Fig 3 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Source publication
A comparison between neural network clustering (NNC), hierarchical clustering (HC) and K-means clustering (KMC) is performed to evaluate the computational superiority of these three machine learning (ML) techniques for organizing large datasets into clusters. For NNC, a self-organizing map (SOM) training was applied to a collection of wavefront sen...
Contexts in source publication
Context 1
... evaluate the number of input vectors linked to each neuron, we generate an SOM sample hits plot (see Fig. 3). The number displayed on a shaded neuron indicates its association with an input vector. This visualization reveals that the highest number of input vectors associated with a single neuron is 1. It is best if the data are fairly evenly distributed across the neurons. In this example, the data are concentrated a little more in the ...
Context 2
... Three key aspects of the SOM-Sample Hits, Neighboring Weight Distances, Weight Planes and Weight Positions-were analyzed to assess the distribution and consolidation of Z 1 − Z 15 Zernike components. The visualization of sample hits suggests that the 16 sampled locations can be categorized into eight distinct clusters based on data conformity (Fig. 3). In addition to this, the dissimilarity among the 16 vectors was assessed through Neighbour Weight Distances as shown in Fig. 2. Approximately 8 clusters can be identified according to high correlation, indicated by yellow areas surrounded by dark red or black patches in the maps. To identify highly correlated Zernike variables, a ...
