Skip to main content
. 2024 Apr 16;24:416. doi: 10.1186/s12909-024-05352-y

Table 1.

Training algorithm – Parallel Batch SOM with sparse binary matrices

graphic file with name 12909_2024_5352_Tab1_HTML.jpg

D - total number of MeSH categories (=29,917–; Da - number of MeSH categories annotating an individual article; N - number of articles (=33,375,866); M - number of nodes (350x350=122,500); –(e) - width of the neighborhood (which changes over training epochs according to the formula 175/(1.7)epoch)