ltb火箭
Two ideas from computer science have intrigued me for a long time: neural networks and evolutionary algorithms. Neural networks are of course everywhere right now: essentially a computer runs a “brain” consisting of simulated nerve cells. They are being used to perform science-fiction-sounding tasks like besting human grand masters in games¹, creating natural looking landscapes from sketches² or even photos of people that don’t exist³. The concept behind evolutionary algorithms is possibly even more fascinating: a population of computer programs is tested against each other in a simulation of natural selection, and over time the descendants of the “survivors” become increasingly better at their required task. Although they are a bit less all-present than neural networks evolutionary algorithms have their applications, and sometimes both are used together.
很长一段时间以来，计算机科学的两个想法吸引了我：神经网络和进化算法。 神经网络当然现在无处不在：本质上，一台计算机运行由模拟神经细胞组成的“大脑”。 它们被用来执行听起来像科幻小说般的任务，例如在游戏中击败人类大师，用草图2甚至是不存在的人的照片3创造出自然的风景。 进化算法背后的概念可能更令人着迷：在自然选择的模拟中，对计算机程序进行了相互测试，随着时间的流逝，“幸存者”的后代在其所需任务上变得越来越出色。 尽管它们比神经网络的通用性要差一些，但进化算法却有其应用，有时两者一起使用。
Neural networks and evolutionary algorithms in their simplest forms can be programmed in surprisingly few lines of fairly straightforward code. Both are notorious for requiring a lot of computing power, but today’s computers should be plenty powerful enough to deal with a small problem even if the program runs in a web browser (or so I assumed). If I had an interesting problem to solve I could make an entertaining personal project that shouldn’t take lots of coding to complete, or so I hoped!
最简单形式的神经网络和进化算法可以用几行相当简单的代码进行编程。 两者都因需要大量的计算能力而臭名昭著，但是，即使程序在Web浏览器中运行(或者我假设是这样)，当今的计算机也应具有足够强大的功能来处理一个小问题。 如果我有一个有趣的问题要解决，那么我可以做一个有趣的个人项目，而无需花费很多代码来完成，所以我希望如此！
版权声明：本文内容由互联网用户自发贡献，该文观点仅代表作者本人。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容，请联系我们，一经查实，本站将立刻删除。
如需转载请保留出处：https://51itzy.com/kjqy/18588.html