一种结合知识挖掘的进化规划算法

进化规划是一种进化计算方法。进化规划主要使用随机化技术来实现优化过程。与其它进化算法相比,进化规划只有变异操作,而没有交叉操作,因此变异操作的有效性对进化规划算法的成功至关重要。在传统进化规划算法中,进化规划的变异操作具有完全的随机性,这虽然有利于避免局部极值,但却导致较大的计算量。在本文算法中,将知识挖掘技术引入进化规划之中。知识挖掘技术主要用于发现规则,然后利用发现的规则指导变异操作过程,提高变异操作的效率。最终加快了进化规划算法的速度,而计算效果没有明显下降。通过知识挖掘技术的引入,进化规划算法有了初步的智能特性。将本文所介绍的新算法应用于视频编码的运动估计实验中,结果表明新算法有良好的计算速度和计算精确性。

A Evolutionary Programming Algorithms Combined With Knowledge-Digging

Evolutionary programming is a kind of evolutionary algorithms. Evolutionary programming use randomization technology to optimize. But different from other evolutionary algorithms, evolutionary programming only has variation operation and don't has crossing operation. So effective variation operation is very important for success of evolutionary programming. In traditional algorithm, the variation operation of evolutionary programming has full randomization. Because of it, evolutionary programming can get good computation result, but it also reduce computing speed. In proposed algorithm, we use knowledge-digging technology in evolutionary programming. Knowledge-digging technology is used to discovery rules. We do variation operation by rules that are discovered by knowledge-digging technology. So we can improve efficiency of variation operation through these rules, which can lead to the improvement of speed and accuracy of the algorithm. Because of these rules, we make traditional algorithm has initial intellect. We use the proposed algorithm to do motion estimation, and the experimental result show that the new algorithms has quick .reliable and accurate performance.