Jan-08-2017, 08:44 PM
Hi all,
I've been working on making a python script for a Hopfield Network for the resolution of the shortest path problem, and I have found no success until now. It has been weeks or so that I am banging my head on the keyboard and no solution yet. I have based my efforts until now on two papers I've read - h ttps://www.justbeamit.com/5g3g4 - in which both had simulations of the network through software instead of hardware. Although I have rewritten my code almost a thousand times, I can't find what is going wrong with it; the outputs of the neurons are always far from anything reasonable and correct.
To have some background, just read one of the articles, they aren't big; essentially, I want to find the shortest distance between any two given nodes in a directed graph (not between every node, as in the travelling salesman problem). That will happen when the network minimizes the energy function, like any other Hopfield ANN.
Source:h ttp://pastebin.com/S2R8Er4m
If any of you who has some experience on neural networks can shed some light upon this, I'd be really thankful. I am developing a project that needs it, and my time is almost up.
Thanks in advance.
I've been working on making a python script for a Hopfield Network for the resolution of the shortest path problem, and I have found no success until now. It has been weeks or so that I am banging my head on the keyboard and no solution yet. I have based my efforts until now on two papers I've read - h ttps://www.justbeamit.com/5g3g4 - in which both had simulations of the network through software instead of hardware. Although I have rewritten my code almost a thousand times, I can't find what is going wrong with it; the outputs of the neurons are always far from anything reasonable and correct.
To have some background, just read one of the articles, they aren't big; essentially, I want to find the shortest distance between any two given nodes in a directed graph (not between every node, as in the travelling salesman problem). That will happen when the network minimizes the energy function, like any other Hopfield ANN.
Source:h ttp://pastebin.com/S2R8Er4m
If any of you who has some experience on neural networks can shed some light upon this, I'd be really thankful. I am developing a project that needs it, and my time is almost up.
Thanks in advance.