I wish to compare network density metrics.
I have 6 densities from 6 different graphs. Each graph has different number of nodes as well different number of edges.
I wanted to compare the 6 densities between the graphs. In doing this I divided the graph densities with number of nodes in the graph. For example a graph with a density of 0.55 was divided with 30 (number of edges in the graph). I did the same with the other other 5 densities. Later I normalised the values and I represented on a bar chart. the chart suggests that graph densities decrease by number of nodes. Eg. high density sco…
I have 6 densities from 6 different graphs. Each graph has different number of nodes as well different number of edges.
I wanted to compare the 6 densities between the graphs. In doing this I divided the graph densities with number of nodes in the graph. For example a graph with a density of 0.55 was divided with 30 (number of edges in the graph). I did the same with the other other 5 densities. Later I normalised the values and I represented on a bar chart. the chart suggests that graph densities decrease by number of nodes. Eg. high density sco…
Why is dropout favored compared to reducing the number of units in hidden layers for the convolutional networks?
If a large set of units leads to overfitting and dropping out "averages" the response units, why not just suppress units?
I have read different questions and answers on the dropout t...
