Skip to content

Commit

Permalink
fix maxout op latex equation (#6303)
Browse files Browse the repository at this point in the history
  • Loading branch information
kexinzhao committed Dec 5, 2017
1 parent 458ffbf commit 2c1270e
Showing 1 changed file with 20 additions and 15 deletions.
35 changes: 20 additions & 15 deletions paddle/operators/maxout_op.cc
Original file line number Diff line number Diff line change
Expand Up @@ -40,23 +40,28 @@ class MaxOutOpMaker : public framework::OpProtoAndCheckerMaker {
"the number of channels divided by groups.."
)DOC");
AddComment(R"DOC(
Assumed the input shape is (N, Ci, H, W).
The output shape is (N, Co, H, W). Then `Co = Ci / groups`.
MaxOut Operator.
math:
y_{si+j} = \max_k x_{gsi + sk + j}
g = groups
s = input.size / num_channels
0 \le i < num_channels / groups
0 \le j < s
0 \le k < groups
Assumed the input shape is (N, Ci, H, W).
The output shape is (N, Co, H, W).
Then $Co = Ci / groups$ and the operator formula is as follows:
Please refer to Paper:
- Maxout Networks: http://www.jmlr.org/proceedings/papers/v28/goodfellow13.pdf
- Multi-digit Number Recognition from Street View \
Imagery using Deep Convolutional Neural Networks: \
https://arxiv.org/pdf/1312.6082v4.pdf
)DOC");
$$
y_{si+j} = \max_k x_{gsi + sk + j} \\
g = groups \\
s = \frac{input.size}{num\_channels} \\
0 \le i < \frac{num\_channels}{groups} \\
0 \le j < s \\
0 \le k < groups
$$
Please refer to Paper:
- Maxout Networks: http://www.jmlr.org/proceedings/papers/v28/goodfellow13.pdf
- Multi-digit Number Recognition from Street View \
Imagery using Deep Convolutional Neural Networks: \
https://arxiv.org/pdf/1312.6082v4.pdf
)DOC");
}
};

Expand Down

0 comments on commit 2c1270e

Please sign in to comment.