Presentation 2020-12-11
An application of Differentiable Neural Architecture Search to Multimodal Neural Networks
Yushiro Funoki, Satoshi Ono,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper proposes a method that designs an architecture of a deep neural network for multimodal sequential data using a gradient-based neural structure search method named Differentiable Neural Architecture Search (DARTS). Experimental results using an emotion recognition dataset containing sequential data showed that the proposed method succeeded inautomatically designing a network structure with competitive performance to manually designed networks.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Neural architecture search / Emotion recognition / Multimodal sequential data
Paper # AI2020-11
Date of Issue 2020-12-03 (AI)

Conference Information
Committee AI
Conference Date 2020/12/10(1days)
Place (in Japanese) (See Japanese page)
Place (in English) Online and HAMAMATSU ACT CITY
Topics (in Japanese) (See Japanese page)
Topics (in English) Foundations and application technologies for AI systems on the new normal
Chair Naoki Fukuta(Shizuoka Univ.)
Vice Chair Yuichi Sei(Univ. of Electro-Comm.) / Yuko Sakurai(AIST)
Secretary Yuichi Sei(Nagoya Inst. of Tech.) / Yuko Sakurai(Tokyo Univ. of Agriculture and Technology)
Assistant

Paper Information
Registration To Technical Committee on Artificial Intelligence and Knowledge-Based Processing
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) An application of Differentiable Neural Architecture Search to Multimodal Neural Networks
Sub Title (in English)
Keyword(1) Neural architecture search
Keyword(2) Emotion recognition
Keyword(3) Multimodal sequential data
1st Author's Name Yushiro Funoki
1st Author's Affiliation Kagoshima University(Kagoshima Univ)
2nd Author's Name Satoshi Ono
2nd Author's Affiliation Kagoshima University(Kagoshima Univ)
Date 2020-12-11
Paper # AI2020-11
Volume (vol) vol.120
Number (no) AI-281
Page pp.pp.52-56(AI),
#Pages 5
Date of Issue 2020-12-03 (AI)