self-organizing map
Table of Contents
, but also:
- update neighbourhood(not max, but ith-max)
- enforce convergence(let learning rate fade to 0)
1. algorithm
The same as , but change the incremental term \(\Delta w_{ji}^t\) to: \(\Delta w_{ji}^t = C(t)(a_i^t - w^t_{ji})\theta(j,j^*)\)
- the \(\theta\) function represent distance from the ith-max neuron to the 1th-max neuron j.
2. referecne
Bibliography
[1]
T. Kohonen, “Essentials of the self-organizing map,” Neural networks, vol. 37, pp. 52–65, Jan. 2013, doi: 10.1016/j.neunet.2012.09.018.
Backlinks
unsupervised learning
unsupervised learning algorithms are learning algorithms that do not use labels.