-
Steps:
- [x] include a theano wheel,
- [x] auto-create `theano-cache.bat` , `theano-nose.bat`, `theano-test.bat`
- [x] passing the basic test
- [x] integrate it back in default build
- [x] include a t…
-
Some are already there and can just be removed in the scikit-learn repo, other have to be ported and made consistent with the ml-benchmarks interface.
-
@ogrisel @fabianp At present scores are being calculated using cross-validation data. Should this not be done using test data for both arcene and madeon?
-
scikit-learn LDA produces incorrect results when used for dimensionality reduction. I went back and implemented an LDA step-by-step, and compared my results with Matlab's and R's LDA functions on the …
rasbt updated
10 years ago
-
There's a benchmark script in `benchmarks/bench_plot_nmf.py` that compares our implementation Lin's projected gradient algorithm against Lee & Seung's old gradient descent algorithm, which it implemen…
-
See here:
http://www.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-24.pdf
-
This popped up when I ran a `CDRegressor(warm_start=True, permute=True, penalty="l1")`::
```
_lasso_fx_selection
est.fit(X, y)
File "/usr/local/lib/python2.7/dist-packages/lightning/impl/prima…
-
I've been repeating some of the recent random forest benchmarks people have done on the forest coverage dataset and have discovered a possible bug that leads to occasional nonsense splits producing la…
-
It seems that we have a memory leak when running `MinibatchKMeans` on sparse data:
``` python
import os, psutil
from sklearn.datasets import fetch_20newsgroups
from sklearn.feature_extraction.text im…
-
The properties for accessing attributes on `Tree` return arrays which don't own their data, and which are malloced and freed by the tree. Therefore:
``` python
from sklearn.tree import DecisionTreeCl…