Information-theoretic Approaches for Causal Reasoning
Interdisciplinary Areas: | Data and Engineering Applications, Autonomous and Connected Systems |
---|
Project Description:
Causal reasoning capabilities are critical to the development of artificial intelligence. Causal reasoning also has applications in generalizable supervised machine learning, explainable decision-making, and algorithmic fairness Recent work demonstrated that information-theoretic approaches extend the known limitations of the existing causal inference algorithms under suitable assumptions [1, 2]. In this project, we will establish novel connections between information theory and causality, and their applications in machine learning.
Start Date:
January 2023
Postdoc Qualifications:
The desired candidate would have Ph.D. in ECE, CS, Stats, or related areas, with experience with both mathematical proofs as well as programming. Top-tier papers related to information theory, causality, or machine learning will be preferred.
Co-Advisors:
Murat Kocaoglu
School of Electrical and Computer Engineering
mkocaoglu@purdue.edu
https://www.muratkocaoglu.com/
Vaneet Aggarwal
School of Industrial Engineering
vaneet@purdue.edu
https://web.ics.purdue.edu/~vaneet/
Bibliography:
[1] https://aaai.org/ocs/index.php/AAAI/AAAI17/paper/view/14218
[2] https://papers.nips.cc/paper/2020/file/a979ca2444b34449a2c80b012749e9cd-Paper.pdf
[3] https://www.aaai.org/AAAI22Papers/AAAI-1666.BaiQ.pdf
[4] https://proceedings.neurips.cc/paper/2018/file/39e98420b5e98bfbdc8a619bef7b8f61-Paper.pdf
[5] https://arxiv.org/pdf/2206.06469.pdf