Add sum mode to Downsample

theano/tensor/signal/downsample.py

@@ -64,10 +64,10 @@ def max_pool_2d(input, ds, ignore_border=False, st=None, padding=(0, 0),
             of the images, pad_h is the size of the top and bottom margins,
             and pad_w is the size of the left and right margins.
     :type padding: tuple of two ints
-    :param mode: 'max', 'average_inc_pad' or 'average_exc_pad'.
-        Operation executed on each window.  `max` always excludes the padding
-        in the computation. `average` gives you the choice to include or
-        exclude it.
+    :param mode: 'max', 'sum', 'average_inc_pad' or 'average_exc_pad'.
+        Operation executed on each window.  `max` and `sum` always exclude
+        the padding in the computation. `average` gives you the choice to
+        include or exclude it.
     :type mode: string
     """
     if input.ndim < 2:
@@ -104,7 +104,7 @@ def max_pool_2d(input, ds, ignore_border=False, st=None, padding=(0, 0),
 class DownsampleFactorMax(Op):
     """For N-dimensional tensors, consider that the last two
     dimensions span images.  This Op downsamples these images by
-    taking the max or average over different patch.
+    taking the max, sum or average over different patch.
 
     """
     __props__ = ('ds', 'ignore_border', 'st', 'padding', 'mode')
@@ -188,7 +188,7 @@ class DownsampleFactorMax(Op):
 
     def __init__(self, ds, ignore_border=False, st=None, padding=(0, 0),
                  mode='max'):
-        """ Take the max or average or different input patches.
+        """ Take the max, sum or average or different input patches.
 
         :param ds: downsample factor over rows and column.
                    ds indicates the pool region size.
@@ -210,8 +210,8 @@ class DownsampleFactorMax(Op):
             and pad_w is the size of the left and right margins.
         :type padding: tuple of two ints
 
-        :param mode: 'max', 'average_inc_pad', 'average_exc_pad'.
-            ('average_inc_pad' exclude the padding from the count,
+        :param mode: 'max', 'sum', 'average_inc_pad', 'average_exc_pad'.
+            ('average_inc_pad' excludes the padding from the count,
             'average_exc_pad' include it)
 
         """
@@ -232,9 +232,9 @@ class DownsampleFactorMax(Op):
         if self.padding[0] >= self.ds[0] or self.padding[1] >= self.ds[1]:
             raise NotImplementedError(
                 'padding_h and padding_w must be smaller than strides')
-        if mode not in ['max', 'average_inc_pad', 'average_exc_pad']:
+        if mode not in ['max', 'average_inc_pad', 'average_exc_pad', 'sum']:
             raise ValueError(
-                "DownsampleFactorMax mode parameter only support 'max',"
+                "DownsampleFactorMax mode parameter only support 'max', 'sum',"
                 " 'average_inc_pad' and 'average_exc_pad'. Got %s" % mode)
         self.mode = mode
 
@@ -277,7 +277,9 @@ class DownsampleFactorMax(Op):
         else:
             y = x
         func = numpy.max
-        if self.mode != 'max':
+        if self.mode == 'sum':
+            func = numpy.sum
+        elif self.mode != 'max':
             func = numpy.average
 
         for n in xrange(x.shape[0]):
@@ -317,7 +319,7 @@ class DownsampleFactorMax(Op):
         return ['<algorithm>']
 
