18:50:26	  Anon. Myrtle:	It seems the recording has stopped
18:50:32	  Anon. Hobart:	^^
18:50:38	  Anon. Murdoch:	^
18:51:28	  Anon. Grandview:	professor … the recording has stopped!!!
18:51:33	  Anon. Rocket:	Lean On Me
18:53:54	  Anon. Spiderman:	yup
18:55:16	  Anon. Superman:	LOM Number:(412)-530-4700
18:56:38	  Anon. Mantis:	no
18:57:15	  Anon. Mantis:	depends on location of the patern in the input
18:58:33	  Anon. Mantis:	no
18:59:57	  Anon. Heimdall:	not unless its trained to
19:00:00	  Anon. Green Lantern:	no
19:00:04	  Anon. Batman:	Different location of flower
19:00:05	  Anon. P.J. McArdle:	no
19:00:15	  Anon. BlackWidow:	different subspace
19:00:27	  Anon. Murray:	The weights and biases have been configured for different input dimensions?
19:00:58	  Anon. Heimdall:	yes
19:02:51	  Anon. Mantis:	scan the input for the pattern
19:02:53	  Anon. BlackWidow:	train an MLP over all the different subspaces?
19:03:51	  Anon. Frew:	max
19:14:19	  Anon. Heimdall:	no
19:27:44	  Anon. Green Lantern:	Can we think of them as one subnet scanning over time, or multiple identical subnets simultaneously capturing information from each subarea, are they equivalent？
19:29:33	  Anon. Murray:	no
19:36:17	  Anon. Hobart:	yes
19:57:08	  Anon. Heimdall:	yes
20:02:21	  Anon. Spiderman:	yeah
20:02:24	  Anon. S. Highland:	yes
20:05:37	  Anon. Spiderman:	so there's one convolution for each neuron, each producing a new map, and all the maps together make up the next layer?
20:14:10	  Anon. Frew:	75%