写一个输入是4*10000的resnet的matlab程序

好的，下面是一个输入为4*10000的ResNet的MATLAB程序示例：

classdef ResNetBlock < nnet.layer.Layer
    properties
        Convolution1DLayer
        BatchNormalizationLayer
        ReLULayer
        Convolution1DLayer_1
        BatchNormalizationLayer_1
        ShortcutConnection
    end
    
    methods
        function this = ResNetBlock(in_channels, out_channels, stride)
            name = "resnet_block_" + in_channels + "_" + out_channels + "_" + stride;
            this.Name = name;

            this.Convolution1DLayer = convolution1dLayer(3, out_channels, "Stride", stride, "Padding", 1, "BiasLearnRateFactor", 0);
            this.BatchNormalizationLayer = batchNormalizationLayer;
            this.ReLULayer = reluLayer;
            this.Convolution1DLayer_1 = convolution1dLayer(3, out_channels, "Padding", 1, "BiasLearnRateFactor", 0);
            this.BatchNormalizationLayer_1 = batchNormalizationLayer;

            if stride ~= 1 || in_channels ~= out_channels
                this.ShortcutConnection = convolution1dLayer(1, out_channels, "Stride", stride, "BiasLearnRateFactor", 0);
                this.ShortcutConnection.Weights = randn([1 1 in_channels out_channels])*sqrt(2/(in_channels*out_channels));
                this.ShortcutConnection.Bias = zeros([1 1 out_channels]);
            else
                this.ShortcutConnection = [];
            end
        end

        function Z = predict(this, X)
            residual = X;

            Z = this.Convolution1DLayer(X);
            Z = this.BatchNormalizationLayer(Z);
            Z = this.ReLULayer(Z);

            Z = this.Convolution1DLayer_1(Z);
            Z = this.BatchNormalizationLayer_1(Z);

            if ~isempty(this.ShortcutConnection)
                residual = this.ShortcutConnection(X);
            end

            Z = Z + residual;
            Z = this.ReLULayer(Z);
        end
    end
end

classdef ResNet < nnet.layer.Layer
    properties
        Convolution1DLayer
        BatchNormalizationLayer
        ReLULayer
        MaxPooling1DLayer
        ResNetBlockLayer1
        ResNetBlockLayer2
        ResNetBlockLayer3
        ResNetBlockLayer4
        AveragePooling1DLayer
        FullyConnectedLayer
    end
    
    methods
        function this = ResNet(numClasses)
            name = "resnet_" + numClasses;
            this.Name = name;

            this.Convolution1DLayer = convolution1dLayer(7, 64, "Stride", 2, "Padding", 3, "BiasLearnRateFactor", 0);
            this.BatchNormalizationLayer = batchNormalizationLayer;
            this.ReLULayer = reluLayer;
            this.MaxPooling1DLayer = maxPooling1dLayer(3, "Stride", 2, "Padding", 1);

            this.ResNetBlockLayer1 = [
                ResNetBlock(4, 64, 1)
                ResNetBlock(64, 64, 1)
                ResNetBlock(64, 64, 1)
            ];
            this.ResNetBlockLayer2 = [
                ResNetBlock(64, 128, 2)
                ResNetBlock(128, 128, 1)
                ResNetBlock(128, 128, 1)
                ResNetBlock(128, 128, 1)
            ];
            this.ResNetBlockLayer3 = [
                ResNetBlock(128, 256, 2)
                ResNetBlock(256, 256, 1)
                ResNetBlock(256, 256, 1)
                ResNetBlock(256, 256, 1)
                ResNetBlock(256, 256, 1)
                ResNetBlock(256, 256, 1)
            ];
            this.ResNetBlockLayer4 = [
                ResNetBlock(256, 512, 2)
                ResNetBlock(512, 512, 1)
                ResNetBlock(512, 512, 1)
            ];

            this.AveragePooling1DLayer = averagePooling1dLayer(10000);
            this.FullyConnectedLayer = fullyConnectedLayer(numClasses);
        end

        function Z = predict(this, X)
            Z = this.Convolution1DLayer(X);
            Z = this.BatchNormalizationLayer(Z);
            Z = this.ReLULayer(Z);
            Z = this.MaxPooling1DLayer(Z);

            Z = this.ResNetBlockLayer1.predict(Z);
            Z = this.ResNetBlockLayer2.predict(Z);
            Z = this.ResNetBlockLayer3.predict(Z);
            Z = this.ResNetBlockLayer4.predict(Z);

            Z = this.AveragePooling1DLayer(Z);
            Z = this.FullyConnectedLayer(Z);
        end
    end
end

numClasses = 10;
inputSize = [4 10000 1];
lgraph = layerGraph();
lgraph = addLayers(lgraph, [
    imageInputLayer(inputSize, "Name", "input")
    ResNet(numClasses)
    classificationLayer()
]);
lgraph = connectLayers(lgraph, "input", "resnet_10");

这里同样提供了一个ResNet50的程序，你可以根据自己的需求进行修改。在这个程序中，我们使用了MATLAB中的深度学习工具箱来定义ResNetBlock和ResNet两个类，其中ResNet由多个ResNetBlock组成的深度神经网络。在这个模型中，每个输入的数据是一个大小为4*10000的三维向量。最后，我们使用layerGraph函数将这些层组成一个网络。