Click here to Download # # DCN_AlexNet.moo # # This model was designed to conform to the implementation of AlexNet # published for the Caffe Deep Learning Framework. # mod_type caffe_net sscale 0.1 (deg/pix) # Number of degrees (of visual angle) per pixel # thus, 1 cyc/deg ~= 1 cyc per conv1 kernel width tscale 0.001 (s/frame) # Number of seconds per stimulus frame xn 227 (pix) # yn 227 (pix) # tn 1 (frames) # weight_filename data/anw.txt # Weights based on Caffe's AlexNet version stim_depth 3 # For R,G,B stimuli <layer> name conv1 type Convolution bottom data top conv1 <param> num_output 96 kernel_size 11 stride 4 </param> outdim 10 96 55 55 </layer> <layer> name relu1 type ReLU bottom conv1 top relu1 outdim 10 96 55 55 </layer> <layer> name pool1 type Pooling bottom relu1 top pool1 <param> pool MAX kernel_size 3 stride 2 </param> outdim 10 96 27 27 </layer> <layer> name norm1 type LRN bottom pool1 top norm1 <param> local_size 5 alpha 0.0001 beta 0.75 </param> outdim 10 96 27 27 </layer> <layer> name conv2 type Convolution bottom norm1 top conv2 <param> num_output 256 pad 2 kernel_size 5 group 2 </param> outdim 10 256 27 27 </layer> <layer> name relu2 type ReLU bottom conv2 top relu2 outdim 10 256 27 27 </layer> <layer> name pool2 type Pooling bottom relu2 top pool2 <param> pool MAX kernel_size 3 stride 2 </param> outdim 10 256 13 13 </layer> <layer> name norm2 type LRN bottom pool2 top norm2 <param> local_size 5 alpha 0.0001 beta 0.75 </param> outdim 10 256 13 13 </layer> <layer> name conv3 type Convolution bottom norm2 top conv3 <param> num_output 384 pad 1 kernel_size 3 </param> outdim 10 384 13 13 </layer> <layer> name relu3 type ReLU bottom conv3 top relu3 outdim 10 384 13 13 </layer> <layer> name conv4 type Convolution bottom relu3 top conv4 <param> num_output 384 pad 1 kernel_size 3 group 2 </param> outdim 10 384 13 13 </layer> <layer> name relu4 type ReLU bottom conv4 top relu4 </layer> <layer> name conv5 type Convolution bottom relu4 top conv5 <param> num_output 256 pad 1 kernel_size 3 group 2 </param> </layer> <layer> name relu5 type ReLU bottom conv5 top relu5 </layer> <layer> name pool5 type Pooling bottom relu5 top pool5 <param> pool MAX kernel_size 3 stride 2 </param> </layer> <layer> name fc6 type InnerProduct bottom pool5 top fc6 <param> num_output 4096 </param> </layer> <layer> name relu6 type ReLU bottom fc6 top relu6 </layer> <layer> name fc7 type InnerProduct bottom relu6 top fc7 <param> num_output 4096 </param> </layer> <layer> name relu7 type ReLU bottom fc7 top relu7 </layer> <layer> name fc8 type InnerProduct bottom relu7 top fc8 <param> num_output 1000 </param> </layer> <layer> name prob type Softmax bottom fc8 top prob </layer>