检索结果-内蒙古大学图书馆

App-Net: A Hybrid Neural Network for Encrypted Mobile Traffic Classification

学校读者我要写书评

暂无评论

App-Net: A Hybrid Neural Network for Encrypted Mobile Traffi...

IEEE Conference on Computer Communications Workshops, INFOCOM Wksps

作者： Xin Wang Shuhui Chen Jinshu Su College of Computer Science and Technology National University of Defense Technology National Key Laboratory for Parallel and Distributed Processing National University of Defense Technology

ISBN: (数字)9781728186955

ISBN: (纸本)9781728186962

With the exponential growth of mobile traffic data, mobile traffic classification is in a great need. It is an essential step to improve the performance of network services such as QoS and security monitoring. However, the widespread use of encrypted protocols, especially the TLS protocol, has posed great challenges to traditional traffic classification techniques. As the rule-based deep packet inspection approaches are ineffective for encrypted traffic classification, various machine learning methods have been studied and used. Recently, deep learning solutions which enable automatic feature extraction are also proposed to classify encrypted traffic. In this paper we propose App-Net, an end-to-end hybrid neural network, to learn effective features from raw TLS flows for mobile app identification. App-Net is designed by combining RNN and CNN in a parallel way. So that it can learn a joint flow-app embedding to characterize both flow sequence patterns and unique app signatures. We evaluate App-Net on a real-world dataset that covering 80 apps. The results show that our method can achieve an excellent performance and outperform the state-of-the-art methods.

关键词： Cryptography Feature extraction Payloads Recurrent neural networks Machine learning Convolutional neural networks

Research on Fine-Grained Sentiment Classification 8th

学校读者我要写书评

暂无评论

Research on Fine-Grained Sentiment Classification

8th CCF International Conference on Natural Language processing and Chinese Computing, NLPCC 2019

作者： Wang, Zhihui Wang, Xiaodong Chang, Tao Lv, Shaohe Guo, Xiaoting National Key Laboratory of Parallel and Distributed Processing College of Computer Science and Technology National University of Defence Technology Changsha410073 China

ISBN: (纸本)9783030322359

Aiming at the fine-grained sentiment classification that distinguishes the emotional intensity, the commonly used dataset SST-1 is analyzed in depth. Through the analysis, it is found that the dataset has serious problems such as data imbalance and small overall scale, which seriously restricts the classification effect. In order to solve the related problems, data augmentation method is adopted to realize the optimization of the dataset. The IMDB and other data which are relatively homologous to the original dataset are annotated, and the focus is to expand the categories with fewer numbers. By this way, the problem of data imbalance is effectively alleviated and the original data scale is expanded. Then, based on the Bidirectional Encoder Representations from Transformers (BERT) model, which has good overall performance on natural language processing, the benchmark classification model is built. Through multiple comparison experiments on the original dataset and the enhanced data, the influence of the deficiency of the original dataset on the classification effect is verified. And, it is fully demonstrated that the enhanced data can effectively improve the test results and solve the problem of large differences in performance between different categories well. © 2019, Springer Nature Switzerland AG.

关键词： Classification (of information)

NUMA-aware FFT-based convolution on ARMv8 many-core CPUs

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Huang, Xiandong Wang, Qinglin Lu, Shuyu Hao, Ruochen Mei, Songzhu Liu, Jie Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha410073 China School of Computer Science National University of Defense Technology Changsha410073 China Department of Biomedical Informatics University of Pittsburgh Pittsburgh United States

Convolutional Neural Networks (CNNs), one of the most representative algorithms of deep learning, are widely used in various artificial intelligence applications. Convolution operations often take most of the computational overhead of CNNs. The FFT-based algorithm can improve the efficiency of convolution by reducing its algorithm complexity, there are a lot of works about the high-performance implementation of FFT-based convolution on many-core CPUs. However, there is no optimization for the non-uniform memory access (NUMA) characteristics in manycore CPUs. In this paper, we present a NUMA-aware FFT-based convolution implementation on ARMv8 many-core CPUs with NUMA architectures. The implementation can reduce a number of remote memory access through the data reordering of FFT transformations and the three-level parallelization of the complex matrix multiplication. The experiment results on a ARMv8 many-core CPU with NUMA architectures demonstrate that our NUMA-aware implementation has much better performance than the state-of-the-art work in most cases. Copyright © 2021, The Authors. All rights reserved.

关键词： Convolutional neural networks

Outlier detection ensemble with embedded feature delection

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Cheng, Li Wang, Yijie Liu, Xinwang Li, Bin Science and Technology on Parallel and Distributed Processing Laboratory College of Computer National University of Defense Technology Changsha China

Feature selection places an important role in improving the performance of outlier detection, especially for noisy data. Existing methods usually perform feature selection and outlier scoring separately, which would select feature subsets that may not optimally serve for outlier detection, leading to unsatisfying performance. In this paper, we propose an outlier detection ensemble framework with embedded feature selection (ODEFS), to address this issue. Specifically, for each random sub-sampling based learning component, ODEFS unifies feature selection and outlier detection into a pairwise ranking formulation to learn feature subsets that are tailored for the outlier detection method. Moreover, we adopt the thresholded self-paced learning to simultaneously optimize feature selection and example selection, which is helpful to improve the reliability of the training set. After that, we design an alternate algorithm with proved convergence to solve the resultant optimization problem. In addition, we analyze the generalization error bound of the proposed framework, which provides theoretical guarantee on the method and insightful practical guidance. Comprehensive experimental results on 12 real-world datasets from diverse domains validate the superiority of the proposed ODEFS. Copyright © 2020, The Authors. All rights reserved.

