chengsoonong crowdastro issues

chengsoonong / crowdastro

Cross-identification of radio objects and host galaxies by applying machine learning on crowdsourced training labels.

MIT License

13 stars 1 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Bump numpy from 1.11.0 to 1.22.0

#249 dependabot[bot] opened 2 years ago
0
Bump numpy from 1.11.0 to 1.21.0

#248 dependabot[bot] closed 2 years ago
1
Bump requests from 2.10.0 to 2.20.0

#247 dependabot[bot] opened 4 years ago
0
pip installation doesn't include crowdastro.crowd

#246 MatthewJA opened 5 years ago
1
Verify that LR weights for image features ~ LR weights for non-image features

#245 MatthewJA opened 7 years ago
0
Serialise random forests to protobuf

#244 MatthewJA opened 7 years ago
0
Fix PyPI missing submodules

#243 MatthewJA closed 7 years ago
0
Train and test classifiers

#242 MatthewJA closed 7 years ago
0
Freeze unweighted RGZ-ATLAS + Norris label sets

#241 MatthewJA closed 7 years ago
0
Train/test Norris/RGZ with logistic regression and random forests on the compact and resolved sets with 5-fold cross-validation

#240 MatthewJA closed 7 years ago
2
Make a feature ablation plot

#239 MatthewJA opened 7 years ago
1
Fix non-primary Franzen component matching to Norris

#238 jbanfield closed 7 years ago
1
Include "No IR" consensus in the big table

#237 jbanfield closed 7 years ago
1
One table to rule them all

#236 chengsoonong closed 7 years ago
0
Every source in RGZ must have a corresponding source in Franzen

#235 chengsoonong closed 7 years ago
1
Clarify why a CNN could be more useful than other methods

#234 MatthewJA opened 7 years ago
0
Find a clear application of host galaxy identification

#233 MatthewJA closed 7 years ago
2
Clarify relationship between Norris/Middelberg cross-identifications and ATLAS catalogues

#232 MatthewJA closed 7 years ago
6
Find confusion matrices to help investigate effect of label noise

#231 MatthewJA closed 7 years ago
2
Plot predicted probabilities histogram for new CNNs and investigate bias term

#230 MatthewJA opened 7 years ago
1
Compute dataset statistics to better inform decisions

#229 MatthewJA closed 7 years ago
11
Investigating normalising convolutional/non-convolutional features

#228 MatthewJA closed 7 years ago
1
Concretely choose a dataset to work with

#227 MatthewJA closed 7 years ago
2
Nearest neighbour component catalog

#226 chengsoonong opened 7 years ago
0
Investigate colour/intensity normalisation

#225 chengsoonong opened 7 years ago
4
Train on 25% and 50%

#224 chengsoonong opened 7 years ago
2
Use logistic regression final layer

#223 chengsoonong closed 7 years ago
0
Make a small test dataset for quick testing

#222 MatthewJA closed 7 years ago
1
Update code to use full Norris table

#221 MatthewJA closed 7 years ago
2
Update documentation index.md

#220 MatthewJA opened 7 years ago
0
Update results in notebooks

#219 chengsoonong opened 7 years ago
0
Updated .gitmodules to try and resolve #217

#218 MatthewJA closed 7 years ago
0
Travis builds broken due to submodule

#217 MatthewJA closed 7 years ago
1
Tidy up/delete wiki

#216 chengsoonong closed 7 years ago
1
How good are individual labelers?

#215 chengsoonong closed 7 years ago
2
Allow crowdastro.train_classifier to accept a training set or use existing training sets

#214 MatthewJA opened 7 years ago
0
Fix train/test submodules to use the new test sets

#213 MatthewJA opened 7 years ago
0
Should we extract crowdastro.crowd into its own project?

#212 chengsoonong opened 7 years ago
2
Subtract background noise from CDFS and ELAIS

#211 chengsoonong opened 7 years ago
2
Classifier without CNN features

#210 chengsoonong closed 7 years ago
4
Balance labels for passive learning test

#209 MatthewJA closed 7 years ago
1
RGZ data use

#208 MatthewJA closed 7 years ago
2
Using raw labels to train directly

#207 MatthewJA closed 7 years ago
1
Developer documentation

#206 chengsoonong opened 7 years ago
0
Investigate if the outstanding performance of LR over RF was a side-effect of CNN peeking

#205 MatthewJA closed 7 years ago
1
Use QBC to plot an active learning curve

#204 MatthewJA closed 7 years ago
0
Test effect of CNN features alone

#203 MatthewJA closed 7 years ago
0
Run feature ablation with no CNN

#202 MatthewJA closed 7 years ago
1
Flip labels on Raykar to check whether behaviour is feature-dependent

#201 MatthewJA closed 7 years ago
2
Confirm correct binary label convention for Raykar

#200 MatthewJA closed 7 years ago
1