Neural networks known for massive parallelism and for pattern recognition and matching. Based on these, can we use the neural networks for matching strings in large texts.
If so, how to convert character to integer as neural inputs, what kind of NN , which training algorithm should be used.
Hope this helps..
http://link.springer.com/chapter/10.1007%2F978-3-540-74272-2_97
http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=718523&queryText%3Dstring+matching+neural+networks
This might be interesting, too:
http://www.ncbi.nlm.nih.gov/pubmed/11709811
I think yes, but problem is how to splits words from string and split each word to character, finally extract features of characters which will be the input to ANN.
If yes then how about converting each letter into its ascii binary value and input into ANN
According to Ling (2006), each graffiti pattern will be divided into a distance segment represented by ten points. Firstly, the five points (Xi,Yi), i=1,3,5,7 and 9 are converted to five numbers (ϭ) by using (ϭi=Xi*Xmax+Yi) while the others are transferred by the formula (ϭi=Yi*Ymax+Xi). Consequently, the feature extraction operation results in ten numbers for each pattern. In the other words, we have 10 inputs for the next level (the neural network).
Forward multilayer neural network with back propagation as supervised algorithm
- Badges
- Science topic