Hier der untere Punkt
Consider the following 2-layer feedforward neural network with a 1-dimensional input, K outputs
a M hidden units. The kth output is given by :
Your task is to derive a weight update rule for the weights wkj and (Mü)j using the backpropagation
algorithm. We only consider the mean squared error (mse) of a single example (x, y). Always use
the same formalism and notation as discussed in the lecture.
• Define the neuron input ak, the layer input zk and the activation function of the neurons hk
in the output layer.
• Define the neuron input aj , the layer input zj and the activation function of the neurons hj
in the hidden layer.
• Calculate (Ro)k for the output neurons and the resulting weight update (Delta)wkj for the weights
in the output layer.
• Calculate (Ro)j for the hidden neurons and the resulting weight update (DeltaMü)j for the weights in
the hidden layer.
Wir kennen uns ganz und garnicht aus und wissen auch nicht was gemeint ist, das Skriptum ist scheisse und hilft auch nicht weiter.
Wäre dankbar wen jemand helfen könnte.
Consider the following 2-layer feedforward neural network with a 1-dimensional input, K outputs
a M hidden units. The kth output is given by :
Your task is to derive a weight update rule for the weights wkj and (Mü)j using the backpropagation
algorithm. We only consider the mean squared error (mse) of a single example (x, y). Always use
the same formalism and notation as discussed in the lecture.
• Define the neuron input ak, the layer input zk and the activation function of the neurons hk
in the output layer.
• Define the neuron input aj , the layer input zj and the activation function of the neurons hj
in the hidden layer.
• Calculate (Ro)k for the output neurons and the resulting weight update (Delta)wkj for the weights
in the output layer.
• Calculate (Ro)j for the hidden neurons and the resulting weight update (DeltaMü)j for the weights in
the hidden layer.
Wir kennen uns ganz und garnicht aus und wissen auch nicht was gemeint ist, das Skriptum ist scheisse und hilft auch nicht weiter.
Wäre dankbar wen jemand helfen könnte.