Echo state network, why does it matter the order of the data when memorizing a sequence?
1 次查看(过去 30 天)
显示 更早的评论
I've got an echo state network an using it to memorize a sequence of binary vectors. To memorize it I am offsetting the data by one on the output. However the number of nodes and iterations changes massively depending on which way I offset the data. For example if I offset the output as follows it converges really quickly with not too many nodes:
Input = [ho,h1,h2,.....hn]
output = [hn,ho,h1,.....hn-1]
but if i set the output to
[h1,h2,h3,......ho] then it takes too long to converge, why is this?
Thanks in advance
0 个评论
回答(0 个)
另请参阅
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!