Please enable JavaScript.
Coggle requires JavaScript to display documents.
Neural Networks - Coggle Diagram
Neural Networks
VOCABULARY
Input Weight
The input weights to a neuron are adjustable parameters on an artificial neuron. They multiply the values of the inputs before the total input is compared to the bias.
In every neuron, the weight multiplies the input. THis means that when a weight is larger, a neuron's outpur changes rapidly for a small change of an input.
The weights control the width of the transition window of the sigmoid neuron. A very large input weight can make a sigmoid behave like a binary neuron.
-
DECISION BOUNDARY
The decision boundary is a line separating the space of possible inputs to a neuron into two regions. For a binary neuron, inputs in one region activate the neuron, and in the other region they don't. The neuron's particular configuration determines the location of the decision boundary.
Scatter Plot
A scatterplot shows how two measurements are related across a sample. A scatterplot is constructed by treating two measurements of a single sample as an ordered pair and then plotting each point in a coordinate plane.
-
bias
The bias is one of the configurable parameters on an artificial neuron. The bias controls the amount of total input needed to activate the neuron.
-
The NOT Gate
A NOT gate has one input, which takes white and black and outputs a binary value, which is the opposite of the inpur.
when the input is (white), the neuron's output should be the opposite (black).
The only way to complete this condition is when the bias is negative and the weight is also negative.
-
The benefit of using a sigmoid neuron is immediately seeing the effect of small changes you make to the weight and bias.
when the input is off or white, only the bias influences the output.
- 1 more item...
-
-
-
-
