machine learning - Training convolutional nets on multi-channel image data sets -
i trying implement convolutional neural network scratch , not able figure out how perform (vectorized)operations on multi-channel images rgb, have 3 dimensions. on following articles , tutorials such this cs231n tutorial , it's pretty clear implement network single input input layer 3d matrix there multiple data points in dataset. so, cannot figure out how implement these networks vectorized operation on entire datsets.
i have implemented network takes 3d matrix input have realized not work on entire dataset have propagate 1 input @ time.i don't know whether conv nets vectorized on entire dataset or not .but if are, how can vectorize convolutional network multi-channel images ?
wiki
Comments
Post a Comment