georgemilosh / Climate-Learning

How to predict extreme events in climate using rare event algorithms and modern tools of machine learning

https://georgemilosh.github.io/Climate-Learning/

georgemilosh/Climate-Learning Issues

Compatibility with newer tensorflow versions
Updated 3 months ago
import params from external folder
Closed 5 months ago1
For the gaussian approximation on CESM matrix `W` is 5.7GB
Closed 5 months ago1
Move useful unspecific things into separate modules
Updated 5 months ago
Allow to enable and disable mods
Closed 5 months ago
cannot load kernels from composite in IIPR
Updated 6 months ago
'monitor' is ignored to compute the score
Updated 6 months ago1
`probabilistic_regression`: put the activation function for sigma in the model architecture rather than in the loss function
Closed 6 months ago
pass metrics from command line
Updated 6 months ago
flexible hypop doesn't remove kwargs at default
Closed 6 months ago1
Support for cascading inheritance
Closed 6 months ago1
`lsm2mask` should be handled better
Updated 7 months ago
CESM France mask could be better
Closed 7 months ago6
ERA_Fields.Plasim_Field.sort_lat doesn't do anything
Updated 7 months ago
It is a bad idea to write to home directory commit #d1ba925cbddd69d00bd297c4cdfc90efc8a5f1f5
Closed 6 months ago
Transfer Learning from outside folder fails
Closed 9 months ago11
dict arguments are not dealt with correctly
Closed a year ago1
Queue file
Updated a year ago
`Learn2_new.py` trainer inheritence
Updated 8 months ago1
Using `lon` and `lat` as inputs to `X` when classifying heatwaves
Updated a year ago
Leap years
Updated a year ago1
xarray friendly `ef.create_mask`
Updated a year ago2
The new runs should carry the label `R` before completion
Closed a year ago
the default behavior should be to load from a repository config file which documents where to look for the data
Updated a year ago8
Reduce the size of the repository
Updated a year ago
Write docstring for gaussian approx
Closed a year ago
Support for single fold run
Updated a year ago1
Put some error message when the year_list exceeds the size of the file we load so that it is clear what is happening
Updated 6 months ago1
Data not normalized prior to being fed into the model when computing Y_pred_unbiased?
Closed a year ago5
`setup.sh` could be outdated for the new versions of tensorflow
Closed a year ago
Testing the network before loading data
Updated a year ago1
The error messages return wrong lines for `Learn2_new.py`
Closed a year ago5
transfer learning on `u`
Updated a year ago3
seems that this condition is restrictive. I don't see a reason why we need to limit the majority class to never be smaller@georgemilosh
Updated a year ago2
Reduce memory usage!
Updated a year ago12
Ability to skip convolutional layers and support for weight decay
Closed a year ago
`field_infos` should be a json file
Closed a year ago1
`nfolds` is kwarg of both `prepare_XY` and `k_fold_cross_val` which can cause problems
Closed a year ago2
Flexible latitude in `committor_projection_NN.py`
Closed a year ago1
Working with CESM
Closed a year ago
Merge the subm2 and main
Closed a year ago1
Save the temperature anomaly so we don't need to recompute it
Closed a year ago2
Make more readable and compact notebooks
Updated 2 years ago
Implement analog method in VAE
Closed 2 years ago
More information in README.md
Updated 2 years ago
Use submodules
Closed 2 years ago2
Enable the possibility to run autoencoder on reduced space using main branch
Closed 2 years ago
Better masked average
Closed 2 years ago1
Better exploitation of xarray features
Updated 2 years ago1
More flexible choice of the data domain
Closed 2 years ago1