我正在尝试使用交叉熵实现RNN。以下是我的代码:
net = layrecnet(1:2,10);
net.performFcn = 'crossentropy';
net.performParam.regularization = 0.1;
net.performParam.normalization = 'none';
[Xs,Xi,Ai,Ts] = preparets(net, featureMatrix, labels);
net = train(net,Xs,Ts,Xi,Ai);
% view(net)
Y = net(Xs,Xi,Ai);
perf = perform(net,Y,Ts);
performParam
来自Matlab的官方doc。然而,在我执行之后,我收到了警告:
Warning: Performance function replaced with squared
error performance.
> In trainlm>formatNet (line 155)
In trainlm (line 65)
In nntraining.setup (line 14)
In network/train (line 335)
即使我执行feedforwardnet,我也得到了相同的警告。以下是我的代码。
[x,t] = simplefit_dataset;
net = feedforwardnet(10);
net.performFcn = 'crossentropy';
net.performParam.regularization = 0.1;
net.performParam.normalization = 'none';
net = train(net,x,t);
view(net)
y = net(x);
perf = perform(net,y,t);
那我怎么能在代码中使用交叉熵呢?
答案 0 :(得分:1)
问题是trainlm
仅适用于使用Jacobian Matrix
的损失函数,如文档中所述:
此函数使用雅可比行列式进行计算,假设为 性能是平方误差的平均值或平方和。因此,网络 使用此功能训练必须使用mse或sse性能 功能
一种解决方案是使用其他培训算法,例如trainrp
或trainscg
。以下作品:
[x,t] = simplefit_dataset;
net = feedforwardnet(10);
net.performFcn = 'crossentropy';
net.performParam.regularization = 0.1;
net.performParam.normalization = 'none';
net.trainFcn = 'trainrp';
net = train(net,x,t);
view(net)
y = net(x);
perf = perform(net,y,t);