Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update alibi docs #3092

Merged
merged 1 commit into from
Mar 29, 2021
Merged

Update alibi docs #3092

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 15 additions & 0 deletions doc/source/analytics/explainers.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,21 @@ Seldon provides model explanations using its [Alibi](https://github.com/SeldonIO

We provide [an example notebook](../examples/explainer_examples.html) showing how to deploy an explainer for Tabular, Text and Image models.


## Available Methods

Seldon Core supports a subset of the methods currently available in [Alibi](https://github.com/SeldonIO/alibi). Presently this the following:


| Method | Explainer Key | Example |
|--------|---------------|---------||
| [Anchor Tabular](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) | `AnchorTabular` | [Example](https://docs.seldon.io/projects/seldon-core/en/latest/examples/explainer_examples.html#Income-Prediction-Model-with-Anchors-Explainer) |
| [Anchor Text](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) | `AnchorText` | [Example](https://docs.seldon.io/projects/seldon-core/en/latest/examples/explainer_examples.html#Movie-Sentiment-Model) |
| [Anchor Images](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) | `AnchorImages` | [Example](https://docs.seldon.io/projects/seldon-core/en/latest/examples/explainer_examples.html#Tensorflow-CIFAR10-Model) |
| [kernel Shap](https://docs.seldon.io/projects/alibi/en/latest/methods/KernelSHAP.html) | `KernelShap` | [Example](https://docs.seldon.io/projects/seldon-core/en/latest/examples/explainer_examples.html#Wine-Prediction-Model-with-Shap-Explainer) |
| [Integrated Gradients](https://docs.seldon.io/projects/alibi/en/latest/methods/IntegratedGradients.html) | `IntegratedGradients` | [Example](https://docs.seldon.io/projects/seldon-core/en/latest/examples/explainer_examples.html#MNIST-Model-with-Integrated-Gradients-Explainer) |
| [Tree Shap](https://docs.seldon.io/projects/alibi/en/latest/methods/TreeSHAP.html) | `TreeShap` | [Example](https://docs.seldon.io/projects/seldon-core/en/latest/examples/explainer_examples.html#XGBoost-Model-with-TreeShap-Explainer) |

## Creating your explainer

For Alibi explainers that need to be trained you should
Expand Down
2 changes: 1 addition & 1 deletion doc/source/examples/notebooks.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -83,7 +83,7 @@ Advanced Machine Learning Insights
:titlesonly:

Real time monitoring of statistical metrics <feedback_reward_custom_metrics>
Tabular, Text and Image Model Explainers <explainer_examples>
Model Explainers <explainer_examples>
Outlier Detection on CIFAR10 <outlier_cifar10>

Batch Processing with Seldon Core
Expand Down
16 changes: 8 additions & 8 deletions doc/source/python/api/seldon_core.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,14 +25,6 @@ seldon\_core.api\_tester module
:undoc-members:
:show-inheritance:

seldon\_core.app module
-----------------------

.. automodule:: seldon_core.app
:members:
:undoc-members:
:show-inheritance:

seldon\_core.batch\_processor module
------------------------------------

Expand All @@ -49,6 +41,14 @@ seldon\_core.flask\_utils module
:undoc-members:
:show-inheritance:

seldon\_core.gunicorn\_utils module
-----------------------------------

.. automodule:: seldon_core.gunicorn_utils
:members:
:undoc-members:
:show-inheritance:

seldon\_core.imports\_helper module
-----------------------------------

Expand Down
40 changes: 21 additions & 19 deletions notebooks/explainer_examples.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,29 +5,31 @@
"metadata": {},
"source": [
"# Example Model Explanations with Seldon\n",
"\n",
"Seldon core supports various out-of-the-box explainers that leverage the [Alibi ML Expalinability](https://github.com/SeldonIO/alibi) open source library.\n",
"\n",
"In this notebook we show how you can use the pre-packaged explainer functionality that simplifies the creation of advanced AI model explainers.\n",
"\n",
"Seldon provides the following out-of-the-box pre-packaged explainers:\n",
"* Anchor Tabular Explainer \n",
" * A black box Explainer that uses the [anchor technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for tabular data\n",
" * It basically answers the question of what are the most \"powerul\" or \"important\" features in a tabular prediction\n",
"* Anchor Image Explainer\n",
" * A black box Explainer that uses the [anchor technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for image data\n",
" * It basically answers the question of what are the most \"powerul\" or \"important\" pixels in an image prediction\n",
"* Anchor Text Explainer\n",
" * A black box Explainer that uses the [anchor technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for text data\n",
" * It basically answers the question of what are the most \"powerul\" or \"important\" tokens in a text prediction\n",
"* Kernel Shap Explainer\n",
" * A black box Explainer that uses the [kernel shap technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for tabular data\n",
" * It provides postive and negative feature attributions that contributed to the predictions\n",
"* Integrated Gradient Explainer\n",
" * A white box explainer that uses the [Integrated Gradients technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for Keras models\n",
" * It provides importance values for each feature\n",
"* Tree Shap Explainer\n",
" * A white box explainer that uses the [TreeShap technqiue](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for tree based models\n",
" * It provides positive and negative feature attributions that contributed to the predictions"
"\n",
" - Anchor Tabular Explainer \n",
" - A black box Explainer that uses the [anchor technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for tabular data\n",
" - It basically answers the question of what are the most \"powerul\" or \"important\" features in a tabular prediction\n",
" - Anchor Image Explainer\n",
" - A black box Explainer that uses the [anchor technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for image data\n",
" - It basically answers the question of what are the most \"powerul\" or \"important\" pixels in an image prediction\n",
" - Anchor Text Explainer\n",
" - A black box Explainer that uses the [anchor technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for text data\n",
" - It basically answers the question of what are the most \"powerul\" or \"important\" tokens in a text prediction\n",
" - Kernel Shap Explainer\n",
" - A black box Explainer that uses the [kernel shap technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for tabular data\n",
" - It provides postive and negative feature attributions that contributed to the predictions\n",
" - Integrated Gradient Explainer\n",
" - A white box explainer that uses the [Integrated Gradients technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for Keras models\n",
" - It provides importance values for each feature\n",
" - Tree Shap Explainer\n",
" - A white box explainer that uses the [TreeShap technique](https://docs.seldon.io/projects/alibi/en/latest/methods/Anchors.html) for tree based models\n",
" - It provides positive and negative feature attributions that contributed to the predictions"
]
},
{
Expand Down Expand Up @@ -2474,7 +2476,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.7.9"
"version": "3.6.8"
},
"varInspector": {
"cols": {
Expand Down