I will start by presenting Adaptive Testing and Debugging of NLP Models (AdaTest), where we propose a process and tool that leverages the complementary strengths of humans and language models to find and fix bugs in NLP models. I will then present a variety of follow up work that builds on similar ideas to test vision models, patch models with natural language, and avoid conflicts and interference between debugging attempts.
Marco Tulio Ribeiro is a Senior Researcher at Microsoft Research. His work is on facilitating the communication between humans and machine learning models, which includes interpretability, trust, debugging, feedback, robustness, testing, etc. He received his PhD from the University of Washington.