ABSTRACT:

I’ll present research on using large language models (LLMs) to build explainable classifiers.   I will show off work from my PhD students and collaborators on several recent research directions:

• Image classification with explainable features  (https://arxiv.org/abs/2211.11158)
• Text classification with explainable features (work in progress)
• The importance of faithfulness in explanations (https://arxiv.org/abs/2209.11326)
• (Time permitting) A faithful “chain of thought” LLM reasoner that produces code in its explanations (https://arxiv.org/abs/2301.13379)

Here’s an example of the automatically generated concepts that we use for image classification in the first

The papers that I’ll present are joint work with:
Adam Stein, Ajay Patel, Ansh Kothary, Artemis Panagopoulou, Daniel Jin, Delip Rao, Eric Wong, Harry Li Zhang, Kathleen McKeown, Marianna Apidianaki, Mark Yatskar, Shenghao Zhou, Shreya Havaldar, Veronica Qing Lyu, Yue Yang, and othe