switches to matconvnet beta17

doc/instructions.md

@@ -649,7 +649,7 @@ Note that we remember the `imageMean` for later use. Note also that the softmaxl
 The next step is to glance at the filters that have been learned:
 ```matlab
 figure(2) ; clf ; colormap gray ;
-vl_imarraysc(squeeze(net.layers{1}.filters),'spacing',2)
+vl_imarraysc(squeeze(net.layers{1}.weights{1}),'spacing',2)
 axis equal ;
 title('filters in the first layer') ;
 ```
@@ -787,6 +787,7 @@ That completes this practical.
 
 ## History
 
+* Used in the Oxford AIMS CDT, 2015-16.
 * Used in the Oxford AIMS CDT, 2014-15.
 
 [^lattice]: A two-dimensional *lattice* is a discrete grid embedded in $R^2$, similar for example to a checkerboard.

exercise4.m

@@ -68,7 +68,7 @@ function exercise4(varargin)
 % -------------------------------------------------------------------------
 
 figure(2) ; clf ; colormap gray ;
-vl_imarraysc(squeeze(net.layers{1}.filters),'spacing',2)
+vl_imarraysc(squeeze(net.layers{1}.weights{1}),'spacing',2)
 axis equal ; title('filters in the first layer') ;
 
 % -------------------------------------------------------------------------

extra/download.sh

@@ -1,5 +1,5 @@
 #!/bin/bash
-wget http://www.vlfeat.org/matconvnet/models/imagenet-vgg-verydeep-16.mat \
+wget http://www.vlfeat.org/sandbox-matconvnet/models/imagenet-vgg-verydeep-16.mat \
     --output-document=data/imagenet-vgg-verydeep-16.mat --continue
 
 if test ! -e vlfeat
@@ -11,8 +11,8 @@ fi
 
 if test ! -e matconvnet
 then
-    wget http://www.vlfeat.org/matconvnet/download/matconvnet-1.0-beta16.tar.gz \
+    wget http://www.vlfeat.org/sandbox-matconvnet/download/matconvnet-1.0-beta17.tar.gz \
          --output-document=data/matconvnet.tar.gz --continue
     tar xzvf data/matconvnet.tar.gz
-    mv matconvnet-1.0-beta16 matconvnet
+    mv matconvnet-1.0-beta17 matconvnet
 fi

initializeCharacterCNN.m

@@ -3,8 +3,7 @@
 f=1/100 ;
 net.layers = {} ;
 net.layers{end+1} = struct('type', 'conv', ...
-                           'filters', f*randn(5,5,1,20, 'single'), ...
-                           'biases', zeros(1, 20, 'single'), ...
+                           'weights', {{f*randn(5,5,1,20, 'single'), zeros(1, 20, 'single')}}, ...
                            'stride', 1, ...
                            'pad', 0) ;
 net.layers{end+1} = struct('type', 'pool', ...
@@ -13,8 +12,7 @@
                            'stride', 2, ...
                            'pad', 0) ;
 net.layers{end+1} = struct('type', 'conv', ...
-                           'filters', f*randn(5,5,20,50, 'single'),...
-                           'biases', zeros(1,50,'single'), ...
+                           'weights', {{f*randn(5,5,20,50, 'single'), zeros(1,50,'single')}}, ...
                            'stride', 1, ...
                            'pad', 0) ;
 net.layers{end+1} = struct('type', 'pool', ...
@@ -23,14 +21,14 @@
                            'stride', 2, ...
                            'pad', 0) ;
 net.layers{end+1} = struct('type', 'conv', ...
-                           'filters', f*randn(4,4,50,500, 'single'),...
-                           'biases', zeros(1,500,'single'), ...
+                           'weights', {{f*randn(4,4,50,500, 'single'), zeros(1,500,'single')}}, ...
                            'stride', 1, ...
                            'pad', 0) ;
 net.layers{end+1} = struct('type', 'relu') ;
 net.layers{end+1} = struct('type', 'conv', ...
-                           'filters', f*randn(2,2,500,26, 'single'),...
-                           'biases', zeros(1,26,'single'), ...
+                           'weights', {{f*randn(2,2,500,26, 'single'), zeros(1,26,'single')}}, ...
                            'stride', 1, ...
                            'pad', 0) ;
 net.layers{end+1} = struct('type', 'softmaxloss') ;
+
+net = vl_simplenn_tidy(net) ;