Neural Sandbox Beta下载

Create, customize, and battle your own neural network AI! Note: This game is still in beta and is being actively developed and improved! Whether it's predicting the stock market, driving a car, or beating the best Go player in the world, there's no doubt that neural networks are the hot new topic in AI. Their inspiration comes from the human brain, and they have the remarkable ability to learn how to overcome even the most traditionally complex tasks for a computer on their own. Perhaps their greatest drawback is how difficult they can be to set up or play around with for yourself. The Neural Sandbox is here to change this! Enter a small cellular world where you play god, and can create and customize neural net controlled creatures for yourself. Train neural nets as you see fit just by hitting a button, no programming required! From there you can test it out in the cell world, and even control a cell in the world yourself. This game is in beta, so there are still many features to be added. If you have any ideas or suggestions, we'd love to hear them! Questions? Comments? Suggestions? Email us at contact@gobtech.io More information about neural nets and how they work in this game: A neural net is a mathematical model inspired by the biological brain. Each one is essentially a web of interconnected neurons, which take in input, modify it and then send out an output to the next neuron. In a large neural net, this can lead to some pretty complex decision making! The neural net cells in this game take in inputs as information such as position and velocity from nearby cells and output a direction for themselves to move. In the training window, thousands of rounds of the game are simulated so that the neural net can "practice" and learn how to play the game based on your specifications for its personality. 创建，定制和战斗的自己的神经网络人工智能！注：本场比赛是仍处于测试阶段，并正在积极建立和完善！ 无论是预测股市，驾驶汽车，或在击败世界上最好的棋手，但毫无疑问的是神经网络是人工智能的热点新课题。他们的灵感来源于人类的大脑，他们必须学会如何克服即使是最复杂的传统任务上自己的一台计算机的非凡能力。 也许，他们最大的缺点是它们是多么困难设立或玩弄自己。神经沙盒在这里改变这个！输入您玩神小蜂窝领域，并且可以创建和自定义神经网络控制的生物自己。训练神经网络，您看合适仅仅通过点击一个按钮，无需编程！从那里，你可以在世界电池测试它，甚至控制在自己世界的小区。 这场比赛是在测试阶段，因此仍然有添加了许多功能。如果您有任何意见或建议，我们很乐意听到他们的声音！ 有问题吗？注释？建议？请发送电子邮件至contact@gobtech.io 关于神经网络以及它们如何在这场比赛中工作的详细信息： 一个神经网络是由生物大脑启发的数学模型。每一个基本上是相互连接的神经元，其采取在输入，修改后再发送出输出到下一个神经元的网。在一个大的神经网络，这可能会导致一些非常复杂的决策！在此游戏中的神经网络的细胞取作为输入信息，例如来自附近小区的位置和速度，并输出自己的方向移动。在训练窗，千轮比赛的进行模拟，使得神经网络可以“实践”，并学习如何发挥基于您的规格为它的个。