Jonas Mueller

icon indicating a website link https://people.csail.mit.edu/jonasmueller/

icon company Cleanlab icon location ML researcher

Results 180 comments of Jonas Mueller

for 'other workflows' datalab tutorial, show how to detect all issue types and use a custom regression model for detecting label errors

first need to address this: https://github.com/cleanlab/cleanlab/issues/920 so we have a place to store the new code

Improve efficiency and runtimes or memory usage (performance optimization)

> DO you have any handy large real-world multiannotator dataset I can use? There are some multiannotator datasets here: https://github.com/cleanlab/multiannotator-benchmarks https://github.com/cleanlab/examples/blob/master/multiannotator_cifar10/multiannotator_cifar10.ipynb

Add method to compute Out-of-bag Data Shapely value

@mahimairaja Awesome thrilled to have your help with this! I recommend contributing your new code in cleanlab.classification as part of the CleanLearning class. That way your method can use any...

add new tutorial: "Other useful Datalab workflows" (datalab_workflows.ipynb)

one issue type to highlight can be class imbalance. Can use this code for that part: ``` { "cell_type": "markdown", "metadata": {}, "source": [ "#### Class Imbalance \n", "\n", "Let's...

add new tutorial: "Other useful Datalab workflows" (datalab_workflows.ipynb)

include the ability to provide clusters for underperforming group issues (from the FAQ) into this tutorial

CI: script that runs all Colab notebooks

Certainly @Steven-Yiran we'd love your help! But be aware this issue is actually fairly complex and will require you to dive pretty deep into how our CI works and builds...

extend active learning with multiple annotators to multi-label datasets

Also requested by another user here: https://github.com/cleanlab/cleanlab/issues/932

adding an spurious_correlation as new issue type

Thanks for your contribution @01PrathamS! We will need you to sign the CLA: https://github.com/cleanlab/cleanlab/pull/872#issuecomment-1762110159 before we can review it, thanks!

adding an spurious_correlation as new issue type

Addresses: https://github.com/cleanlab/cleanlab/issues/860

adding an spurious_correlation as new issue type

> hello @jwmueller , please checkout my submitted PR [new issue type spurious_correlation changes added](https://github.com/cleanlab/cleanlab/pull/872/commits/f6af4d4ad76593b159abd0cd84b22d6de0e0ac2a) So this PR is now up-to-date and I close the [other one](https://github.com/cleanlab/cleanlab/pull/873)? I'll take a...