     def c_code(self, node, name, inp, out, sub):
-        if self.mode not in ('max', 'average_exc_pad', 'average_inc_pad'):
+        if self.mode not in ('max', 'sum', 'average_exc_pad', 'average_inc_pad'):
             raise theano.gof.utils.MethodNotDefined()
         x, = inp
         z, = out
@@ -448,7 +450,7 @@ class DownsampleFactorMax(Op):
         """
         if self.mode == 'max':
             ccode += """
-                    // use the first element as the initial value of maximum
+                    // use the first element as the initial value of collector
                     collector = ((dtype_%(x)s*)(PyArray_GETPTR4(%(x)s,b,k,r_st,c_st)))[0];
                     // go through the pooled region in the unpadded input
                     for(int m=r_st; m<r_end; m++)
@@ -461,7 +463,7 @@ class DownsampleFactorMax(Op):
                     }
                     z[0] = collector;
             """
-        elif self.mode == 'average_exc_pad' or self.mode == 'average_inc_pad':
+        elif self.mode in ('sum', 'average_exc_pad', 'average_inc_pad'):
             ccode += """
                     // initialize the sum at zero
                     collector = ((dtype_%(x)s)(0));
@@ -475,7 +477,11 @@ class DownsampleFactorMax(Op):
                       }
                     }
             """
-            if self.mode == 'average_inc_pad' and self.ignore_border:
+            if self.mode == "sum":
+                ccode += """
+                    z[0] = collector;
+                """
+            elif self.mode == 'average_inc_pad' and self.ignore_border:
                 ccode += """
                     z[0] = collector / (%(ds0)s * %(ds1)s);
                 """
@@ -493,7 +499,7 @@ class DownsampleFactorMax(Op):
         return ccode % locals()
 
     def c_code_cache_version(self):
-        return (0, 6, 8, 1)
+        return (0, 6, 8, 3)
 
 class DownsampleFactorMaxGrad(Op):
     __props__ = ('ds', 'ignore_border', 'st', 'padding', 'mode')
@@ -505,9 +511,9 @@ class DownsampleFactorMaxGrad(Op):
             st = ds
         self.st = tuple(st)
         self.padding = tuple(padding)
-        if mode not in ['max', 'average_inc_pad', 'average_exc_pad']:
+        if mode not in ['max', 'sum', 'average_inc_pad', 'average_exc_pad']:
             raise ValueError(
-                "DownsampleFactorMax mode parameter only support 'max',"
+                "DownsampleFactorMax mode parameter only support 'max', 'sum',"
                 " 'average_inc_pad' and 'average_exc_pad'. Got %s" % mode)
         self.mode = mode
 
@@ -524,7 +530,7 @@ class DownsampleFactorMaxGrad(Op):
         return Apply(self, [x, maxout, gz], [x.type()])
 
     def perform(self, node, inp, out):
-        if self.mode != 'max' and self.padding != (0, 0):
+        if self.mode not in ('max', 'sum') and self.padding != (0, 0):
             raise NotImplementedError()
         x, maxout, gz = inp
         gx_stg, = out
@@ -539,6 +545,7 @@ class DownsampleFactorMaxGrad(Op):
         img_rows = x.shape[-2] + 2 * pad_h
         img_cols = x.shape[-1] + 2 * pad_w
         inc_pad = self.mode == 'average_inc_pad'
+        sum_mode = self.mode == 'sum'
 
         # pad the image
         if self.padding != (0, 0):
@@ -566,18 +573,21 @@ class DownsampleFactorMaxGrad(Op):
             for n in xrange(x.shape[0]):
                 for k in xrange(x.shape[1]):
                     for r in xrange(pr):
-                        if inc_pad:
+                        if sum_mode or inc_pad:
                             row_st = r * st0
                         else:
                             row_st = __builtin__.max(r * st0, self.padding[0])
                         row_end = __builtin__.min(row_st + ds0, img_rows)
                         for c in xrange(pc):
-                            if inc_pad:
+                            if sum_mode or inc_pad:
                                 col_st = c * st1
                             else:
                                 col_st = __builtin__.max(c * st1,
                                                          self.padding[1])
                             col_end = __builtin__.min(col_st + ds1, img_cols)
+                            if sum_mode:
+                              val = gz[n, k, r, c]
+                            else:
                               val = gz[n, k, r, c] / ((row_end - row_st) *
                                                       (col_end - col_st))
                             gx[n, k, row_st:row_end, col_st:col_end] += val

theano/tensor/signal/tests/test_downsample.py

@@ -33,7 +33,9 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
         out_shp.append(input.shape[-1] / ds[1] + yi)
         output_val = numpy.zeros(out_shp)
         func = numpy.max
-        if mode != 'max':
+        if mode == 'sum':
+            func = numpy.sum
+        elif mode != 'max':
             func = numpy.average
 
         for k in numpy.ndindex(*input.shape[:-2]):
@@ -76,7 +78,9 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
         tt = []
         y = pad_img(x)
         func = numpy.max
-        if mode != 'max':
+        if mode == 'sum':
+            func = numpy.sum
+        elif mode != 'max':
             func = numpy.average
         inc_pad = mode == 'average_inc_pad'
 
