⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 generate_internal_weights.m

📁 回声状态神经网络(ESN)是一种性能优异的递归神经网络
💻 M
字号:
function internalWeights = generate_internal_weights(nInternalUnits, ...                                                  connectivity)% GENERATE_INTERNAL_WEIGHTS creates a random reservoir for an ESN%  % inputs:% nInternalUnits = the number of internal units in the ESN% connectivity \in [0,1], says how many weights should be non-zero %% output:% internalWeights = matrix of size nInternalUnits x nInternalUnits% internalWeights(i,j) = value of weight(synapse) from unit i to unit j% internalWeights(i,j) might be different from internalWeights(j,i)%% Created April 30, 2006, D. Popovici% Copyright: Fraunhofer IAIS 2006 / Patent pending% Revision 1, Feb 23, 2007, H. Jaeger% Revision 2, March 10, 2007, H. Jaeger (replaced eigs by myeigs)success = 0 ;                                               while success == 0    % following block might fail, thus we repeat until we obtain a valid    % internalWeights matrix    try,        internalWeights = sprand(nInternalUnits, nInternalUnits, connectivity);        internalWeights(internalWeights ~= 0) = internalWeights(internalWeights ~= 0)  - 0.5;        maxVal = max(abs(myeigs(internalWeights,1)));        internalWeights = internalWeights/maxVal;        success = 1 ;    catch,        success = 0 ;     endend

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -