matlab深度学习神经网络特征融合遇到了问题

​

本人深度学习小白，想用matlab做两组传感器数据的神经网络特征融合，不知道我的想法对不对，我在工具箱里这么编写了一个基于AlexNet的神经网络模型，两组数据经过相同的卷积层，再通过一个additionLayer连接层绑两组数据连接起来，再通过三个全连接层，最后输出三个分类。
代码是通过深度学习工具箱生成的，如下

lgraph = layerGraph();
tempLayers = [
    imageInputLayer([227 227 3],"Name","data1")
    convolution2dLayer([11 11],96,"Name","conv1_1","BiasLearnRateFactor",2,"Stride",[4 4])
    reluLayer("Name","relu1_1")
    crossChannelNormalizationLayer(5,"Name","norm1_1","K",1)
    maxPooling2dLayer([3 3],"Name","pool1_1","Stride",[2 2])
    groupedConvolution2dLayer([5 5],128,2,"Name","conv1_2","BiasLearnRateFactor",2,"Padding",[2 2 2 2])
    reluLayer("Name","relu1_2")
    crossChannelNormalizationLayer(5,"Name","norm1_2","K",1)
    maxPooling2dLayer([3 3],"Name","pool1_2","Stride",[2 2])
    convolution2dLayer([3 3],384,"Name","conv1_3","BiasLearnRateFactor",2,"Padding",[1 1 1 1])
    reluLayer("Name","relu1_3")
    groupedConvolution2dLayer([3 3],192,2,"Name","conv1_4","BiasLearnRateFactor",2,"Padding",[1 1 1 1])
    reluLayer("Name","relu1_4")
    groupedConvolution2dLayer([3 3],128,2,"Name","conv1_5","BiasLearnRateFactor",2,"Padding",[1 1 1 1])
    reluLayer("Name","relu1_5")
    maxPooling2dLayer([3 3],"Name","pool1_5","Stride",[2 2])];
lgraph = addLayers(lgraph,tempLayers);

tempLayers = [
    imageInputLayer([227 227 3],"Name","data2")
    convolution2dLayer([11 11],96,"Name","conv2_1","BiasLearnRateFactor",2,"Stride",[4 4])
    reluLayer("Name","relu2_1")
    crossChannelNormalizationLayer(5,"Name","norm2_1","K",1)
    maxPooling2dLayer([3 3],"Name","pool2_1","Stride",[2 2])
    groupedConvolution2dLayer([5 5],128,2,"Name","conv2_2","BiasLearnRateFactor",2,"Padding",[2 2 2 2])
    reluLayer("Name","relu2_2")
    crossChannelNormalizationLayer(5,"Name","norm2_2","K",1)
    maxPooling2dLayer([3 3],"Name","pool2_2","Stride",[2 2])
    convolution2dLayer([3 3],384,"Name","conv2_3","BiasLearnRateFactor",2,"Padding",[1 1 1 1])
    reluLayer("Name","relu2_3")
    groupedConvolution2dLayer([3 3],192,2,"Name","conv2_4","BiasLearnRateFactor",2,"Padding",[1 1 1 1])
    reluLayer("Name","relu2_4")
    groupedConvolution2dLayer([3 3],128,2,"Name","conv2_5","BiasLearnRateFactor",2,"Padding",[1 1 1 1])
    reluLayer("Name","relu2_5")
    maxPooling2dLayer([3 3],"Name","pool2_5","Stride",[2 2])];
lgraph = addLayers(lgraph,tempLayers);

tempLayers = [
    additionLayer(2,"Name","addition")
    fullyConnectedLayer(4096,"Name","fc6","BiasLearnRateFactor",2)
    reluLayer("Name","relu6")
    dropoutLayer(0.5,"Name","drop6")
    fullyConnectedLayer(4096,"Name","fc7","BiasLearnRateFactor",2)
    reluLayer("Name","relu7")
    dropoutLayer(0.5,"Name","drop7")
    fullyConnectedLayer(3,"Name","fc8","BiasLearnRateFactor",2)
    softmaxLayer("Name","prob")
    classificationLayer("Name","classoutput")];
lgraph = addLayers(lgraph,tempLayers);

% 清理辅助变量
clear tempLayers;
lgraph = connectLayers(lgraph,"pool1_5","addition/in1");
lgraph = connectLayers(lgraph,"pool2_5","addition/in2");
figure
plot(lgraph);
layers = lgraph.Layers;

画出来的图如下

输入图片读取处理的代码如下

%% 训练数据存储；
allImages1 = imageDatastore("E:\Data\old-new\CWT\1\",...
    'IncludeSubfolders',true,...
    'LabelSource','foldernames');
allImages2= imageDatastore("E:\Data\old-new\CWT\2\",...
    'IncludeSubfolders',true,...
    'LabelSource','foldernames');
%% 数据处理；
rng default
[imgsTrain1,imgsValidation1] = splitEachLabel(allImages1,0.8,'randomized');%按比例拆分 ImageDatastore 标签
disp(['Number of training images: ',num2str(numel(imgsTrain1.Files))]);
disp(['Number of validation images: ',num2str(numel(imgsValidation1.Files))]);

[imgsTrain2,imgsValidation2] = splitEachLabel(allImages2,0.8,'randomized');%按比例拆分 ImageDatastore 标签
disp(['Number of training images: ',num2str(numel(imgsTrain2.Files))]);
disp(['Number of validation images: ',num2str(numel(imgsValidation2.Files))]);

%% 处理输入的图片；
inputSize = layers(1).InputSize;
augimgsTrain1 = augmentedImageDatastore(inputSize(1:2),imgsTrain1); % 调整图像大小以匹配网络输入层
augimgsValidation1 = augmentedImageDatastore(inputSize(1:2),imgsValidation1);

augimgsTrain2 = augmentedImageDatastore(inputSize(1:2),imgsTrain2); % 调整图像大小以匹配网络输入层
augimgsValidation2 = augmentedImageDatastore(inputSize(1:2),imgsValidation2);

augimgsTrain = combine(augimgsTrain1,augimgsTrain2);
% augimgsValidation = combine(augimgsValidation1,augimgsValidation2);

但是训练时会一直报错，ValidationData设置为空因为我不知道怎么把augimgsValidation1 augimgsValidation2两个结构体的验证集放到一块，‘opts =’ 这行会报错误，所以上面也就注掉了

训练的代码如下

%% 设置参数进行训练
rng default
mbSize = 40;
mxEpochs = 40;
ilr = 1e-4;
plt = 'training-progress';

opts = trainingOptions('sgdm',...
    'InitialLearnRate',ilr, ...
    'MaxEpochs',mxEpochs ,...
    'MiniBatchSize',mbSize, ...
    'ValidationData',[],...
    'ExecutionEnvironment','multi-gpu',...
    'Plots',plt);

[trainedAN,info2] = trainNetwork(augimgsTrain,layers,opts);

报的错误如下图

有没有会用matlab做深度学习的大佬，帮我看看这个，万分感谢！

​