Science Links Japan | Developmental Word Grounding through a Growing Neural Network with a Humanoid Robot

Developmental Word Grounding through a Growing Neural Network with a Humanoid Robot

Accession number;05A0661051

Title;Developmental Word Grounding through a Growing Neural Network with a Humanoid Robot

Author; KOJIMA RYO (Tokyo Inst. of Technol.) HASEGAWA OSAMU (Tokyo Inst. of Technol.) HASEGAWA OSAMU (Jst-presto)

Journal Title;Proceedings of the Annual Conference on JSAI (CD-ROM)

Journal Code:X0580B

ISSN:

VOL.19th;NO.;PAGE.1B3-03(2005)

Figure&Table&Reference;FIG.5, TBL.1, REF.5

Pub. Country;Japan

Language;Japanese

Abstract;This paper presents an unsupervised approach of integrating speech and visual information without using any prepared data, which enables a humanoid robot to learn words with their meanings. The approach is different from most other existing approaches in that it learns online from audio-visual input, rather than from stationary data provided in advance. In addition, it is capable of learning incrementally which is considered to be indispensable to lifelong learning. A noise-robust self-organized growing neural network is developed to represent the topological structure of unsupervised online data. We are also developing an active learning mechanism, called "desire for knowledge", to let the robot select the object with the least information for subsequent learning. Experimental results show that it makes the learning process more efficient. (author abst.)

Related Articles;