How does a mathmatical model of a neuron look like?
Caclulate:
Why is this basic formula called a linear classifier?
How does the training algorithm in general work like?
How does the number of neurons influence the weight matrix?
Number of rows
How does matrix multiplication work?
Which is the historically most used activation function?
Sigmoid
Which other activation functions are there?
How to calc lernable parameters in a 2 layer ANN?
What happens when we build a neural network without activiation function?
Linear Classifier
Whats the problem deriving the loss function on paper? How to solve it?
Lots of calculus
When we change loss func. we need to start from scratch
Not feasible for complex models
Computational Graph —> Backprop.
Whats the upstream gradient?
Whats the downstream gradient?
and how to calc?
Whats the local gradient?
Zuletzt geändertvor 2 Monaten
