BytheCPP
Hello,
How are the connections between two neurons in a Ki (with e.g. 10 neurons) calculated during training?
For example, by looking at which connection has the lowest error rate/highest score and then setting the connections?
Thanks for an answer
Just ask me if you don't understand the question.
TheCPP
(1 votes)
Loading...If so, in which specific situations do they help you the most – and where do you reach your limits? 🤖📚
Even my calculator has an AI. Why can I talk to my calculator?
Hey, I have an assignment on Bubblesort – of course I don't want you to do my homework, but I'm stuck at the moment because I'm a bit confused and have a question about it. So, here you're supposed to perform a manual simulation, and as I understand it, Bubble Sort runs through the loop…
If I factor out x^4, what is e^x then? So x^4*(??
Connections between neurons are not discrete (on or off) but are simply numerical values that indicate the strength of the connection. A value of 0 means that the activations of the previous layers are multiplied by 0—as if the connection were not present. The values of the connections can be arbitrarily large and can even be negative.
Neural networks are typically represented using linear algebra. So, each layer of neurons is a vector containing the values of the neurons. If you have A neurons in one layer and B neurons in the next, you can represent the connections with a BxA matrix.
To find the correct values for the connections, the so-called loss function (which measures how good or bad the network is) is derived from the individual connection values (called weights). Then, one performs "gradient descent." This procedure is called backpropagation.
Das funktioniert überhaupt nicht so. Neuronen sind als mathematische Funktion definiert, die “Verbindungen” dahin sind Funktionsparameter, Eingaben sozusagen.
Man guckt auch nicht welche “Verbindung” die niedrigste Fehlerrate erreicht. Das Netz wird als globales optimiert. Der normalerweise genutzte Ansatz nennt sich Backpropagation. Letztendlich sind das mathematisch betrachtet Ableitungen.