91 lines
1.0 KiB
Plaintext
91 lines
1.0 KiB
Plaintext
<subsection Scaling>
|
|
<frame>
|
|
|
|
<split>
|
|
<que>
|
|
<list>
|
|
<e>Now lets increase the graph size(gs)</e>
|
|
<e>plot auc against it</e>
|
|
<e>compare to QCDorWhat results</e>
|
|
</list>
|
|
</que>
|
|
<que>
|
|
<i f="lowscale"></i>
|
|
</que>
|
|
</split>
|
|
|
|
</frame>
|
|
<frame>
|
|
<i f="trivscale"></i>
|
|
</frame>
|
|
<frame>
|
|
<i f="compscale"></i>
|
|
</frame>
|
|
<frame>
|
|
<i f="compscale_zoom"></i>
|
|
</frame>
|
|
<frame>
|
|
<split>
|
|
<que w="0.4">
|
|
<list>
|
|
<e>Now instead of training on network with #4*n# nodes, train #n# networks on #4# nodes each and combine them into one #4*n# network</e>
|
|
<e>if the network would be supervised, this would hurt, since there is no interaction between particles possible</e>
|
|
<e>but here it actually helps</e>
|
|
</list>
|
|
</que>
|
|
<que w="0.58">
|
|
<i f="splitscale"></i>
|
|
</que>
|
|
</split>
|
|
</frame>
|
|
|
|
|
|
|
|
<ignore>
|
|
|
|
<split>
|
|
<que>
|
|
<list>
|
|
<e></e>
|
|
<e></e>
|
|
<e></e>
|
|
</list>
|
|
</que>
|
|
<que>
|
|
|
|
</que>
|
|
</split>
|
|
|
|
<frame>
|
|
|
|
<split>
|
|
<que>
|
|
<list>
|
|
<e></e>
|
|
<e></e>
|
|
<e></e>
|
|
</list>
|
|
</que>
|
|
<que>
|
|
<i f="none"></i>
|
|
</que>
|
|
</split>
|
|
|
|
</frame>
|
|
|
|
<frame>
|
|
|
|
<list>
|
|
<e></e>
|
|
<e></e>
|
|
<e></e>
|
|
</list>
|
|
|
|
</frame>
|
|
|
|
|
|
</ignore>
|
|
|
|
|
|
|