Weighted Densely Connected Convolutional Networks for Reinforcement Learning

Xia, Min; Song, Wenzhu; Sun, Xudong; Liu, Jia; Tao, Ye; Xu, Yiqing

doi:10.1142/s0218001420520011

Cited by 12 publications

(5 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Most of the existing network frameworks of action recognition algorithms based on deep learning [11] are developed from the convolutional neural network [12][13][14][15]. Because action recognition objects are video sequences, they increase time-series information compared with a single image.…”

Section: Introductionmentioning

confidence: 99%

An LSTM with Differential Structure and Its Application in Action Recognition

Chen

Zheng

Gao

et al. 2022

Mathematical Problems in Engineering

View full text Add to dashboard Cite

Because of the broad application of human action recognition technology, action recognition has always been a hot spot in computer vision research. The Long Short-Term Memory (LSTM) network is a classic action recognition algorithm, and many effective hybrid algorithms have been proposed based on basic LSTM infrastructure. Although some progress has been made in accuracy, most of those hybrid algorithms have to have more and more complex structures and deeper network levels. After analyzing the structure of the classic LSTM from the perspective of control theory, we determined that the classic LSTM could strengthen the differential characteristics of human action recognition technology to reflect the change of speed. Thus, an improved LSTM structure with an input differential characteristic module is proposed. Furthermore, in this article, we considered the influence of first-order and second-order differential on the extraction of movement pose information, that is, the influence of movement speed and acceleration on action recognition. We designed four different LSTM units with first-order and second-order differential. Moreover, the experiments were performed for the four units on three common datasets repeatedly. We found that the LSTM network with the input differential feature module proposed in this article can effectively improve action recognition accuracy and stability without deepening the complexity of the network and can be used as a new basic LSTM network architecture.

show abstract

Section: Introductionmentioning

confidence: 99%

An LSTM with Differential Structure and Its Application in Action Recognition

Chen

Zheng

Gao

et al. 2022

Mathematical Problems in Engineering

View full text Add to dashboard Cite

show abstract

“…However, DenseNet ignores the problem that the output characteristic information of each layer is often different for the diagnosis task, and the DenseNet treats the characteristic information of each layer equally. So, many improvements are coming [30][31][32]. This paper, inspired by the squeeze-and-excitation networks (SENet) in the literature [33], proposes a weighted densely connected convolutional network (WDenseNet).…”

Section: Introductionmentioning

confidence: 99%

A fault diagnosis method for an electro-hydraulic directional valve based on intrinsic mode functions and weighted densely connected convolutional networks

Shi¹,

Ren²,

Tang³

et al. 2021

Meas. Sci. Technol.

View full text Add to dashboard Cite

Wear caused by contaminated oil or frequent reciprocating of the spool is the chief source of internal leakage in electro-hydraulic directional valves. It is difficult to detect the location and extent of the internal wear of directional valves because the hydraulic transmission works in a closed system. Therefore, this paper focuses on the internal leakage fault diagnosis caused by the wear based on intrinsic mode functions (IMFs) and weighted densely connected convolutional networks (WDenseNets), especially for electro-hydraulic directional valves. First, the empirical mode decomposition is performed in three subsections, respectively, and a series of IMF components are obtained correspondingly. Combined with the ReliefF feature selection algorithm, the optimal IMF component is selected. The weights assigned to the optimal IMF mode are normalized and then weighted as the subsequent diagnostic feature vector. Then, a WDenseNet inspired by squeeze-and-excitation networks is designed. Finally, the weighted optimal IMF component is used as the input for the WDenseNet for fault identification and classification. In addition, an experimental method to investigate the internal leakage faults associated with wear is developed. Theoretical and experimental results indicate that the fault diagnosis method can be used to identify the type of internal leakage and wear location in electro-hydraulic directional valves. Compared with other intelligent diagnosis methods, this method has higher diagnostic accuracy and better training stability. The effectiveness of the proposed method is also verified through an experimental study of the bearing fault in an axial piston pump.

show abstract

“…Although existing semantic segmentation algorithms perform well in remote sensing image extraction, there is a problem of gradient vanishing due to the deepening of convolution layers during training process, and thus the network performance degrades and its image segmentation accuracy is affected. In addition, a DNN normally has many convolution kernels, which increases the parameter numbers of the training network, and thus makes the training time-consuming and difficult [24]. To solve these problems, a separable residual SegNet (SR-SegNet) is proposed in this paper.…”

Section: Introductionmentioning

confidence: 99%

Water Areas Segmentation from Remote Sensing Images Using a Separable Residual SegNet Network

Weng

Xia

et al. 2020

IJGI

Self Cite

View full text Add to dashboard Cite

Changes on lakes and rivers are of great significance for the study of global climate change. Accurate segmentation of lakes and rivers is critical to the study of their changes. However, traditional water area segmentation methods almost all share the following deficiencies: high computational requirements, poor generalization performance, and low extraction accuracy. In recent years, semantic segmentation algorithms based on deep learning have been emerging. Addressing problems associated to a very large number of parameters, low accuracy, and network degradation during training process, this paper proposes a separable residual SegNet (SR-SegNet) to perform the water area segmentation using remote sensing images. On the one hand, without compromising the ability of feature extraction, the problem of network degradation is alleviated by adding modified residual blocks into the encoder, the number of parameters is limited by introducing depthwise separable convolutions, and the ability of feature extraction is improved by using dilated convolutions to expand the receptive field. On the other hand, SR-SegNet removes the convolution layers with relatively more convolution kernels in the encoding stage, and uses the cascading method to fuse the low-level and high-level features of the image. As a result, the whole network can obtain more spatial information. Experimental results show that the proposed method exhibits significant improvements over several traditional methods, including FCN, DeconvNet, and SegNet.

show abstract

Weighted Densely Connected Convolutional Networks for Reinforcement Learning

Cited by 12 publications

References 17 publications

An LSTM with Differential Structure and Its Application in Action Recognition

An LSTM with Differential Structure and Its Application in Action Recognition

A fault diagnosis method for an electro-hydraulic directional valve based on intrinsic mode functions and weighted densely connected convolutional networks

Water Areas Segmentation from Remote Sensing Images Using a Separable Residual SegNet Network

Contact Info

Product

Resources

About