DeepDiffusion: Unsupervised Learning of Retrieval-Adapted Representations via Diffusion-Based Ranking on Latent Feature Manifold
Unsupervised learning of feature representations is a challenging yet important problem for analyzing a large collection of multimedia data that do not have semantic labels. Recently proposed neural network-based unsupervised learning approaches have succeeded in obtaining features appropriate for c...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9934898/ |
| _version_ | 1797988354405433344 |
|---|---|
| author | Takahiko Furuya Ryutarou Ohbuchi |
| author_facet | Takahiko Furuya Ryutarou Ohbuchi |
| author_sort | Takahiko Furuya |
| collection | DOAJ |
| description | Unsupervised learning of feature representations is a challenging yet important problem for analyzing a large collection of multimedia data that do not have semantic labels. Recently proposed neural network-based unsupervised learning approaches have succeeded in obtaining features appropriate for classification of multimedia data. However, unsupervised learning of feature representations adapted to content-based matching, comparison, or retrieval of multimedia data has not been explored well. To obtain such retrieval-adapted features, we introduce the idea of combining diffusion distance on a feature manifold with neural network-based unsupervised feature learning. This idea is realized as a novel algorithm called DeepDiffusion (DD). DD simultaneously optimizes two components, a feature embedding by a deep neural network and a distance metric that leverages diffusion on a latent feature manifold, together. DD relies on its loss function but not encoder architecture. It can thus be applied to diverse multimedia data types with their respective encoder architectures. Experimental evaluation using 3D shapes and 2D images demonstrates versatility as well as high accuracy of the DD algorithm. Code is available at <uri>https://github.com/takahikof/DeepDiffusion</uri> |
| first_indexed | 2024-04-11T08:02:42Z |
| format | Article |
| id | doaj.art-6e41df7e137a422097b69fca11914f17 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-11T08:02:42Z |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-6e41df7e137a422097b69fca11914f172022-12-22T04:35:40ZengIEEEIEEE Access2169-35362022-01-011011628711630110.1109/ACCESS.2022.32189099934898DeepDiffusion: Unsupervised Learning of Retrieval-Adapted Representations via Diffusion-Based Ranking on Latent Feature ManifoldTakahiko Furuya0https://orcid.org/0000-0001-9976-0330Ryutarou Ohbuchi1https://orcid.org/0000-0002-7605-9135Department of Computer Science and Engineering, University of Yamanashi, Kofu-shi, JapanDepartment of Computer Science and Engineering, University of Yamanashi, Kofu-shi, JapanUnsupervised learning of feature representations is a challenging yet important problem for analyzing a large collection of multimedia data that do not have semantic labels. Recently proposed neural network-based unsupervised learning approaches have succeeded in obtaining features appropriate for classification of multimedia data. However, unsupervised learning of feature representations adapted to content-based matching, comparison, or retrieval of multimedia data has not been explored well. To obtain such retrieval-adapted features, we introduce the idea of combining diffusion distance on a feature manifold with neural network-based unsupervised feature learning. This idea is realized as a novel algorithm called DeepDiffusion (DD). DD simultaneously optimizes two components, a feature embedding by a deep neural network and a distance metric that leverages diffusion on a latent feature manifold, together. DD relies on its loss function but not encoder architecture. It can thus be applied to diverse multimedia data types with their respective encoder architectures. Experimental evaluation using 3D shapes and 2D images demonstrates versatility as well as high accuracy of the DD algorithm. Code is available at <uri>https://github.com/takahikof/DeepDiffusion</uri>https://ieeexplore.ieee.org/document/9934898/Unsupervised representation learningmultimedia information retrievaldeep learning |
| spellingShingle | Takahiko Furuya Ryutarou Ohbuchi DeepDiffusion: Unsupervised Learning of Retrieval-Adapted Representations via Diffusion-Based Ranking on Latent Feature Manifold IEEE Access Unsupervised representation learning multimedia information retrieval deep learning |
| title | DeepDiffusion: Unsupervised Learning of Retrieval-Adapted Representations via Diffusion-Based Ranking on Latent Feature Manifold |
| title_full | DeepDiffusion: Unsupervised Learning of Retrieval-Adapted Representations via Diffusion-Based Ranking on Latent Feature Manifold |
| title_fullStr | DeepDiffusion: Unsupervised Learning of Retrieval-Adapted Representations via Diffusion-Based Ranking on Latent Feature Manifold |
| title_full_unstemmed | DeepDiffusion: Unsupervised Learning of Retrieval-Adapted Representations via Diffusion-Based Ranking on Latent Feature Manifold |
| title_short | DeepDiffusion: Unsupervised Learning of Retrieval-Adapted Representations via Diffusion-Based Ranking on Latent Feature Manifold |
| title_sort | deepdiffusion unsupervised learning of retrieval adapted representations via diffusion based ranking on latent feature manifold |
| topic | Unsupervised representation learning multimedia information retrieval deep learning |
| url | https://ieeexplore.ieee.org/document/9934898/ |
| work_keys_str_mv | AT takahikofuruya deepdiffusionunsupervisedlearningofretrievaladaptedrepresentationsviadiffusionbasedrankingonlatentfeaturemanifold AT ryutarouohbuchi deepdiffusionunsupervisedlearningofretrievaladaptedrepresentationsviadiffusionbasedrankingonlatentfeaturemanifold |