Browsing by Author "Chang, Jia-Ren"

Now showing 1 - 3 of 3

Deep Neural Network with Attention Mechanism for Classification of Motor Imagery EEG
(2019-01-01)
Huang, Yen-Cheng
;
Chang, Jia-Ren
;
Chen, Li-Fen
;
Chen, Yong-Sheng
;
交大名義發表
This paper presents a deep neural network architecture for the classification of motor imagery electroencephalographic recordings. This classification task usually encounters difficulties such as data with poor signal-to-noise ratio, contamination from muscle activity, body movements, and external interferences, and both intra-subject and intersubject variability. Through the spatiotemporal features automatically learned from training data, deep neural networks continue to demonstrate their good performance, versatility, and adaptation capability. In this work, we developed a novel neural network model which can extract signal features from multiple electrodes in a manner similar to that of conventional signal processing methods, such as common spatial patterns and common temporal patterns. The proposed neural network model comprises an attention mechanism, which calculates the importance of each electrode, and a spatial convolution layer. Compared to the results obtained using a variety of state-of-the-art deep learning techniques, the proposed scheme represents a considerable advancement in classification accuracy when applied to the BCI competition IV dataset 2a. By training with data of all subjects, the proposed universal neural network model outperforms state-of-the-art methods in terms of classification accuracy.
NEUROSCIENCE-INSPIRED RECURRENT NETWORK FOR OBJECT RECOGNITION
(2017-01-01)
Chang, Jia-Ren
;
Kuo, Po-Chih
;
Chen, Yong-Sheng
;
資訊工程學系
;
Department of Computer Science
Deep neural networks inspired by the connections among biological neurons have been highly effective in the advancement of computer vision. Recent research into recurrent neural networks has taken account of forward as well as recurrent connections in a neural network architecture. In this study, we proposed a model that mirrors the architecture of the human ventral pathway with separate layers representing brain regions connected using long-range recurrent links. MAR 10/100 datasets were used to assess the performance of object recognition using the proposed model and the results were compared with those obtained using state-of-the-art methods. We demonstrated that the classification accuracy increased as the number of recurrences increased. Our results suggest that the proposed neuroscience-inspired model can facilitate object recognition in computer vision and may help to elucidate neurological mechanisms in the human brain.
Pyramid Stereo Matching Network
(2018-01-01)
Chang, Jia-Ren
;
Chen, Yong-Sheng
;
資訊工程學系
;
Department of Computer Science
Recent work has shown that depth estimation from a stereo pair of images can be formulated as a supervised learning task to be resolved with convolutional neural networks (CNNs). However, current architectures rely on patch-based Siamese networks, lacking the means to exploit context information for finding correspondence in ill-posed regions. To tackle this problem, we propose PSMNet, a pyramid stereo matching network consisting of two main modules: spatial pyramid pooling and 3D CNN. The spatial pyramid pooling module takes advantage of the capacity of global context information by aggregating context in different scales and locations to form a cost volume. The 3D CNN learns to regularize cost volume using stacked multiple hourglass networks in conjunction with intermediate supervision. The proposed approach was evaluated on several benchmark datasets. Our method ranked first in the KITTI 2012 and 2015 leaderboards before March 18, 2018. The codes of PSMNet are available at: https://github.com/JiaRenChang/PSMNet.