Skip to content

Commit 51b4b24

Browse files
lhklhk
committed
adding ipython notebook for 2d convolution
1 parent 048914c commit 51b4b24

File tree

1 file changed

+344
-0
lines changed

1 file changed

+344
-0
lines changed

‎conv.ipynb

+344
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,344 @@
1+
{
2+
"cells": [
3+
{
4+
"cell_type": "code",
5+
"execution_count": 1,
6+
"metadata": {
7+
"collapsed": true,
8+
"deletable": true,
9+
"editable": true
10+
},
11+
"outputs": [],
12+
"source": [
13+
"import numpy as np"
14+
]
15+
},
16+
{
17+
"cell_type": "code",
18+
"execution_count": 2,
19+
"metadata": {
20+
"collapsed": true,
21+
"deletable": true,
22+
"editable": true
23+
},
24+
"outputs": [],
25+
"source": [
26+
"dim_x, dim_y, colors=10,11,5\n",
27+
"kernel_x, kernel_y, in_colors, out_colors=2,3,colors, 4"
28+
]
29+
},
30+
{
31+
"cell_type": "code",
32+
"execution_count": 3,
33+
"metadata": {
34+
"collapsed": false,
35+
"deletable": true,
36+
"editable": true
37+
},
38+
"outputs": [],
39+
"source": [
40+
"im=np.random.rand(dim_x, dim_y, colors)\n",
41+
"weights=np.random.rand(kernel_x, kernel_y, in_colors, out_colors)\n",
42+
"\n",
43+
"debug_weights=np.random.rand(*(dim_x-kernel_x+1, dim_y-kernel_y+1, out_colors))"
44+
]
45+
},
46+
{
47+
"cell_type": "code",
48+
"execution_count": 4,
49+
"metadata": {
50+
"collapsed": false,
51+
"deletable": true,
52+
"editable": true
53+
},
54+
"outputs": [
55+
{
56+
"data": {
57+
"text/plain": [
58+
"(10, 11, 5)"
59+
]
60+
},
61+
"execution_count": 4,
62+
"metadata": {},
63+
"output_type": "execute_result"
64+
}
65+
],
66+
"source": [
67+
"im.shape"
68+
]
69+
},
70+
{
71+
"cell_type": "code",
72+
"execution_count": 5,
73+
"metadata": {
74+
"collapsed": true,
75+
"deletable": true,
76+
"editable": true
77+
},
78+
"outputs": [],
79+
"source": [
80+
"def forward(im, weights):\n",
81+
" kernel_x, kernel_y, in_colors, out_colors=weights.shape\n",
82+
" dim_x, dim_y, colors=im.shape\n",
83+
" \n",
84+
" \n",
85+
" out=np.empty((dim_x-kernel_x+1, dim_y-kernel_y+1, out_colors))\n",
86+
" \n",
87+
" for i in range(out.shape[0]):\n",
88+
" for j in range(out.shape[1]):\n",
89+
" crop=im[i:i+kernel_x, j:j+kernel_y]\n",
90+
" #expand crops so the dimensions match\n",
91+
" crop=np.expand_dims(crop, axis=-1)\n",
92+
" \n",
93+
" res=crop*weights\n",
94+
" \n",
95+
" #sum everything except the output\n",
96+
" res=np.apply_over_axes(np.sum, res, [0,1,2]).reshape(-1)\n",
97+
" \n",
98+
" out[i,j]=res\n",
99+
" \n",
100+
" return out"
101+
]
102+
},
103+
{
104+
"cell_type": "code",
105+
"execution_count": 6,
106+
"metadata": {
107+
"collapsed": false,
108+
"deletable": true,
109+
"editable": true
110+
},
111+
"outputs": [],
112+
"source": [
113+
"out=forward(im, weights)"
114+
]
115+
},
116+
{
117+
"cell_type": "code",
118+
"execution_count": 7,
119+
"metadata": {
120+
"collapsed": false
121+
},
122+
"outputs": [
123+
{
124+
"data": {
125+
"text/plain": [
126+
"(9, 9, 4)"
127+
]
128+
},
129+
"execution_count": 7,
130+
"metadata": {},
131+
"output_type": "execute_result"
132+
}
133+
],
134+
"source": [
135+
"out.shape"
136+
]
137+
},
138+
{
139+
"cell_type": "code",
140+
"execution_count": 8,
141+
"metadata": {
142+
"collapsed": false,
143+
"deletable": true,
144+
"editable": true
145+
},
146+
"outputs": [],
147+
"source": [
148+
"debug_out=np.sum(out*debug_weights)"
149+
]
150+
},
151+
{
152+
"cell_type": "code",
153+
"execution_count": 9,
154+
"metadata": {
155+
"collapsed": false,
156+
"deletable": true,
157+
"editable": true
158+
},
159+
"outputs": [
160+
{
161+
"data": {
162+
"text/plain": [
163+
"1202.8937344394294"
164+
]
165+
},
166+
"execution_count": 9,
167+
"metadata": {},
168+
"output_type": "execute_result"
169+
}
170+
],
171+
"source": [
172+
