If you have a hammer, everything looks like a nail.

Just because neural networks (NN) are in fashion, they are not the silver bullet for solving all tasks on earth.

Neural networks are inherently fuzzy, which is a very good property for machine learning. However, you'd need to featurize your input to fit in a fixed size input layer, and train the network on given classifications.

Motifs are well-defined subgraph patterns in larger graphs, which are not restricted in size. They are not fuzzy, and you just need to find and count them. There are efficient implementations for certain kinds of motifs that you might be able to use out of the box.

Thus: NNs are not suitable for motif finding.

Thank you , but as we know that by using the neural network especially when use a layered structure will achieve low computational complexity which one of the purposes of any algorithms that finding the motif, so it achieves a benefit in term of decreasing the complexity.

Explain more detailed, Reema, what is the complexity f finding motifs and how would NNs help? I don't see it, for this application.

I mean that the neural network is considered as one of the machine learning techniques  , it is used for finding the motif in DNA and protein sequences , one of its advantages that achieve low computational complexity since it  use of layered structure.I know how the neural network works in general but  How does it work exactly for finding the motif  ? that is what I want to know after I get an answer for my question that will give how it works exactly and why it get low computational complexity .

Rema, NNs are not suitable for exact motif finding./counting as far as I know.Also, Motif counting is usually not solved using machine learning techniques.

Exact Motif matching is NP-hard in the general case, so you won't be able to get lower complexity without paying a price on accuracy, Do you have a reference for using NNs in Motif finding?

thank you

but I have this reference that talk  about the machine learning that is important  for motif finding in short details and show that  the neural network achieve low computational complexity.

maybe the exact motif finding is NP-hard but how can NN achieve low complexity when dealing with the motif finding?

you see this paper:

Hi Reema,

I am an undergraduate student and am by no means an expert on machine learning, but I know of at least one example (DeepBind, http://www.nature.com/nbt/journal/v33/n8/full/nbt.3300.html)  where neural networks worked quite well at finding DNA- and RNA-binding protein motifs.