Figure 4

Test result of LVQ after successful training of the architecture to classify N-MORB, E/P-MORB and OIB using 25, 50 and 75 neurons. (a) The use of 25 neurons for identification of different basalts indicates that the network performed excellently and gives 100% satisfactory result for N-MORB, E/P-MORB and OIB. (b) A 50 neuron architecture shows a 100% satisfactory result for N-MORB whereas a mis-classification is noticeable for E/P-MORB and OIB. (c) A 75 neuron architecture performs 100% for N-MORB and OIB while a mis-classification occurs for the E/P-MORB.