Scholarlink

Causal AI is reshaping machine learning (ML) by uncovering cause-effect relationships that enhance model robustness, interpretability, and generalizability. This paper presents a comprehensive review of causal inference applications in three key domains: Natural Language Processing (NLP), Computer Vision (CV), and Time-Series Analysis. We explore advanced techniques for handling confounders, encoding high-dimensional data, and leveraging causal models to address domain-specific challenges. Practical use cases in healthcare, autonomous systems, and marketing illustrate the transformative potential of causal AI, while highlighting persistent challenges, including managing unobserved confounders and ensuring out-of-domain robustness. Recent advancements in text, image, and temporal data processing are discussed, alongside emerging topics such as model fairness and causal reinforcement learning. This review synthesizes current progress, identifies search gaps, and proposes a roadmap for future developments in causal AI.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

STATE-Of-The-Art Causal AI Applications in NLP, CV, and Time-Series Data

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Asmae Elidrissi,
  • My Abdelouahed Sabri

摘要

Causal AI is reshaping machine learning (ML) by uncovering cause-effect relationships that enhance model robustness, interpretability, and generalizability. This paper presents a comprehensive review of causal inference applications in three key domains: Natural Language Processing (NLP), Computer Vision (CV), and Time-Series Analysis. We explore advanced techniques for handling confounders, encoding high-dimensional data, and leveraging causal models to address domain-specific challenges. Practical use cases in healthcare, autonomous systems, and marketing illustrate the transformative potential of causal AI, while highlighting persistent challenges, including managing unobserved confounders and ensuring out-of-domain robustness. Recent advancements in text, image, and temporal data processing are discussed, alongside emerging topics such as model fairness and causal reinforcement learning. This review synthesizes current progress, identifies search gaps, and proposes a roadmap for future developments in causal AI.