FazBrowse GitHub Viewer
|
Trending
|
URL:
|
Home
Tools:
[Download Repo ZIP]
[View Raw Code]
[Original HTTPS Page]
arrayfire-python/arrayfire/ml.py at master · arrayfire/arrayfire-python · GitHub
Uh oh!
There was an error while loading.
Please reload this page
.
arrayfire
/
arrayfire-python
Public
Notifications
You must be signed in to change notification settings
Fork
64
Star
422
Code
Issues
57
Pull requests
3
Actions
Projects
Wiki
Security and quality
0
Insights
Additional navigation options
Code
Issues
Pull requests
Actions
Projects
Wiki
Security and quality
Insights
Expand file tree
Breadcrumbs
arrayfire-python
/
arrayfire
/
ml.py
Copy path
More file actions
More file actions
Latest commit
History
History
History
83 lines (63 loc) · 2.9 KB
Breadcrumbs
arrayfire-python
/
arrayfire
/
ml.py
Copy path
File metadata and controls
83 lines (63 loc) · 2.9 KB
Raw
Copy raw file
Download raw file
Open symbols panel
Edit and raw actions
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
#######################################################
# Copyright (c) 2020, ArrayFire
# All rights reserved.
#
# This file is distributed under 3-clause BSD license.
# The complete license agreement can be obtained at:
# http://arrayfire.com/licenses/BSD-3-Clause
########################################################
"""
Machine learning functions
- Pool 2D, ND, maxpooling, minpooling, meanpooling
- Forward and backward convolution passes
"""
from
.
library
import
*
from
.
array
import
*
def
convolve2GradientNN
(
incoming_gradient
,
original_signal
,
original_kernel
,
convolved_output
,
stride
=
(
1
,
1
),
padding
=
(
0
,
0
),
dilation
=
(
1
,
1
),
gradType
=
CONV_GRADIENT
.
DEFAULT
):
"""
Function for calculating backward pass gradient of 2D convolution.
This function calculates the gradient with respect to the output of the
\r
ef convolve2NN() function that uses the machine learning formulation
for the dimensions of the signals and filters
Multiple signals and filters can be batched against each other, however
their dimensions must match.
Example:
Signals with dimensions: d0 x d1 x d2 x Ns
Filters with dimensions: d0 x d1 x d2 x Nf
Resulting Convolution: d0 x d1 x Nf x Ns
Parameters
-----------
incoming_gradient: af.Array
- Gradients to be distributed in backwards pass
original_signal: af.Array
- A 2 dimensional signal or batch of 2 dimensional signals.
original_kernel: af.Array
- A 2 dimensional kernel or batch of 2 dimensional kernels.
convolved_output: af.Array
- output of forward pass of convolution
stride: tuple of ints. default: (1, 1).
- Specifies how much to stride along each dimension
padding: tuple of ints. default: (0, 0).
- Specifies signal padding along each dimension
dilation: tuple of ints. default: (1, 1).
- Specifies how much to dilate kernel along each dimension before convolution
Returns
--------
output: af.Array
- Gradient wrt/requested gradient type
"""
output
=
Array
()
stride_dim
=
dim4
(
stride
[
0
],
stride
[
1
])
padding_dim
=
dim4
(
padding
[
0
],
padding
[
1
])
dilation_dim
=
dim4
(
dilation
[
0
],
dilation
[
1
])
safe_call
(
backend
.
get
().
af_convolve2_gradient_nn
(
c_pointer
(
output
.
arr
),
incoming_gradient
.
arr
,
original_signal
.
arr
,
original_kernel
.
arr
,
convolved_output
.
arr
,
2
,
c_pointer
(
stride_dim
),
2
,
c_pointer
(
padding_dim
),
2
,
c_pointer
(
dilation_dim
),
gradType
.
value
))
return
output
Back
|
FazBrowse Home
|
New Git URL