Paper ID: 2210.13984

Abductive Action Inference

Clement Tan, Chai Kiat Yeo, Cheston Tan, Basura Fernando

Abductive reasoning aims to make the most likely inference for a given set of incomplete observations. In this paper, we introduce a novel research task known as "abductive action inference" which addresses the question of which actions were executed by a human to reach a specific state shown in a single snapshot. The research explores three key abductive inference problems: action set prediction, action sequence prediction, and abductive action verification. To tackle these challenging tasks, we investigate various models, including established ones such as Transformers, Graph Neural Networks, CLIP, BLIP, GPT3, end-to-end trained Slow-Fast, Resnet50-3D, and ViT models. Furthermore, the paper introduces several innovative models tailored for abductive action inference, including a relational graph neural network, a relational bilinear pooling model, a relational rule-based inference model, a relational GPT-3 prompt method, and a relational Transformer model. Notably, the newly proposed object-relational bilinear graph encoder-decoder (BiGED) model emerges as the most effective among all methods evaluated, demonstrating good proficiency in handling the intricacies of the Action Genome dataset. The contributions of this research offer significant progress toward comprehending the implications of human actions and making highly plausible inferences concerning the outcomes of these actions.

Submitted: Oct 24, 2022