神经网络中的竞争性学习
我正在玩一些神经网络模拟。我想让两个神经网络共享输入和输出节点（其他节点是不同的并且是两条不同路线的一部分）来竞争。有什么我应该看的示例/标准算法吗？这个问题适合本网站吗？
现在我正在使用阈值来区分两条路线，但我想同时激活它们并让它们通过使用遍历每条路线所花费的时间来决定（“这个模拟对于我们两个人来说不够大”）因素。
更新：
谢谢 Gacek 和 Amro，
Gacek - 我不是机器学习的学生.../这是我第一次实现神经网络的经验...那么“质量系数”是什么意思？
Amro - 抱歉...我不应该在问题中使用“竞争性学习”...会尝试改变这一点，也许会添加一些数据。 我想做的是建立两个网络，它们共享输入并产生相同的输出（不是定性的）......它们实际上连接到相同的输出神经元。 也许您可以将其视为具有两条路线或路径的单个网络，我试图让该事物根据信息沿着两条路线从刺激节点传播到响应神经元所需的时间做出选择。
I am playing with some neural network simulations. I'd like to get two neural networks sharing the input and output nodes (with other nodes being distinct and part of two different routes) to compete. Are there any examples/standard algorithms I should look at? Is this an appropriate question for this site?
Right now I'm using a threshold to distinguish between two routes, but I want to activate them simultaneously and let them decide ('this simulation isn't big enough for the two of us') by using time taken to traverse each route as the factor.
Update:
Thanks Gacek and Amro,
Gacek - I am not a machine learning student.../and this is my first experience with implementing neural networks...so what do you mean by 'quality coefficients'?
Amro - sorry...I shouldn't have ujsed 'competitive learning' in the question...will try to change that and maybe add some data.
What I am trying to do is set up two networks which share inputs and produce the same output (not qualitatively)...they are literally connected to the same output neuron.
Maybe you could look at it as a single network with two routes or pathways, and I am trying to make the thing make a choice based on the time it takes information to travel from stimulus node to response neuron along the two routes.
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
AFAIK，“竞争学习”这个词指的是一种特定类型的网络，其中神经元竞争对输入做出响应，获胜的神经元的输出为 1，所有其他神经元的输出为 0。
根据我的理解（没有看到任何代码），您所描述的就像只是在相同的训练数据上训练两个具有相同结构（但初始化不同）的人工神经网络，并最终选择最好的一个（就性能而言）。
AFAIK, the word Competitive Learning refers to a specific type of networks where neurons compete to respond to an input, with the winning neuron's output being 1, and all others zeros.
From what I understood (without seeing any code), what you describe is rather like just training two ANN's of the same structure (but initialized differently) on the same training data, and eventually picking the best one (in terms of performance).