hi friends,
i have selected my final year project as "Handwritten character segmentation using neural n/w" So u guys if have any idea of neural n/w implementation in c++" i will be thankful to u.
u can mail me at [email protected]
Printable View
hi friends,
i have selected my final year project as "Handwritten character segmentation using neural n/w" So u guys if have any idea of neural n/w implementation in c++" i will be thankful to u.
u can mail me at [email protected]
I did work on something like this.
The handwriting was stored as the picture of course, but also a list of parameters giving the number of backwards movements, upper movements (used for l, k, h, b, f ...), down movements (used for g, j, q ...), etc...
I don't remember if we used neural networks there...
Marina
Some resources I have seen for c++ are:
- Some ugly c++ classes -- not well written, but gets the idea across
- Neural Network Objects -- this is a pdf explaining an implementation
- XOR Net back propagation classes -- only bp, but its a start...
Also, there are two good books I have read. Practical Neural Network Recipes in C++ by Timothy Masters and Object-Oriented Neural Networks in C++ by Joey Rogers. These are really useful references. However, I might add that much of current research on neural net algorithm at the academic level is done in Lisp, as it is a more natural language to express these data structures and their evolution.