Development of learning subspace-based methods for pattern recognition
Project/Area Number |
22K17960
|
Research Category |
Grant-in-Aid for Early-Career Scientists
|
Allocation Type | Multi-year Fund |
Review Section |
Basic Section 61030:Intelligent informatics-related
|
Research Institution | National Institute of Advanced Industrial Science and Technology |
Principal Investigator |
SALESDESOUZA LINCON 国立研究開発法人産業技術総合研究所, 情報・人間工学領域, 研究員 (40912481)
|
Project Period (FY) |
2022-04-01 – 2026-03-31
|
Project Status |
Granted (Fiscal Year 2022)
|
Budget Amount *help |
¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000)
Fiscal Year 2025: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2024: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2023: ¥1,820,000 (Direct Cost: ¥1,400,000、Indirect Cost: ¥420,000)
Fiscal Year 2022: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
|
Keywords | Subspace learning / Deep neural networks / Manifold optimization / Subspace methods / Pattern recognition |
Outline of Research at the Start |
We research a new algorithm for pattern recognition, which are computer programs that allow a machine to automatically recognize regularities in data, such as target objects and events. We mainly focus on the case of recognizing patterns in given multiple images of one object, addressing some inabilities of the current technology called deep learning.
|
Outline of Annual Research Achievements |
In fiscal year 2022, we worked to address the problem that traditional deep neural network frameworks process image sets independently, without considering the underlying feature distribution and the variance of the images in the set. To overcome this limitation, we devised a new subspace learning method called Grassmannian learning mutual subspace method (G-LMSM), which is an NN layer that can be integrated into deep neural networks. G-LMSM maps the image set into a low-dimensional input subspace representation, which is then matched with dictionary subspaces using a similarity metric of their canonical angles, an interpretable and computationally efficient metric. The key idea of G-LMSM is to learn dictionary subspaces as points on the Grassmann manifold, which is a smooth, non-linear manifold that captures the geometric structure of subspaces. This learning is optimized with Riemannian stochastic gradient descent, which is stable, efficient, and theoretically well-grounded. The proposed method was evaluated on three different tasks: hand shape recognition, face identification, and facial emotion recognition. Our experimental results showed that G-LMSM outperformed state-of-the-art methods on all three tasks, demonstrating its potential to improve the performance of deep frameworks for object recognition from image sets.
|
Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
Reason: We were able to combine subspace learning and deep neural networks to improve the performance in tasks of image set recognition.
|
Strategy for Future Research Activity |
We will work on new ways to combine subspace learning and deep neural network that can address their problems and improve performance.
|
Report
(1 results)
Research Products
(4 results)