TUTORIALS

Efficient Pre-trained Model Tuning via Model Reprogramming

Website:

https://trustworthy-ml.wixsite.com/acmlmr

Abstract:

Knowledge shouldn’t be limited to those who can pay," said Robert C. May, chair of UC's Academic Senate. In machine learning, this is particularly relevant, as recent foundation models—pre-trained on massive datasets—have widened the gap between those who can afford to access and control these models and those who cannot. While these models hold promise for global challenges like healthcare and education, their size and cost make fine-tuning prohibitive, especially for institutions in developing regions. Model reprogramming (MR) offers a cost-effective solution by repurposing pre-trained models without expensive retraining. This tutorial will explain MR’s input reprogramming and output transformation, highlighting real-world applications, especially in resource-limited settings.

Organizers:

Feng Liu

The University of Melbourne & RIKEN Center for Advanced Intelligence Project (AIP)

Causality and Large Models

Website:

https://sites.google.com/view/calm-acml24

Abstract:

Our tutorial explores the synergies between causality and large models, also known as “foundation models,” which have demonstrated remarkable capabilities for helping machine learning in healthcare, finance, and education. However, there are increasing concerns about the trustworthiness and interpretability of these complex ”black-box” LLMs behind the promising performance in machine learning domains. A growing community of researchers is turning towards a more principled framework to address these concerns, better understand the behavior of large models, and improve their reliability and interpretability. Specifically, this tutorial will focus on three directions: causal agents for decision-making, LLMs for causality, and benefiting LLMs with causality. Besides, we introduce some open challenges and potential future directions for this area. We hope this tutorial will stimulate more ideas on this topic and facilitate the development of causality-aware large models.

Organizers: