Comments (2)
Thanks for your comments. This code is experimental and there were numerous bugs. I just fixed some of them. Now BWN
should be around 97%
and XNOR
around 96%
.
Regarding to the alpha
issue, it is indeed computed in the forward and backward of BinaryConvolution
layer but not in the BinaryActivation
layer. The scale factor of the binary inputs is not considered in this code, which is the parameter beta
in equation (10) of the XNOR-Net paper. If you included beta
, it should produce better results.
from xnor-net.
Just updated it again! With Adam
optimizer and lr=0.0002
, the results are better.
python main.py --network=mnist_xnor
Namespace(batch_size=100, data_path='../data/MNIST_data/', learning_rate=0.1, momentum=0.9, network='mnist_xnor', num_epoch=5)
INFO:root:Epoch[0] Batch [200] Speed: 2333.27 samples/sec accuracy=0.893284
INFO:root:Epoch[0] Batch [400] Speed: 2348.61 samples/sec accuracy=0.973850
INFO:root:Epoch[0] Train-accuracy=0.976533
INFO:root:Epoch[0] Time cost=25.585
INFO:root:Epoch[0] Validation-accuracy=0.979300
INFO:root:Epoch[1] Batch [200] Speed: 2370.62 samples/sec accuracy=0.979751
INFO:root:Epoch[1] Batch [400] Speed: 2381.95 samples/sec accuracy=0.983550
INFO:root:Epoch[1] Train-accuracy=0.982764
INFO:root:Epoch[1] Time cost=25.247
INFO:root:Epoch[1] Validation-accuracy=0.980700
INFO:root:Epoch[2] Batch [200] Speed: 2384.56 samples/sec accuracy=0.982736
INFO:root:Epoch[2] Batch [400] Speed: 2369.69 samples/sec accuracy=0.985950
INFO:root:Epoch[2] Train-accuracy=0.984472
INFO:root:Epoch[2] Time cost=25.245
INFO:root:Epoch[2] Validation-accuracy=0.982100
INFO:root:Epoch[3] Batch [200] Speed: 2362.13 samples/sec accuracy=0.985920
INFO:root:Epoch[3] Batch [400] Speed: 2351.47 samples/sec accuracy=0.987700
INFO:root:Epoch[3] Train-accuracy=0.986834
INFO:root:Epoch[3] Time cost=25.455
INFO:root:Epoch[3] Validation-accuracy=0.980200
INFO:root:Epoch[4] Batch [200] Speed: 2355.15 samples/sec accuracy=0.986169
INFO:root:Epoch[4] Batch [400] Speed: 2338.48 samples/sec accuracy=0.987750
INFO:root:Epoch[4] Train-accuracy=0.988342
INFO:root:Epoch[4] Time cost=25.685
INFO:root:Epoch[4] Validation-accuracy=0.982600
Testing accuracy: 98.26%
from xnor-net.
Related Issues (2)
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from xnor-net.