Typos

matlab/src/bits/im2col_gpu.cu

@@ -232,9 +232,9 @@ __global__ void col2im_gpu_kernel(T* data,
      u(x) = x_data - (x * strideX - padLeft)
      v(y) = y_data - (y * strideY - padRight)
 
-     Now we can comptute the indeces of the elements of stacked[] to accumulate:
+     Now we can compute the indices of the elements of stacked[] to accumulate:
 
-     stackedIndex(x,y) = 
+     stackedIndex(x,y) =
          (y * numPatchesX + x) +                 // column offset
          ((z * windowHeight + v(y)) * windowWidth + u(x)) *  // within patch offset
             (numPatchesX*numPatchesY)

matlab/vl_nnconv.m

@@ -6,10 +6,10 @@
 %    biases as well as performing downsampling and padding as explained
 %    below.
 %
-%    [DXDY, DXDF, DXDB] = VL_NNCONV(X, F, B, DZDY) computes the
-%    derivatives of the nework output Z w.r.t. the data X and
+%    [DZDX, DZDF, DZDB] = VL_NNCONV(X, F, B, DZDY) computes the
+%    derivatives of the network output Z w.r.t. the data X and
 %    parameters F, B given the derivative w.r.t the output Y. If B is
-%    the empty matrix, then DXDB is also empty.
+%    the empty matrix, then DZDB is also empty.
 %
 %    X is a SINGLE array of dimension H x W x D x N where (H,W) are
 %    the height and width of the map stack, D is the image depth
@@ -44,14 +44,15 @@
 %      1 <= FW <= W + 2*(PADLEFT+PADRIGHT).
 %
 %    The output a is a SINGLE array of dimension YH x YW x K x N of
-%    N images with K challens and size:
+%    N images with K channels and size:
 %
 %      YH = floor((H + (PADTOP+PADBOTTOM) - FH)/STRIDEY) + 1,
 %      YW = floor((W + (PADLEFT+PADRIGHT) - FW)/STRIDEX) + 1.
 %
 %    The derivative DZDY has the same dimension of the output Y,
-%    the derivative DZDX has the same dimension as the input X, and
-%    the derivative DZDF has the the same dimenson as F.
+%    the derivative DZDX has the same dimension as the input X,
+%    the derivative DZDF has the the same dimenson as F, and
+%    the derivative DZDB has the the same dimenson as B.
 
 % Copyright (C) 2014 Andrea Vedaldi and Max Jaderberg.
 % All rights reserved.

matlab/vl_nnloss.m

@@ -42,7 +42,7 @@
   assert(isequal(sz_, [sz(1) sz(2) 1 sz(4)])) ;
 end
 
-% convert to indeces
+% convert to indices
 c_ = 0:numel(c)-1 ;
 c_ = 1 + ...
   mod(c_, sz(1)*sz(2)) + ...

matlab/vl_nnpool.m

@@ -1,4 +1,4 @@
-% VL_NNPOOL  CNN poolinng
+% VL_NNPOOL  CNN pooling
 %    Y = VL_NNPOOL(X, POOL) applies the pooling operator to all
 %    channels of the data X using a square filter of size POOL. X is a
 %    SINGLE array of dimension H x W x D x N where (H,W) are the
@@ -9,7 +9,7 @@
 %    height POOLY and width POOLX.
 %
 %    DZDX = VL_NNPOOL(X, POOL, DZDY) computes the derivatives of
-%    the nework output Z w.r.t. the data X given the derivative DZDY
+%    the network output Z w.r.t. the data X given the derivative DZDY
 %    w.r.t the max-pooling output Y.
 %
 %    VL_NNCONV(..., 'option', value, ...) takes the following options:
@@ -37,7 +37,7 @@
 %      1 <= POOLX <= WIDTH + (PADLEFT + PADRIGHT).
 %
 %    The output a is a SINGLE array of dimension YH x YW x K x N of N
-%    images with K challens and size:
+%    images with K channels and size:
 %
 %      YH = floor((H + (PADTOP+PADBOTTOM) - POOLY)/STRIDEY) + 1,
 %      YW = floor((W + (PADLEFT+PADRIGHT) - POOLX)/STRIDEX) + 1.

matlab/vl_nnsoftmax.m

@@ -10,7 +10,7 @@
 %    convolutionally at all spatial locations.
 %
 %    DZDX = VL_NNSOFTMAX(X, DZDY) computes the derivative DZDX of the
-%    CNN otuoutwith respect to the input X given the derivative DZDY
+%    CNN output Z with respect to the input X given the derivative DZDY
 %    with respect to the block output Y. DZDX has the same dimension
 %    as X.
 

matlab/vl_nnsoftmaxloss.m

@@ -16,7 +16,7 @@
 %    convolutionally at all spatial locations.
 %
 %    DZDX = VL_NNSOFTMAXLOSS(X, C, DZDY) computes the derivative DZDX
-%    of the CNN with respect to the input X given the derivative DZDY
+%    of the CNN output Z with respect to the input X given the derivative DZDY
 %    with respect to the block output Y. DZDX has the same dimension
 %    as X.
 
@@ -42,7 +42,7 @@
   assert(isequal(sz_, [sz(1) sz(2) 1 sz(4)])) ;
 end
 
-% convert to indeces
+% convert to indices
 c_ = 0:numel(c)-1 ;
 c_ = 1 + ...
   mod(c_, sz(1)*sz(2)) + ...