Reset the network state between sequence predictions.

Load JapaneseVowelsNet, a pretrained long short-term memory (LSTM) network trained on the Japanese Vowels data set as described in [1] and [2]. This network was trained on the sequences sorted by sequence length with a mini-batch size of 27.

load JapaneseVowelsNet

View the network architecture.

net.Layers

ans = 
  5x1 Layer array with layers:

     1   'sequenceinput'   Sequence Input          Sequence input with 12 dimensions
     2   'lstm'            LSTM                    LSTM with 100 hidden units
     3   'fc'              Fully Connected         9 fully connected layer
     4   'softmax'         Softmax                 softmax
     5   'classoutput'     Classification Output   crossentropyex with '1' and 8 other classes

Load the test data.

[XTest,YTest] = japaneseVowelsTestData;

Classify a sequence and update the network state. For reproducibility, set rng to 'shuffle'.

rng('shuffle')
X = XTest{94};
[net,label] = classifyAndUpdateState(net,X);
label

label = categorical
     3 

Classify another sequence using the updated network.

X = XTest{1};
label = classify(net,X)

label = categorical
     7 

Compare the final prediction with the true label.

trueLabel = YTest(1)

trueLabel = categorical
     1 

The updated state of the network may have negatively influenced the classification. Reset the network state and predict on the sequence again.

net = resetState(net);
label = classify(net,XTest{1})

label = categorical
     1