@@ -145,7 +149,9 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
         out_shp.append(out_c)
 
         func = numpy.max
-        if mode != 'max':
+        if mode == 'sum':
+            func = numpy.sum
+        elif mode != 'max':
             func = numpy.average
 
         output_val = numpy.zeros(out_shp)
@@ -169,6 +175,7 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
         for maxpoolshp, ignore_border, mode in product(maxpoolshps,
                                                        [True, False],
                                                        ['max',
+                                                        'sum',
                                                         'average_inc_pad',
                                                         'average_exc_pad']):
                 # print 'maxpoolshp =', maxpoolshp
@@ -198,23 +205,23 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
         stridesizes = ((1, 1), (3, 3), (5, 7))
         # generate random images
         imval = rng.rand(4, 10, 16, 16)
+        # The same for each mode
         outputshps = ((4, 10, 16, 16), (4, 10, 6, 6), (4, 10, 4, 3),
                       (4, 10, 16, 16), (4, 10, 6, 6), (4, 10, 4, 3),
                       (4, 10, 14, 14), (4, 10, 5, 5), (4, 10, 3, 2),
                       (4, 10, 14, 14), (4, 10, 6, 6), (4, 10, 4, 3),
                       (4, 10, 12, 14), (4, 10, 4, 5), (4, 10, 3, 2),
                       (4, 10, 12, 14), (4, 10, 5, 6), (4, 10, 4, 3))
-        # The same for each mode
-        outputshps = outputshps + outputshps + outputshps
         images = tensor.dtensor4()
         indx = 0
         for mode, maxpoolshp, ignore_border in product(['max',
+                                                        'sum',
                                                         'average_inc_pad',
                                                         'average_exc_pad'],
                                                        maxpoolshps,
                                                        [True, False]):
                 for stride in stridesizes:
-                    outputshp = outputshps[indx]
+                    outputshp = outputshps[indx % len(outputshps)]
                     indx += 1
                     # DownsampleFactorMax op
                     numpy_output_val = \
@@ -251,7 +258,8 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
             stride = stridesizes[indx]
             maxpoolshp = maxpoolshps[indx]
             for ignore_border, mode in product([True, False],
-                                               ['max', 'average_inc_pad',
+                                               ['max', 'sum',
+                                                'average_inc_pad',
                                                 'average_exc_pad']):
                 indx_out = indx * 2
                 if not ignore_border:
@@ -283,7 +291,7 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
         c = 2  # channel size
         images = tensor.dtensor4()
         for indx, mode in product(numpy.arange(len(maxpoolsizes)),
-                                  ['max', 'average_inc_pad',
+                                  ['max', 'sum', 'average_inc_pad',
                                    'average_exc_pad']):
             imgsize = imgsizes[indx]
             imval = rng.rand(m, c, imgsize[0], imgsize[1]) - 0.5
@@ -486,7 +494,7 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
 
         for maxpoolshp, ignore_border, mode in product(maxpoolshps,
                                                        [True, False],
-                                                       ['max',
+                                                       ['max', 'sum',
                                                         'average_inc_pad',
                                                         'average_exc_pad']):
                 # print 'maxpoolshp =', maxpoolshp
@@ -537,7 +545,7 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
 
         for maxpoolshp, ignore_border, mode in product(maxpoolshps,
                                                        [True, False],
-                                                       ['max',
+                                                       ['max', 'sum',
                                                         'average_inc_pad',
                                                         'average_exc_pad']):
                 # print 'maxpoolshp =', maxpoolshp
@@ -574,7 +582,7 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
 
         for maxpoolshp, ignore_border, mode in product(maxpoolshps,
                                                        [True, False],
-                                                       ['max',
+                                                       ['max', 'sum',
                                                         'average_inc_pad',
                                                         'average_exc_pad']):
                 # print 'maxpoolshp =', maxpoolshp