Comments (5)
I would like to see these changes incorporated as well. Is this project still
maintained?
Original comment by [email protected]
on 24 Dec 2014 at 9:56
from neurolab.
Yes, this project is maintaine. I look your code, its good job!
But I think that the regularization need to move out of error function.
I'll thinck how do it better.
Thanks for your work.
Original comment by [email protected]
on 24 Dec 2014 at 10:20
- Changed state: Started
from neurolab.
About CEE
Do you really sure in the method of calculating the derivateve of CCE function?
What sources did you use?
Original comment by [email protected]
on 25 Dec 2014 at 4:15
from neurolab.
"I'll think how do it better."
Great, thanks for maintaining this package!
"About CEE. Do you really sure in the method of calculating the derivateve of
CCE function?
What sources did you use?"
I did not perform the calculation myself. I was inspired by knowing that the
gradient descent learning rule for logistic regression (which uses CEE) is the
same as that for linear regression using MSE or SSE (as explained in this
document: http://cs229.stanford.edu/notes/cs229-notes1.pdf). I performed the
following experiment to determine whether CEE was working properly:
https://github.com/kwecht/NeuroLab/blob/master/test_cee.py
In summary: I show that a neural network using CEE, one logistic-sigmoid output
node, and no hidden layers produces the same results as logistic regression.
The optimal cost function value is the same to 6 significant figures, and
network weights/regression coefficients are the same to 3 significant digits. I
use this as confirmation that CEE and the CEE derivative are working properly.
If they were not, I don't know how I could get the same results.
This results of the experiment above was convincing to me, but if the results
are not definitive evidence of the correct CEE and derivative, please let me
know, I would like to understand why.
Kevin
Original comment by [email protected]
on 31 Dec 2014 at 12:23
from neurolab.
I added regularization to Neurolob v 0.3.5 and CEE.
For regularization, I did't change error funcs. All changes in train process.
Example
http://nbviewer.ipython.org/urls/neurolab.googlecode.com/svn/trunk/example/Neuro
lab%20-%20Regularization.ipynb
Thx Kevin. I referred you in the CHANGELOG
Original comment by [email protected]
on 26 Jan 2015 at 4:35
- Changed state: Fixed
from neurolab.
Related Issues (20)
- Failing to add Levenberg-Marquardt-training HOT 5
- Problem when using a modified network property HOT 1
- feedforward network not learning HOT 12
- PureLin in outputl layer does not work HOT 2
- Setup issue HOT 3
- strange result HOT 13
- Cannot install successfully for Python 3.2 HOT 5
- Learning Rate is not present HOT 2
- Multiprocessing can not pickle unbound function HOT 6
- cannot save nn HOT 2
- output norm and different resutls HOT 1
- Citing neurolab HOT 3
- Parameters Ignored in Training Function Construction HOT 2
- Training Fails for Non-Default Activation Functions HOT 1
- Linear Activation Leads to NaN minmax HOT 2
- Support for weight decay (regularization parameter) HOT 2
- Missing newelm example in doc HOT 3
- fmin_bfgs() got an unexpected keyword argument 'lr', train func does not take lr as parameter HOT 2
- 0.3.5 version not available in pypi HOT 1
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 neurolab.