WebSHAP: Towards Explaining Any Machine Learning Models Anywhere

crown jewel figure
WebSHAP is the first open-source tool for explaining any machine learning (ML) models in browsers. Adopting the state-of-the-art explainability technique Kernel SHAP to the Web. WebSHAP offers private, ubiquitous, and interactive explanations. For example, researchers can use WebSHAP to build a client-side application that explains ML-based loan approval decisions to applicants, increasing their trust in ML models. (A) With this application, users can experiment with different feature inputs through the Web UI. (B) The application then updates the model prediction via in-browser inference. (C) WebSHAP leverages modern Web technologies to compute feature importance in real time, delivering interactive and engaging explanations.
Demo Video
As machine learning (ML) is increasingly integrated into our every- day Web experience, there is a call for transparent and explainable web-based ML. However, existing explainability techniques often require dedicated backend servers, which limit their usefulness as the Web community moves toward in-browser ML for lower latency and greater privacy. To address the pressing need for a client-side explainability solution, we present WebSHAP, the first in-browser tool that adapts the state-of-the-art model-agnostic explainability technique SHAP to the Web environment. Our open-source tool is developed with modern Web technologies such as WebGL that leverage client-side hardware capabilities and make it easy to integrate into existing Web ML applications. We demonstrate WebSHAP in a usage scenario of explaining ML-based loan approval decisions to loan applicants. Reflecting on our work, we discuss the opportunities and challenges for future research on transparent Web ML. WebSHAP is available at https://github.com/poloclub/webshap.
WebSHAP: Towards Explaining Any Machine Learning Models Anywhere
  title = {{{WebSHAP}}: {{Towards Explaining Any Machine Learning Models Anywhere}}},
  shorttitle = {{{WebSHAP}}},
  booktitle = {Companion {{Proceedings}} of the {{Web Conference}} 2023},
  author = {Wang, Zijie J. and Chau, Duen Horng},
  year = {2023},
  doi = {10.1145/3543873.3587362},