SPRING: Situated Conversation Agent Pretrained with Multimodal Questions from Incremental Layout Graph

TL;DR

SPRING is a multimodal conversation agent pretrained with questions from incremental layout graphs, enabling better reasoning of spatial relations and attributes in complex scenarios, leading to superior performance on benchmark datasets.

Contribution

The paper introduces a novel pretraining method using multimodal questions from incremental layout graphs to enhance reasoning in situated conversation agents.

Findings

Outperforms state-of-the-art on SIMMC 1.0 and SIMMC 2.0 datasets.

Effectively models multi-hop spatial relations and visual attributes.

Uses curriculum learning with automatically annotated difficulty levels.

Abstract

Existing multimodal conversation agents have shown impressive abilities to locate absolute positions or retrieve attributes in simple scenarios, but they fail to perform well when complex relative positions and information alignments are involved, which poses a bottleneck in response quality. In this paper, we propose a Situated Conversation Agent Petrained with Multimodal Questions from INcremental Layout Graph (SPRING) with abilities of reasoning multi-hops spatial relations and connecting them with visual attributes in crowded situated scenarios. Specifically, we design two types of Multimodal Question Answering (MQA) tasks to pretrain the agent. All QA pairs utilized during pretraining are generated from novel Incremental Layout Graphs (ILG). QA pair difficulty labels automatically annotated by ILG are used to promote MQA-based Curriculum Learning. Experimental results verify the SPRING's effectiveness, showing that it significantly outperforms state-of-the-art approaches on both SIMMC 1.0 and SIMMC 2.0 datasets.