Interactive Interior Design Recommendation via Coarse-to-fine Multimodal Reinforcement Learning

TL;DR

This paper introduces an interactive reinforcement learning system for interior design that personalizes recommendations by actively engaging with users and effectively capturing their preferences, leading to higher satisfaction.

Contribution

The paper proposes a novel interactive RL-based interior design recommendation system with a coarse-to-fine policy network and object-aware feedback generation for improved personalization.

Findings

Outperforms traditional methods in recommendation accuracy.

Achieves higher user satisfaction in real-world tests.

Demonstrates effective preference mining through interaction.

Abstract

Personalized interior decoration design often incurs high labor costs. Recent efforts in developing intelligent interior design systems have focused on generating textual requirement-based decoration designs while neglecting the problem of how to mine homeowner's hidden preferences and choose the proper initial design. To fill this gap, we propose an Interactive Interior Design Recommendation System (IIDRS) based on reinforcement learning (RL). IIDRS aims to find an ideal plan by interacting with the user, who provides feedback on the gap between the recommended plan and their ideal one. To improve decision-making efficiency and effectiveness in large decoration spaces, we propose a Decoration Recommendation Coarse-to-Fine Policy Network (DecorRCFN). Additionally, to enhance generalization in online scenarios, we propose an object-aware feedback generation method that augments model training with diversified and dynamic textual feedback. Extensive experiments on a real-world dataset demonstrate our method outperforms traditional methods by a large margin in terms of recommendation accuracy. Further user studies demonstrate that our method reaches higher real-world user satisfaction than baseline methods.