|
|
Learning rule based on minimizing the error of a single layer net in which the output units may have any differentiable function for their activation function. (The standard delta rule assumes that the output units have the identity function. For their activation function during the training process.) |
|
|
A vector that represents the patterns placed on a cluster: this may be formed by the neural net during training, as is SOM. Or specified in advance, as in the Hamming net. |
|
Connection link between two neurons with a positive weight : it serves to increase the response of the unit that receives the signal, In contrast. See inhibitory connection. |
|
|
The Euclidean distance, D, between vectors (x1, x2, … xn )
And ( y1, y2, …, yn ) is defined by:
2 n 2
D = ∑ ( xi – yi )
i = 1 |
|
One presentation of each training pattern |
|
|
|
|
<< Start < Prev 1 2 3 4 5 6 7 8 9 10 Next > End >>
|
|
Page 1 of 10 |
