This software is developed for the ERA group at TUM. It provides simple frameworks for SuS, BUS, aBUS, CE-method, SIS and FORM in matlab and python.
For further information visit: www.era.bgu.tum.de/software
Software developed for ERA Group (TUM)
Home Page: https://www.era.bgu.tum.de/software
This software is developed for the ERA group at TUM. It provides simple frameworks for SuS, BUS, aBUS, CE-method, SIS and FORM in matlab and python.
For further information visit: www.era.bgu.tum.de/software
Write a proper description with input/output/reference for all files:
Please make sure the format is consistent throughout all files:
import numpy as np
and import scipy as sp
##
to # %%
samplesX
directly (i.e. samplesX[i]
instead of samplesX['total'][i]
)Edit the EM algorithm according to Sebastian's template.
I get this bug:
C:\Users\Willer\Anaconda3\lib\site-packages\numpy\core\_methods.py:70: RuntimeWarning: invalid value encountered in double_scalars ret = ret.dtype.type(ret / rcount)
c:\Users\Willer\Desktop\ERA-software\03_aBUS\python\aCS_BUS.py:109: RuntimeWarning: invalid value encountered in minimum mu_acc[i] = mu_acc[i] + np.minimum(1, np.mean(acc[idx+1:idx+Nchain[k]])) # min problem
c:\Users\Willer\Desktop\ERA-software\03_aBUS\python\aCS_BUS.py:120: RuntimeWarning: invalid value encountered in minimum sigma = np.minimum(lam[i+1]*sigma_0, np.ones((n))) # Ref. 1 Eq. 23
Probably something with aCS? Can you please check it out, Matthias?
Implement random initialization of EMvMFNM also in Python
Translate the vMFNM to Python
Edit CE and SIS such that they take nGM/k
as an input. Call it k_init
.
Add Examples 3, 4 and 5 in all the reliability methods
Matlab:
Python:
Python does that by default!
Hi Matthias,
please put also as output of the aBUS functions (matlab and python):
As a final final test, please compare the results of the matlab/python code with those of the Example 2 of Wolfgang's paper (Table 1). It is easy to implement and we can finally check the code.
Thanks a lot,
Should we add a transformation in SuS? Until now we only have samples in X-space.
And be aware of the fact that shape (100,) != (100,1). Use reshape()
!
Check the theory of aBUS and compare it with algorithm 5!
All examples of CEIS do not converge to the correct solution.
Additionally in the kmeans algorithm of EMGM throws warnings, if one cluster is empty.
The llh[t] value doesn't change in python
In main_example_2DOF.py
fzero
is substituted by scipy.optimize.fsolve
but doesn't lead to the same solution.
Something seems not quite accurate. Maybe recheck the EMGM algorithm
Could not find the the python equivalent to gamcdf
(matlab) in main_example_2.py
The EMGM needs improvement, such that it works for other diemensions than d=2.
Do I miss one level, if I just take all samples in 1:l
?
Is it on purpose that the ERANataf.pdf()
(Matlab) is called ERANataf.jointpdf()
in Python?
The Matlab version of SIS doesn't converge to the correct solution in Example 2.
And it takes very long in python. But the Pf seems to be more accurate.
The example file of BUS sometimes produces some weird warnings/errors. Check on that!
If I run the BUS_SuS example, I sometimes get this error message:
c:\Users\Willer\Desktop\ERA-software\02_BUS\python\aCS.py:97: RuntimeWarning: invalid value encountered in less_equal if He <= b:
It probably has something to do with the np.percentile()
function?
The transformation is missing in the CE-method. Use the ERANataf
-class to transform the LSF in the Gaussian space!
Set up other example cases for CE and SIS.
Check the example with 12D and also the example for 1D for aBUS. The results seem weird.
If everything else is ready, again run all examples:
Matlab:
Python:
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.