In[]:=
(*Applythetransformationtoeachelementinthedata*)​​transformedDataWithImages="Input"
,"Output"{"1"1,"2"0,"3"0,"4"0,"5"0,"6"0,"7"0,"8"0,"9"0,"10"0,"11"0,"12"0,"13"0,"14"1,"15"0,"16"0,"17"0,"18"0,"19"0,"20"0,"21"0,"22"0,"23"0,"24"0,"25"0,"26"0,"27"0},"Input"
,"Output"{"1"1,"2"0,"3"0,"4"0,"5"0,"6"0,"7"0,"8"0,"9"0,"10"0,"11"0,"12"0,"13"0,"14"0,"15"1,"16"0,"17"0,"18"0,"19"0,"20"0,"21"0,"22"0,"23"0,"24"0,"25"0,"26"0,"27"0}​​​​(*Modifythenetwork*)​​mobilenet=NetModel["MobileNet V2 Trained on ImageNet Competition Data"];​​​​modifiedNet=NetReplacePart[mobilenet,{"Input"->NetEncoder[{"Image",{224,224},ColorSpace->"RGB"}],"Output"->NetDecoder[{"Class",Range[1,27]}],(*Adjustedtomatch27classes*)"reshape"->LinearLayer[27],(*27outputnodesforyourparameters*)"prob_softmax"->SoftmaxLayer[]}];​​​​(*Trainthenetwork*)​​trainedNet=NetTrain[modifiedNet,trainingData];
Out[]=
Input
,Output{11,20,30,40,50,60,70,80,90,100,110,120,130,141,150,160,170,180,190,200,210,220,230,240,250,260,270},Input
,Output{11,20,30,40,50,60,70,80,90,100,110,120,130,140,151,160,170,180,190,200,210,220,230,240,250,260,270}
NetTrain
:Training data should be an association of lists, or a rule from input to output examples.