Skip to content

Latest commit

 

History

History
88 lines (51 loc) · 2.17 KB

README.md

File metadata and controls

88 lines (51 loc) · 2.17 KB

MLADS Session 215: Using Interpretable Machine Learning for Responsible AI

This is a repo for use in our MLADs Lab in Fall 2019.

1. Access Github repo

Please use Incognito Chrome to open!

IMPORTANT: Browse to https://aka.ms/mlads215

2. Get access to the notebook VM

Registration URL : http://bit.ly/32ZYLoZ

Fill out name, email, etc. Select Launch Lab. You should be quickly taken to a page with relevant passwords.

Open the JuypterhubURL with the admin username and password.

3. Setup the kernel and environment

In the New menu, create a Python 3.5 notebook:

New menu

Select Python 3.5

Enter this in the first cell of notebook:

!git clone https://github.com/bethz/MLADs2019Fall-Interpretability.git
 
!python -m pip install azureml-contrib-interpret azureml-sdk --upgrade

Then, execute the cell by typing shift-enter.

At this point, you have the files needed for the lab.

4. Start the local notebook

From the jupyter root directory, go to:

/MLADs2019Fall-Interpretability/MLADS-Local/

Launch this file:

blackbox-simple-feature-transformations-explain-local.ipynb

5. Setup the Azure notebook

Go back to the email you received from CloudLabs and reference:

  • the On Demand Lab Azure Machine Learning ODSC username and password
  • the amlWorkspaceName

Go to portal.azure.com It will ask for username and password. Enter the username and password. Note: do NOT use the AdminUsername and adminPassword.

In the portal, search for the amlWorkspaceName that you copied from the CloudLabs email (quick-starts-ws-xxxxxx). Select the amlWorkspaceName.

download

Download config.json by selecting download config.json.

Upload it to root.

6. Launch the Azure notebook

From the jupyter root directory, go to:

/MLADs2019Fall-Interpretability/MLADS-Azure/

Launch blackbox-train-explain-model-on-amlcompute-and-deploy.ipynb

7. References

The interpret and interpret-community repos are located at:

https://github.com/interpretml/interpret

https://github.com/interpretml/interpret-community

8. For further questions feel free to reach out to either of:

cjon, eddeleon, bzeran @microsoft.com or by teams