Predicted values are all the same in CNN regression
3 次查看(过去 30 天)
显示 更早的评论
Hello, I am new to CNN, and I am trying to use regression using CNN on 1D data. My data has 105 samples, each with 458 data points. However, whenever I train my network and use that to predict values, all the prediction values turn out to be the same. I have tried using different solvers, filter sizes, or pooling layers, but none of it seems to produce any difference. I am including my code here:
XTrain = xlsread('1stderv_preprocessed_training_sample_spectra.xlsx');
height = 458;
width = 1;
channels = 1;
samples = 105;
CNN_TrainingData = reshape(XTrain,[height, width, channels, samples]);
YTrain = xlsread('trng ref data_caffeine-35x3_105 sample.xls');
CNN_TrainingLabels = YTrain;
layers = [
imageInputLayer([height, width, channels])
convolution2dLayer([5 1],100, 'stride',5)
batchNormalizationLayer
reluLayer
averagePooling2dLayer([50 1],'Stride',2)
dropoutLayer(0.3)
fullyConnectedLayer(100)
fullyConnectedLayer(50)
fullyConnectedLayer(20)
fullyConnectedLayer(5)
fullyConnectedLayer(1)
regressionLayer];
miniBatchSize = 5;
validationFrequency = floor(numel(YTrain)/miniBatchSize);
options = trainingOptions('sgdm', ...
'MiniBatchSize',miniBatchSize, ...
'MaxEpochs',50, ...
'InitialLearnRate',1e-4, ...
'LearnRateSchedule','piecewise', ...
'LearnRateDropFactor',0.1, ...
'LearnRateDropPeriod',45);
net = trainNetwork(CNN_TrainingData,CNN_TrainingLabels,layers,options);
YPrediction = predict(net,CNN_TestingData);
Please help me out, I'm at my wits' end.
0 个评论
回答(0 个)
另请参阅
类别
在 Help Center 和 File Exchange 中查找有关 Image Data Workflows 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!