Difference between Convolution Neural Network & ANN

hi Everyone! I have some confusion
In CNN specific neurons takes specific input data like we have input x1,x2, so in CNN 1st neuron may take 30%, 2nd 40%, 3rd may b 30% of x1 and similarly other three neurons will take data in same manner apart from fix percentage. so, is not similar to ANN where, we can split x1 and x2 in more arrays and increase the number of neurons in input layer rather than just use 2 neurons.

Hi @Muhammad_Azhar_Ghaur,

Is x1 a sample, a feature, or an input layer? What makes a neuron selective on x1? How can we divide a x1 into three groups (30%, 40%, 30%)? Can you give a simple architecture of the CNN and a simple architecture of the ANN to aid the discussion?

I may have more questions for understanding the picture in your mind but I will stop for now.

Raymond

hello, like in simple ANN we have just 1 neuron in input layer, two neuron in output layer and 1 neuron at output layer. see we have x1 an array of 1x1000. it is the simple architecture of ANN.
In video respected professor said that In CNN input neuron not to see all the data but just a small piece of data, other pieces of data see by the other neuron. right.
so my question same thing we can do in ANN, where we divide our data in 3 arrays and increase number of neurons at input layer from 1 to 3. is it possible?
so how we can differentialize ANN and CNN.

Can you share the name of the video, and the timestamp (which minute and which second of the video) Professor said what you quoted? I would like to watch that part and I think it will help me understand your question better.

Raymond

1 Like

Sorry I was wrong, I recall the lecture, inputs is about hidden layer, it is not about input layer. In CNN, single neuron doesn’t need to take input from all neurons, rather to take input from specific neuron.

Hello @Muhammad_Azhar_Ghaur, it’s totally fine :slight_smile: However, I still don’t get the idea behind a neuron being selective on its inputs, and if you still want to talk about this, I will really need the video name and the timestamp so that I can see what you see.

Cheers,
Raymond