Discover the world's scientific knowledge
With 135+ million publication pages, 20+ million researchers and 1+ million questions, this is where everyone can access science
You can use AND, OR, NOT, "" and () to specify your search.
I recently started to deal with neural networks. I read about recurrent neural networks and convolutional neural networks. Is there anyone who can tell me which one (or the combination of both) is better suited for a question-answering system (more concretely for text classification)? Does someone have experience in this field or even publications?
Thanks a lot for your help.
When training a CNN,how will channels effect convolutional layer. Some say that when finish conv, it will generate 3 feature maps when the sample is RGB,and then 3 feature maps will add up and turn into 1 feature map.What is the process mean.
I find it difficult in understanding the concept of convolutional neural networks enough so that I can code it in MATLAB. Can someone help me present the idea algorithmically?
What I understood so far: A face embedding is a list of n (for example n=128) measures (numbers) that describe a human face. Feeding triplets to the neural network (two images from class a and an image from class b) - the neural network modifies the measurements so that the ones for the images in class a are more similar than those for image b.
What I don't understand: what are there measures exactly and how does the neural network modify them?