Creation of Episodic Memory Based on Context and Facial Expressions

2016 Fiscal Year Final Research Report

• Project/Area Number: 26330312
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Kansei informatics
• Research Institution: Akita Prefectural University
• Principal Investigator: Madokoro Hirokazu 秋田県立大学, システム科学技術学部, 准教授 (10373218)
• Research Collaborator: Istvan Lovani
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: 適応共鳴理論 / エピソード記憶 / 人間共生ロボット / コンテクスト / カテゴリマップ / 表情空間 / 分類粒度評価 / 並列動作シミュレーション

Outline of Final Research Achievements

For this study, we proposed a novel method to create episodic memory adaptively and autonomously based on dynamic diversity of facial expression and its context. We conducted evaluation experiments in an actual environment using our autonomous mobile robot prototype as a platform. The proposed method actualized unsupervised learning for counter propagation networks using labels created using adaptive resonance theory network. The evaluation experiment revealed that real-time generation of facial expression maps and context maps were actualized as episodic memory using our original benchmark dataset.

Free Research Field

情報工学