"debug_out"
173+
]
174+
},
175+
{
176+
"cell_type": "code",
177+
"execution_count": 10,
178+
"metadata": {
179+
"collapsed": false,
180+
"deletable": true,
181+
"editable": true
182+
},
183+
"outputs": [],
184+
"source": [
185+
"debug_shape=debug_weights.shape\n",
186+
"\n",
187+
"padded=np.zeros((debug_shape[0]+2*(kernel_x-1), debug_shape[1]+2*(kernel_y-1), out_colors))\n",
188+
"padded[kernel_x-1:kernel_x-1+debug_shape[0], kernel_y-1:kernel_y-1+debug_shape[1]]=debug_weights"
189+
]
190+
},
191+
{
192+
"cell_type": "code",
193+
"execution_count": 11,
194+
"metadata": {
195+
"collapsed": false,
196+
"deletable": true,
197+
"editable": true
198+
},
199+
"outputs": [],
200+
"source": [
201+
"import matplotlib.pyplot as plt\n",
202+
"%matplotlib inline"
203+
]
204+
},
205+
{
206+
"cell_type": "code",
207+
"execution_count": 12,
208+
"metadata": {
209+
"collapsed": false,
210+
"deletable": true,
211+
"editable": true
212+
},
213+
"outputs": [
214+
{
215+
"data": {
216+
"text/plain": [
217+
"<matplotlib.image.AxesImage at 0x7f6ce875cac8>"
218+
]
219+
},
220+
"execution_count": 12,
221+
"metadata": {},
222+
"output_type": "execute_result"
223+
},
224+
{
225+
"data": {
226+
"image/png": "iVBORw0KGgoAAAANSUhEUgAAASYAAAD8CAYAAADaFgknAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhklSO4ZJUjuGSVI7/wPX24/f+sdr0AAAAABJRU5ErkJggg==\n",
227+
"text/plain": [
228+
"<matplotlib.figure.Figure at 0x7f6ceaa4e0f0>"
229+
]
230+
},
231+
"metadata": {},
232+
"output_type": "display_data"
233+
}
234+
],
235+
"source": [
236+
"plt.imshow((padded>0).sum(axis=-1))"
237+
]
238+
},
239+
{
240+
"cell_type": "code",
241+
"execution_count": 13,
242+
"metadata": {
243+
"collapsed": true,
244+
"deletable": true,
245+
"editable": true
246+
},
247+
"outputs": [],
248+
"source": [
249+
"backward_weights=weights[::-1,::-1].transpose((0,1,3,2))"
250+
]
251+
},
252+
{
253+
"cell_type": "code",
254+
"execution_count": 14,
255+
"metadata": {
256+
"collapsed": false,
257+
"deletable": true,
258+
"editable": true
259+
},
260+
"outputs": [],
261+
"source": [
262+
"grads=forward(padded, backward_weights)"
263+
]
264+
},
265+
{
266+
"cell_type": "code",
267+
"execution_count": 15,
268+
"metadata": {
269+
"collapsed": false,
270+
"deletable": true,
271+
"editable": true
272+
},
273+
"outputs": [
274+
{
275+
"data": {
276+
"text/plain": [
277+
"(10, 11, 5)"
278+
]
279+
},
280+
"execution_count": 15,
281+
"metadata": {},
282+
"output_type": "execute_result"
283+
}
284+
],
285+
"source": [
286+
"grads.shape"
287+
]
288+
},
289+
{
290+
"cell_type": "code",
291+
"execution_count": 18,
292+
"metadata": {
293+
"collapsed": false,
294+
"deletable": true,
295+
"editable": true
296+
},
297+
"outputs": [],
298+
"source": [
299+
"eps=1e-4\n",
300+
"tol=1e-8\n",
301+
"\n",
302+
"for idx in np.ndindex(im.shape):\n",
303+
" d_im=im.copy()\n",
304+
" d_im[idx]+=eps\n",
305+
" d_out=forward(d_im, weights)\n",
306+
" d_debug_out=np.sum(d_out*debug_weights)\n",
307+
"\n",
308+
" grad=(d_debug_out-debug_out)/eps\n",
309+
" \n",
310+
" assert np.abs(grad-grads[idx])<tol, idx"
311+
]
312+
},
313+
{
314+
"cell_type": "code",
315+
"execution_count": null,
316+
"metadata": {
317+
"collapsed": true
318+
},
319+
"outputs": [],
320+
"source": []
321+
}
322+
],
323+
"metadata": {
324+
"kernelspec": {
325+
"display_name": "Python 3",
326+
"language": "python",
327+
"name": "python3"
328+
},
329+
"language_info": {
330+
"codemirror_mode": {
331+
"name": "ipython",
332+
"version": 3
333+
},
334+
"file_extension": ".py",
335+
"mimetype": "text/x-python",
336+
"name": "python",
337+
"nbconvert_exporter": "python",
338+
"pygments_lexer": "ipython3",
339+
"version": "3.5.2"
340+
}
341+
},
342+
"nbformat": 4,
343+
"nbformat_minor": 2
344+
}

0 commit comments

Comments
 (0)