Neural Network for finding characteristic points in data

6 views (last 30 days)
Tillmann on 4 Oct 2016
Commented: Tillmann on 5 Oct 2016
I'm trying to find characteristics within a data vector in matlab. The data is the force of a stamping process. The function should be able to identify process characteristics for different materials, stroke and sample rates and obviously different states of wear. Because of all these boundaries, it's difficult to implement mathematical models for these characteristics. That's why I was wondering if I could use a neural network for identifying these characteristics. I've checked out the matlab help and google for the neural toolbox, but didn't really find what I was looking for.
So here is what I'm looking for:
• The input vector has different length.
• Output should be points of the input data.
• One group of characteristics consists of a starting and an endpoint within the dataset.
Could anyone
1. tell me if it is possible to do this with neural networks?
2. post some helpful links to tutorials / papers / books / articles etc.
thanks for your help in advance. Attached you find some sample data for clarification.

Greg Heath on 4 Oct 2016
You need N pairs of I-dimensional inputs and corresponding O-dimensional targets contained in 2 matrices
[ I N ] = size(input)
[ O N ] = size(target)
Then see the regression function documentation:
help fitnet
doc fitnet
Hope this helps
Thank you for formally accepting my answer
Greg

1 Comment

Tillmann on 5 Oct 2016
Hey Greg,
Thanks for your help! I'm having some trouble defining the output matrix. Should the vectors contain the x-values of the data points that I'm looking for, or should the vectors be of length of the input vectors and contain only zeros except for the relevant data points?
Thanks for your help and sorry if this seems trivial to you, for me it's all new :-/