| Budget Amount *help |
¥4,550,000 (Direct Cost: ¥3,500,000、Indirect Cost: ¥1,050,000)
Fiscal Year 2016: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2015: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2014: ¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000)
|
|---|
| Outline of Final Research Achievements |
In this research, we propose a design method of lifting wavelet filter adaptive to each characteristic in order to analyze enormous signal data acquired from various kinds of sensors at high speed. In the proposed method, a new compact deep neural network (DNN) architecture based on lifting complex wavelets is proposed. The proposed DNN architecture (LcwtNet) is composed of multiple layers in addition to a CNN architecture. Complex wavelet and lifting wavelet layers are introduced as the lower layers of LcwtNet, which can reduce the number of parameters while maintaining high performance similar to that of CNN models. In simulations, the effectiveness of LcwtNet is demonstrated by several test results using the MNIST dataset. By virtue of the proposed method, we can realize efficient extraction of features on the sensor device side and development of a privacy-aware multi-sensing data analysis system.
|
