Explainability-Guided Model Routing

Using XAI to Predict and Minimize Compute Cost

Project Overview

This project explores how explainable AI (XAI) can make deep learning systems more compute-efficient. Instead of running every input through a large, energy-hungry model, the system uses explainability signals to predict how difficult an input is and routes it to the most efficient model that can handle it.

The core idea: the AI analyzes the image, explains what it sees, and decides whether it needs a big brain or a small one. This combines responsible AI, interpretability, and green AI efficiency.

Demo Link:

https://xai-model-routing-fdbpxcdxgeeqetarojktcd.streamlit.app/

Project summary:

https://sharmilnk.github.io/Reduce-Compute-and-Latency-using-XAI/