Deep metric learning through local attention and generative loss

• Project/Area Number: 21K17761
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 61010:Perceptual information processing-related
• Research Institution: Tokyo University of Agriculture and Technology
• Principal Investigator: NGUYENTUAN CUONG 東京農工大学, 工学(系)研究科(研究院), 特任助教 (10814246)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Project Status: Granted (Fiscal Year 2021)
• Budget Amount: ¥4,420,000 (Direct Cost: ¥3,400,000、Indirect Cost: ¥1,020,000); Fiscal Year 2023: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000); Fiscal Year 2022: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000); Fiscal Year 2021: ¥2,860,000 (Direct Cost: ¥2,200,000、Indirect Cost: ¥660,000)
• Keywords: deep metric learning / clustering / mathematical expression / handwriting

Outline of Research at the Start

Learning structural representation of data samples through their sub-structures and the relations between these sub-structures is efficient since the sub-structures and relations are shared among the data samples. As supervised learning models requires annotations of these sub-structures and relations, which is costly and unfeasible for some type of data, automatic learning of these sub-structures and relations are necessary. We propose a deep metric learning method based on attention generative model for automatically focus on learning the discriminative local features as the sub-structures.