Please enable JavaScript.

Coggle requires JavaScript to display documents.

NN concepts - Coggle Diagram

- - - - Detect patterns in images.
      - Filters/Kernels - detect patterns
        
        shapes, objects, edges
        
        Edge detector filter, squares, circles
        
        Relative small matrix
        
        Initialised with random no's.
  - - - Non-linear activn fn.
    - - Down sampling
        
        Features are reduced. 26x26x32 becomes 13x13x32
    - - Why do we need them?
        
        Input is a flattened layerf from teh output of the final pooling layer
        
        13x13x32 becomes flattened, i.e 5408x1
- - - - Each input to each output within its layer
    - - Tyjpe of data : When you work with Image data
    - - With time series data
  - - - Here Dense(5, -> means 5 neurons in that layer. Second Dense(2, -> means 2 neurons which is an output layer.