Causal Intersectionality and Dual Form of Gradient Descent for Multimodal Analysis: a Case Study on Hateful Memes

arXiv:2308.11585v2 Announce Type: replace-cross
Abstract: Amidst the rapid expansion of Machine Learning (ML) and Large Language Models (LLMs), understanding the semantics within their mechanisms is vital. Causal analyses define semantics, while gradient-based methods are essential to eXplainable AI (XAI), interpreting the model's 'black box'. Integrating these, we investigate how a model's mechanisms reveal its causal effect on evidence-based decision-making. Research indicates intersectionality - the combined impact of an individual's demographics - can be framed as an Average Treatment Effect (ATE). …

abstract analysis arxiv case case study causal cs.ai cs.cl expansion explainable ai form gradient language language models large language large language models llms machine machine learning memes multimodal semantics study type understanding vital xai