This text is part 21 of “Hivan’s AI Compendium — AI Core Foundations”
Whats up, I’m Hivan.
Within the final class, we mentioned the rules of multi-layer neural networks and understood that the rationale knowledge can’t exceed three layers is because of complexity.
We used a diagram to clarify the whole chain rule differentiation course of:
So, how will we use the node relationships on this diagram to acquire the differentiation course of?
Suppose we outline a perform get_output
:
def get_output(graph, node):
outputs = []
for n, hyperlinks in graph.gadgets():
if node == n:
outputs += hyperlinks
return outputsget_output(computing_graph, 'k1')
---
['L1']
We are able to get l1
from k1
.