Inference of Evolving Mental States from Irregular Action Events to Understand Human Behaviors
Keywords: temporal point process, logic rule, human-AI collaboration
Abstract: Inference of latent human mental processes, such as belief, intention, or desire, is crucial for developing AI with human-like intelligence, enabling more effective and timely collaboration. In this paper, we introduce a versatile encoder-decoder model designed to infer evolving mental processes based on irregularly observed action events and predict future occurrences. The primary challenges arise from two factors: both actions and mental processes are irregular events, and the observed action data is often limited. To address the irregularity of these events, we leverage a temporal point process model within the encoder-decoder framework, effectively capturing the dynamics of both action and mental events. Additionally, we implement a backtracking mechanism in the decoder to enhance the accuracy of predicting future actions and evolving mental states. To tackle the issue of limited data, our model incorporates logic rules as priors, enabling accurate inferences from just a few observed samples. These logic rules can be refined and updated as needed, providing flexibility to the model. Overall, our approach enhances the understanding of human behavior by predicting when actions will occur and how mental processes evolve. Experiments on both synthetic and real-world datasets demonstrate the strong performance of our model in inferring mental states and predicting future actions, contributing to the development of more human-centric AI systems.
Primary Area: interpretability and explainable AI
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 13211
Loading