Understanding LSTM Sequence to Sequence Mathworks example

Hi, in this example for a Sequence to Sequence regression LSTM there's a call to a function named "prepareDataTrain"

here's the function:

function [XTrain,YTrain] = prepareDataTrain(filenamePredictors)

dataTrain = dlmread(filenamePredictors);

numObservations = max(dataTrain(:,1));

XTrain = cell(numObservations,1);
YTrain = cell(numObservations,1);
for i = 1:numObservations
    idx = dataTrain(:,1) == i;
    
    X = dataTrain(idx,3:end)';
    XTrain{i} = X;
    
    timeSteps = dataTrain(idx,2)';
    Y = fliplr(timeSteps);
    YTrain{i} = Y;
end

end

the dataTrain table looks like this:

1 1 -0.0007 -0.0004 100.0 518.67 641.82 1589.70 1400.60 14.62 21.61 554.36 2388.06 9046.19 1.30 47.47 521.66 2388.02 8138.62 8.4195 0.03 392 2388 100.00 39.06

1 2 0.0019 -0.0003 100.0 518.67 642.15 1591.82 1403.14 14.62 21.61 553.75 2388.04 9044.07 1.30 47.49 522.28 2388.07 8131.49 8.4318 0.03 392 2388 100.00 39.00

1 3 -0.0043 0.0003 100.0 518.67 642.35 1587.99 1404.20 14.62 21.61 554.26 2388.08 9052.94 1.30 47.27 522.42 2388.03 8133.23 8.4178 0.03 390 2388 100.00 38.95

1 4 0.0007 0.0000 100.0 518.67 642.35 1582.79 1401.87 14.62 21.61 554.45 2388.11 9049.48 1.30 47.13 522.86 2388.08 8133.83 8.3682 0.03 392 2388 100.00 38.88

1 5 -0.0019 -0.0002 100.0 518.67 642.37 1582.85 1406.22 14.62 21.61 554.00 2388.06 9055.15 1.30 47.28 522.19 2388.04 8133.80 8.4294 0.03 393 2388 100.00 38.90

1 6 -0.0043 -0.0001 100.0 518.67 642.10 1584.47 1398.37 14.62 21.61 554.67 2388.02 9049.68 1.30 47.16 521.68 2388.03 8132.85 8.4108 0.03 391 2388 100.00 38.98

1 7 0.0010 0.0001 100.0 518.67 642.48 1592.32 1397.77 14.62 21.61 554.34 2388.02 9059.13 1.30 47.36 522.32 2388.03 8132.32 8.3974 0.03 392 2388 100.00 39.10

1 8 -0.0034 0.0003 100.0 518.67 642.56 1582.96 1400.97 14.62 21.61 553.85 2388.00 9040.80 1.30 47.24 522.47 2388.03 8131.07 8.4076 0.03 391 2388 100.00 38.97

1 9 0.0008 0.0001 100.0 518.67 642.12 1590.98 1394.80 14.62 21.61 553.69 2388.05 9046.46 1.30 47.29 521.79 2388.05 8125.69 8.3728 0.03 392 2388 100.00 39.05

What I don't understand in this function is why is flipping the timeSteps when they are correctly ordered. Could someone please explain this to me?

Thank you for your time.

Answer 1

[TED MOSBY]

Hi,

The function "fliplr" is used to reverse array elements and it is being used here to turn the ascending cycle index [1 2 … N] into the remaining-cycle count [N N-1 … 1]. The example trains a sequence-to-sequence LSTM whose target at every time step is the remaining useful life (RUL) of the engine, not the “current cycle number”. Reversing the vector is the quickest way to obtain RUL from the raw time–step column the dataset provides. fliplr reverses the row vector, giving the exact remaining-cycles-to-failure value for each time step t.

Hope it helped!

[juan pedrosa]

wow thank you for taking your time with such an old question.