I am planning to implement Convolutional Neural Network model. Is CNN supports only image classification applications? Can I apply CNN modelling on other regular dataset for classification or prediction?
Among the numerous possible general explanations, convolution can be thought as “looking at a function’s surroundings to make better/accurate predictions of its outcome.”
Below are two concrete examples.
If we want to predict the speed of a bullet at a given time (t) considering the speeds of the bullet at a few time intervals before (t-1, t-2, t-3) makes the prediction better.
Similarly, instead of looking at a whole picture to see certain structures, it will be better to look at small parts of the picture.
Hence, CNNs can be used at any place where there’s a location relationship among the features. The dimensionality of the problem will change according to the problem. In the first example it’s 1-D whereas in the second example it’s 2-D.
Natural language processing tasks, time series tasks, and proteomics are some of the areas where CNN can be used.
Yes, CNN can also use for text classification. This link is one of the paper example : http://www.aaai.org/ocs/index.php/AAAI/AAAI15/paper/download/9745/9552
I think your question is not properly understood by some of the above repliers. Aren't texts also images? Are we not taking images for the dealing of diseases like cancer? You may make your question more clear.