一种优化算法
[英]An optimization algorithm
问题描述
Condition 1 : I have 5800 one dimensional signals, where I choose 30 of them and make a picture using those 30 signals that I picked, via a process called X (Here, this process doesn't effect the question).
条件 1 ：我有 5800 个一维信号，我选择其中的 30 个并使用我选择的这 30 个信号通过一个称为X的过程制作图片（这里，这个过程不会影响问题）。
Condition 2 : I have a model that can evaluate a picture and return a number between 0 and 1, where 1 means the picture is very desired and 0 meaning it is not accurate at all.条件 2 ：我有一个 model 可以评估图片并返回 0 到 1 之间的数字，其中 1 表示图片非常需要，0 表示完全不准确。
Condition 3 : Now I want to choose 30 signals from those 5800 original signals, in such a way that if I evaluate its corresponding image with the model in Condition 1 , it is close to 1 as possible.条件3 ：现在我想从这5800个原始信号中选择30个信号，这样如果我在条件1中用model评估它的对应图像，它就尽可能接近1。
To be more precise, choosing all of the 30 out of 5800 possibilities is very computational expensive.更准确地说，从 5800 种可能性中选择所有 30 种可能性是非常昂贵的。 I'm struggling to come up with an algorithm to kind of replace only some of those 30 signals that I originally -randomly- picked, based on the evaluation of the model in Condition 1 .根据对条件 1中 model 的评估，我正在努力想出一种算法来仅替换我最初随机选择的 30 个信号中的一些。
1 个解决方案
解决方案1
1 已采纳 2020-11-27 13:51:40
Try to use genetic algorithms:尝试使用遗传算法：
- calculate a set of 30 signals that cover all 5800 signals.计算一组涵盖所有 5800 个信号的 30 个信号。
- Mix best solutions.混合最佳解决方案。
声明:本站的技术帖子网页，遵循CC BY-SA 4.0协议，如果您需要转载，请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.