回归预测 | MATLAB实现CNN-LSTM卷积长短期记忆神经网络多输入多输出

预测效果

基本介绍

MATLAB实现CNN-LSTM卷积长短期记忆神经网络多输入多输出，运行环境Matlab2020及以上。采用特征融合的方法,通过卷积网络提取出浅层特征与深层特征并进行联接,对特征通过卷积进行融合,将获得的矢量信息输入LSTM单元。

程序设计

• 完整程序和数据下载方式1：同等程序价值兑换。
• 完整程序和数据下载方式2(订阅《组合优化》专栏，数据和程序订阅后私信我获取)：MATLAB实现CNN-LSTM多输入多输出

miniBatchSize = 32;

options = trainingOptions("adam", ...
    MaxEpochs=3, ...
    MiniBatchSize=miniBatchSize, ...
    InitialLearnRate=0.005, ...
    LearnRateDropPeriod=2, ...
    LearnRateSchedule="piecewise", ...
    L2Regularization=5e-4, ...
    SequencePaddingDirection="left", ...
    Shuffle="every-epoch", ...
    ValidationFrequency=floor(numel(featuresTrain)/miniBatchSize), ...
    ValidationData={featuresValidation,labelsValidation}, ...
    Verbose=false, ...
    Plots="training-progress");
net = trainNetwork(featuresTrain,labelsTrain,layers,options);    
function features = extractFeatures(X,afe)

features = log(extract(afe,X) + eps);
features = permute(features, [2 3 1]);
features = {features};

end

往期精彩

MATLAB实现RBF径向基神经网络多输入多输出预测
MATLAB实现BP神经网络多输入多输出预测
MATLAB实现DNN神经网络多输入多输出预测

参考资料

[1] https://blog.csdn.net/kjm13182345320/article/details/116377961
[2] https://blog.csdn.net/kjm13182345320/article/details/127931217
[3] https://blog.csdn.net/kjm13182345320/article/details/127894261