CODEWITHSUNDEEP
inputneural-networknormalization
Charinda Nanayakkara
Charinda Nanayakkara

Reputation: 23

How to normalize coordinates inputs for neural networks

I am doing project regrading signature recognition using Neural Networks.So in there I did all feature extraction activities.In result of feature extraction,there are some features which are as set of coordinates(x and y).So I know before put those as neural network input it should be normalize.So I don't know the exact way to normalize coordinates before it give as NN input.Is there are any method to approach that problem? Thank you.

Upvotes: 1

Views: 1681

Answers (1)

jorgenkg
jorgenkg

Reputation: 4275

Serialize 'em!


Given that your cluster algorithm returns data in this manner (or something similar)

# output from        |output coordinates
# cluster algorithm  |
[_ X X _ _]           (0,0) (0,1) (0,2) (0,3) (0,4)
[_ X _ X _]           (1,0) (1,1) (1,2) (1,3) (1,4)
[_ X X _ _]           (2,0) (2,1) (2,2) (2,3) (2,4)
[_ X _ _ _]           (3,0) (3,1) (3,2) (3,3) (3,4)
[_ X _ _ _]           (4,0) (4,1) (4,2) (4,3) (4,4)


Then we serialize it (that is: just concatinate the lines)

# output from cluster algorithm, serialized:
[_ X X _ _ _ X _ X _ _ X X _ _ _ X _ _ _ _ X _ _ _ ]

# the corresponding coordinates for instructional purposes:
[(0,0) (0,1) (0,2) (0,3) (0,4) (1,0) (1,1) (1,2) (1,3) (1,4) (2,0) (2,1) (2,2) (2,3) (2,4) (3,0) (3,1) (3,2) (3,3) (3,4) (4,0) (4,1) (4,2) (4,3) (4,4)]


And feed this into your network!

When it comes to normalization

Your clustering algorithm should "cluster your input images enough", so that the number of inputs is acceptable.

Upvotes: 2

Related Questions