i am trying to convert weka multilayer perceptron output to matlab code so that i can use this code for classify the new inputs.

weka output is shown below. i know the theory about MLP algorithm. but dont know how can i convert weka output to matlab code. what is the equation for sigmoid node 0 . out put is shown below. i try in this way .

``node0= node4(weight)+node5(weight)+node6(weight);node0 = 1 / (1 + exp(-node0));if(node0 < threshold)node0 = 0;end``

i did it for all nodes but the output is not correct.

``Sigmoid Node 0Inputs WeightsThreshold 0.9406969488131344Node 4 -2.1903200701922616Node 5 -15.897044602974933Node 6 0.3041300920465129Sigmoid Node 1Inputs WeightsThreshold -0.47458154982115386Node 4 -2.404682382816731Node 5 -1.0175108003621611Node 6 -1.248161956451173Sigmoid Node 2Inputs WeightsThreshold -2.494927166085809Node 4 -2.5916900088001404Node 5 2.6569952694128207Node 6 3.6695765419956974Sigmoid Node 3Inputs WeightsThreshold -2.8679148239112555Node 4 3.548745253783415Node 5 1.8555144089470257Node 6 -3.5768482414822405Sigmoid Node 4Inputs WeightsThreshold 2.5705895442897604Attrib length 19.329302599023382Attrib width 0.46083507698821957Sigmoid Node 5Inputs WeightsThreshold 23.728757673706202Attrib length 26.251966218668777Attrib width 3.4888088475120576Sigmoid Node 6Inputs WeightsThreshold -3.6494152797182733Attrib length 10.163410415907084Attrib width -3.6153885660597807Class pedeInputNode 0Class bikeInputNode 1Class autoInputNode 2Class lkwInputNode 3``

Top