Skip to first unread message
Iman Akbari
Dec 24, 2014, 4:17:30 AM12/24/14
to caffe...@googlegroups.com
Hi
It seems that labels are integer numbers. In the examples (e.g: mnist, imagenet) a single integer is used as label for each input. I need to set a vector as output not just a number. A number per an output neuron.
If labels must be integers, how can I determine the output vector? For example, my net has three output neurons and I want to set 001, 010, 110, 111 and 100 as output for five of inputs. how can I do this? Is it possible in Caffe?
On the other hand, when we have 4 classes, the last layer must have 4 neurons? If we have 6 classes Caffe gives error? What about when number of classes are fewer than number of neurons?
ThanksIt seems that labels are integer numbers. In the examples (e.g: mnist, imagenet) a single integer is used as label for each input. I need to set a vector as output not just a number. A number per an output neuron.
If labels must be integers, how can I determine the output vector? For example, my net has three output neurons and I want to set 001, 010, 110, 111 and 100 as output for five of inputs. how can I do this? Is it possible in Caffe?
On the other hand, when we have 4 classes, the last layer must have 4 neurons? If we have 6 classes Caffe gives error? What about when number of classes are fewer than number of neurons?
Charles Shang
Dec 24, 2014, 10:00:14 AM12/24/14
to caffe...@googlegroups.com
check this.
https://github.com/BVLC/caffe/issues/881#issuecomment-53533978
在 2014年12月24日星期三UTC+8下午5时17分30秒，Iman Akbari写道：
https://github.com/BVLC/caffe/issues/881#issuecomment-53533978
在 2014年12月24日星期三UTC+8下午5时17分30秒，Iman Akbari写道：
Reply all
Reply to author
Forward
0 new messages