andy
64276f270d
Affected files: .obsidian/graph.json .obsidian/workspace-mobile.json .obsidian/workspace.json STEM/AI/Literature.md STEM/AI/Neural Networks/MLP.md STEM/AI/Properties.md STEM/Quantum/Orbitals.md STEM/Quantum/Schrödinger.md STEM/Quantum/Wave Function.md STEM/Signal Proc/Convolution.md STEM/Signal Proc/Image/Image Processing.md STEM/img/hydrogen-electron-density.png STEM/img/hydrogen-wave-function.png STEM/img/orbitals-radius.png STEM/img/radial-equations.png STEM/img/radius-electron-density-wf.png STEM/img/wave-function-nodes.png STEM/img/wave-function-polar-segment.png STEM/img/wave-function-polar.png
22 lines
787 B
Markdown
22 lines
787 B
Markdown
- Feed-forward
|
|
- Single hidden layer can learn any function
|
|
- Universal approximation theorem
|
|
- Each hidden layer can operate as a different feature extraction layer
|
|
- Lots of weights to learn
|
|
- [[Back-Propagation]] is supervised
|
|
|
|
![[mlp-arch.png]]
|
|
|
|
# Universal Approximation Theory
|
|
A finite feed-forward MLP with 1 hidden layer can in theory approximate any mathematical function
|
|
- In practice not trainable with [[Back-Propagation|BP]]
|
|
|
|
![[activation-function.png]]
|
|
![[mlp-arch-diagram.png]]
|
|
## Weight Matrix
|
|
- Use matrix multiplication for layer output
|
|
- TLU is hard limiter
|
|
![[tlu.png]]
|
|
- $o_1$ to $o_4$ must all be one to overcome -3.5 bias and force output to 1
|
|
![[mlp-non-linear-decision.png]]
|
|
- Can generate a non-linear [[Decision Boundary|decision boundary]] |