[totierne2]

Back in 1992 I was using a neural network library in Pascal. I did not know the admins were killing long running nohup tasks, rather than the program failing, so I could not run it on the UNIX boxes in the background, hence lots of foreground and usenet reading. The neural network use was applied, 'how did number of nodes, number of layers matter?' to analysing fetal heart rate data. Backpropagation was built in. I was recovering from a nervous breakdown, so my virtual neural network having breakdowns was a little bit strange. I got a 2:1 for the project i.e. good, but a 2.2 for the whole final year, i.e. very average.