Xinrong Wu

Xinrong Wu

Auto-generated profile, based on publicly available information

Xinrong Wu

Army Engineering University of PLA, Nanjing, China

Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
15 October 2021
Verified
Conference Paper
Closed access
YZYanmin Zhu
Yanmin Zhu
XZxiang zheng
xiang zheng
XWXinrong Wu
Xinrong Wu
WLWanning Liu
Wanning Liu
LPLei Pi
Lei Pi
MCMeiju Chen
Meiju Chen
Dual-Path RNN(DPRNN) has achieved great progress in single-channel speech separation. However, RNN-based model needs to pass information through intermediate states and it does not allow parallel computing. Meanwhile, inter-chunk modeling of DPRNN only modeled between multiples of the chunk length, which means the underutilization of contextual information. To address these problems, we propose a multi-scale dual-path temporal convolutional network, DPTCN-ATPP. DPTCN-ATPP utilizes a stacked of one-dimensional dilated convolutions (1-D CNNs) instead of RNN to learn local and global information. Besides, DPTCN-ATPP applies parallel encoders to extract multi-scale input features. Finally, DPTCN-ATPP introduces the ATPP block to summarize multi-scale deep features. The experiment results show that DPTCN-ATPP achieves 19.6dB and 19.8dB on the metrics of scale-invariant source-to-noise ratio(SI-SNR) and source-to-distortion (SDR) respectively, outperforming baseline DPRNN and single-scale TCN (Conv-TasNet).

10 December 2021
Verified
Conference Paper
Closed access
LFLinge Fan
Linge Fan
XWXinrong Wu
Xinrong Wu
WTWei Tong
Wei Tong
WZWeijun Zeng
Weijun Zeng
Unsupervised feature selection (UFS) has been widely used in many machine learning applications, since it can alleviate burdens on the curse of high dimensional data. However, existing works design the UFS methods with the assumption that the complete information of dataset can be observed. In this paper, we propose a UFS method from incomplete data by further introducing ℓ 2,1 -norm minimization. In order to exploit feature relevance for UFS, we show that ℓ 2,1 -norm minimization should be elaborately dealt with unobserved information, which can be finely described with the use of an indicator matrix. The proposed ℓ 2,1 -norm minimization UFS from incomplete data (LUFS) can be directly solved by a convex optimization problem, and we design an alternative algorithm to optimize the convex problem, where the convergence of LUFS algorithm is also proved theoretically and experimentally. Empirical results demonstrate the effectiveness of LUFS with the state-of-the-art methods on real incomplete datasets.

3 November 2022
Verified
Conference Paper
Open access
DHDejun He
Dejun He
XWXinrong Wu
Xinrong Wu
YLYu Lu
Yu Lu
TWTianchi Wang
Tianchi Wang
The recognition of communication relationships under Non-cooperative conditions is significant for understanding the network composition of unknown targets, inferring network topology, and identifying key nodes, which is a prerequisite and basis for conducting efficient electronic countermeasures. However, under Non-cooperative conditions, for prior knowledge related to the target network is difficult to obtain, the communication relationships recognition faces enormous challenges. To address this issue, we construct a system model, analyze the mechanism of wireless communication interaction, extract feature series of signals from spectrum monitoring data, and propose a Transformer-based algorithm for recognizing target network communication relationships. This paper conducts simulation experiments in different scenarios to compare the Transformer-based communication relation recognition algorithm with the other four methods, such as SVM, CNN-based recognition algorithm, ResNet-based recognition algorithm, and LSTM-based recognition algorithm, respectively. And results demonstrate that the proposed algorithm shows high recognition accuracy, good anti-interference performance, and robustness.


10 December 2021
Verified
Conference Paper
Closed access
MYMengmeng Yin
Mengmeng Yin
LWLei Wang
Lei Wang
CYChangHua Yao
ChangHua Yao
XWXinrong Wu
Xinrong Wu
YSYulong Song
Yulong Song
YZYu Zhang
Yu Zhang
In order to resist the cascading failure phenomenon that often occurs in real networks and to enhance the reliability of the network, combined with the improved scale-free network, a cascading failure model with adjustable parameters is constructed. In this model, combining node degree information and node betweenness information and introducing the proportional parameters of the two, are α and β respectively, redefine the initial load and set the different proportional relationship between node load and node capacity. In view of the load spread of failed nodes, the maximum load capacity allocation mechanism is quickly developed and then through the analysis of the network cascade failure, the evolution process of the load parameters is deduced, and the influence of the parameters in the model on the robustness of the network is obtained. Finally, experiments verify the effectiveness of the proposed method.

28 October 2020
Verified
Conference Paper
Closed access
MDMengxue Deng
Mengxue Deng
XWXinrong Wu
Xinrong Wu
PCPu Chen
Pu Chen
WZWeijun Zeng
Weijun Zeng
One of the more representative tasks to detect large-scale network behavior data is to use support vector machine (SVM) to extract the characteristics of network data to establish Intrusion detection model. However, the traditional "ℓ 2 -norm" SVM can fit the model to some extent, it is prone to over fitting. In this paper, the sparse support vector machine (S-SVM) algorithm obtained by "ℓ 1 -norm" is used to extract and classify the network behavior data set based on the sparseness of the network behavior data set. Column and constraint generation algorithm is introduced to solve the S-SVM model, and the one-to-one method is used to expand it into a multi-classification sparse support vector machine. Finally, the KDD (knowledge discovery and data mining) CUP 1999 data set was used for experiments. By comparing with the SGD (stochastic gradient descent) and Barrier algorithm, the efficiency and accuracy of this multi-classification S-SVM algorithm in NBAD (network behavior anomaly detection) were verified.