and can be sped up considerably by employing a very simple substitute objective function instead of game simulations, making it a nearly universal, simple, and transparent technique.
The creation of interesting opponents for human players in computer games is an interesting and challenging task. In contrast to up-to-date computer games, e.g. real time strategy games, learning of non-player-charact...
详细信息
The creation of interesting opponents for human players in computer games is an interesting and challenging task. In contrast to up-to-date computer games, e.g. real time strategy games, learning of non-player-character strategies for older games seems to be easier and not that time-consuming. This way, older games, like the famous arcade game Pac-Man, serve as a test bed for the creation of strategies that are fun to play against. The paper at hand uses computational intelligence methods to accomplish this challenge, namely evolutionary algorithms (EA) and artificial neural networks (ANN). The latter are trained on a model of the game whereas the EA learn good behavior by playing. The performance of these two approaches is compared on the original Pac-Man level as well as on other maps with different properties to test the ability of generalizing the learned strategies.
暂无评论