Test Run - Understanding LSTM Cells Using C# RRS feed

  • General discussion

  • Long short-term memory (LSTM) cells can be used to construct LSTM recurrent neural networks. These networks have been responsible for major advances in prediction systems that work with sequence data. James McCaffrey demonstrate how to implement an LSTM cell using C#.

    Read this article in the April 2018 issue of MSDN Magazine

    Monday, April 2, 2018 7:50 PM

All replies

  • Wonderful! I read these articles by McCaffrey and I use code from them for a long time. I think on the basis of these articles, can even create a training course if they are slightly expanded.

    May be James McCafrey collect all these articles in a heap and write a book: Solving optimization problems using .NET or Machine learning with C#?

    Monday, April 23, 2018 1:57 PM
  • Hi, I'm reading the article and I'm a little bit confused.

    This is the first time I read something about LSTM cells and reading the formula, I see this:

    So, we are multiplying a matrix and a vector and the result is a vector and after that we are adding vectors.

    The question is:

    Does MatSig() method transform a vector into a Matrix?

    float[][] ft = MatSig(MatSum(MatProd(Wf, xt),
                     MatProd(Uf, h_prev), bf));

    En la teoría, la teoría es igual a la práctica; en la práctica no

    follow me on twitter.

    Tuesday, April 24, 2018 2:36 PM
  • Can you help us to understand the code for finding the error in the network? 
    Wednesday, July 24, 2019 8:03 PM