Please enable JavaScript.

Coggle requires JavaScript to display documents.

Neural Networks, Artificial Neural Networks, inspired_by, Challenges …

- - - - Hebbian Rule
        
        A becomes stronger due to it's connection to B
      - Hebbian Learning
        
        increased efficiency
        
        Erkki Oja
        
        OJas Learning
        
        Increases in weight, decreases weight elsewhere
  - - - Father of deep learning
        
        Perceptron
        
        multi-perceptron neuron
        
        an ANN
        
        weighted connections
        
        conceptual limitations
        
        challenges :warning:
        
        technological limitations
        
        single layer
  - - - efficient learning
    - - self driving cars
  - - - godfather of AI
      - multi-layered NN
        
        deep NN
      - challenges
        
        not enough data
        
        not enough compute power
      - 2006
        
        internet: lots of data
        
        World catches up :check:
        
        super fast chips
  - - - handwriting recognition
- - - - sigmund
      - logistic
      - normalise output
        
        1 or 0
      - relus units
        
        0 or linear
  - - - learning
        
        changing the connection weights