研究課題/領域番号 |
20H04249
|
研究種目 |
基盤研究(B)
|
配分区分 | 補助金 |
応募区分 | 一般 |
審査区分 |
小区分61030:知能情報学関連
|
研究機関 | 国立研究開発法人理化学研究所 |
研究代表者 |
ZHAO QIBIN 国立研究開発法人理化学研究所, 革新知能統合研究センター, チームリーダー (30599618)
|
研究分担者 |
曹 建庭 埼玉工業大学, 工学部, 教授 (20306989)
横田 達也 名古屋工業大学, 工学(系)研究科(研究院), 准教授 (80733964)
|
研究期間 (年度) |
2020-04-01 – 2024-03-31
|
研究課題ステータス |
交付 (2023年度)
|
配分額 *注記 |
17,550千円 (直接経費: 13,500千円、間接経費: 4,050千円)
2023年度: 4,290千円 (直接経費: 3,300千円、間接経費: 990千円)
2022年度: 4,290千円 (直接経費: 3,300千円、間接経費: 990千円)
2021年度: 4,290千円 (直接経費: 3,300千円、間接経費: 990千円)
2020年度: 4,680千円 (直接経費: 3,600千円、間接経費: 1,080千円)
|
キーワード | Tensor networks / machine learning / Tensor Networks / Machine Learning / Robustness / tensor network |
研究開始時の研究の概要 |
Tensor networks (TNs) have recently gained increasing attentions in machine learning, data mining and computer vision fields due to its effectiveness in efficient computation and model compression in deep learning. However, there are many open problems that are still unexplored, which limits its impact in machine learning. Therefore, our research aims to investigate the fundamental theory and develop scalable and efficient learning algorithms for TN. Moreover, we will further explore what challenging problems in machine learning can be solved by TN technology.
|
研究実績の概要 |
We have developed several new tensor network decomposition and completion algorithms, and also developed the tensor network based neural network models and learning algorithms. These methods have been applied to several computer vision tasks.
Specifically, we have developed tensorized RNN model that can achieve long term memory and reduced model size; we also studied Bayesian latent factor models to understand how tensor network is able to achieve model compression; our proposed tensor fusion layer can be applied to image denoting tasks with improvement performance, which can be also applied to the development of multimodal sentimental analysis. We also developed an efficient algorithm for classification on incomplete data samples, which has practical applications when the high-quality dataset is difficult to be obtained.
From theoretical perspective, we have studied tensor nuclear norm and proposed several new definition of tensor norm, which has guarantee for exact recovery to tensor. In addition, we proposed a new type tensor network, called fully connected tensor network, which shows great flexibility on modeling complex interaction between tensor modes. The effectiveness of our theory and model is validated extensively on tensor completion tasks.
|
現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
Some experimental evaluation are postponed.
|
今後の研究の推進方策 |
We will further study tensor network representation ability when applied to neural networks.
|