关键词： Feature extraction

Research on Integrated Detection of SQL Injection Behavior Based on Text Features and Traffic Features

学校读者我要写书评

暂无评论

Research on Integrated Detection of SQL Injection Behavior B...

作者： Ming Li Bo Liu Guangsheng Xing Xiaodong Wang Zhihui Wang College of Intelligence Science and Technology National University of Defence Technology National Key Laboratory of Parallel and Distributed Processing College of Computer Science and Technology National University of Defence Technology

With the rapid development of Internet technology, various network attack methods come out one after the other. SQL injection has become one of the most severe threats to Web applications and seriously threatens various Web application services and users' data security. There are both traditional detection methods and emerging methods based on deep learning technology with higher detection accuracy for the detection of SQL injection. However, they are all for detecting a single statement and cannot determine the stage of the attack. To further improve the effect of SQL injection detection, this paper proposes an integrated detection framework for SQL injection behavior based on both text features and traffic features. We propose a SQL-LSTM model based on deep learning technology as the detection model at the text features level. Meanwhile,the features of the data traffic are merged. By this integrated method, the detection effect of SQL injection is further improved.

关键词： SQL injection Integrated method SQL-LSTM Traffic features

A Prompting-based Approach for Adversarial Example Generation and Robustness Enhancement

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Yang, Yuting Huang, Pei Cao, Juan Li, Jintao Lin, Yun Dong, Jin Song Ma, Feifei Zhang, Jian Key Lab of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China Beijing China National University of Singapore Singapore Laboratory of Parallel Software and Computational Science ISCAS Beijing China

Recent years have seen the wide application of NLP models in crucial areas such as finance, medical treatment, and news media, raising concerns of the model robustness and vulnerabilities. In this paper, we propose a novel prompt-based adversarial attack to compromise NLP models and robustness enhancement technique. We first construct malicious prompts for each instance and generate adversarial examples via mask-and-filling under the effect of a malicious purpose. Our attack technique targets the inherent vulnerabilities of NLP models, allowing us to generate samples even without interacting with the victim NLP model, as long as it is based on pre-trained language models (PLMs). Furthermore, we design a prompt-based adversarial training method to improve the robustness of PLMs. As our training method does not actually generate adversarial samples, it can be applied to large-scale training sets efficiently. The experimental results show that our attack method can achieve a high attack success rate with more diverse, fluent and natural adversarial examples. In addition, our robustness enhancement method can significantly improve the robustness of models to resist adversarial attacks. Our work indicates that prompting paradigm has great potential in probing some fundamental flaws of PLMs and fine-tuning them for downstream tasks. Copyright © 2022, The Authors. All rights reserved.

关键词： Natural language processing systems

DFTracker: detecting double-fetch bugs by multi-taint parallel tracking

学校读者我要写书评

暂无评论

Frontiers of Computer science 2019年第2期13卷 247-263页

作者： Pengfei WANG Kai LU Gen LI Xu ZHOU Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha 410073 China College of Computer National University of Defense Technology Changsha 410073 China Collaborative Innovation Center of High Performance Computing National University of Defense Technology Changsha 410073 China

A race condition is a common trigger for concurrency bugs. As a special case, a race condition can also occur across the kernel and user space causing a doublefetch bug, which is a field that has received little research attention. In our work, we first analyzed real-world doublefetch bug cases and extracted two specific patterns for doublefetch bugs. Based on these patter ns, we proposed an approach of multi-taint parallel tracking to detect double-fetch bugs. We also implemented a prototype called DFTracker (doublefetch bug tracker), and we evaluated it with our test suite. Our experiments demonstrated that it could effectively find all the double-fetch bugs in the test suite including eight realworld cases with no false negatives and minor false positives. In addition, we tested it on Linux kernel and found a new double-fetch bug. The execution overhead is approximately 2x for single-file cases and approximately 9x for the whole kernel test, which is acceptable. To the best of the authors1 knowledge, this work is the first to introduce multi-taint parallel tracking to double-fetch bug detection—an innovative method that is specific to double-fetch bug features—and has better path coverage as well as lower runtime overhead than the widely used dynamic approaches.

关键词： multi-taint parallel tracking double fetch race condition between kernel and user time of check to time of use real-world case analysis Clang Static Analyzer

Multi-representation knowledge distillation for audio classification

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Gao, Liang Xu, Kele Wang, Huaimin Peng, Yuxing National Key Laboratory of Parallel and Distributed Processing College of Computer National University of Defense Technology Changsha410073 China

As an important component of multimedia analysis tasks, audio classification aims to discriminate between different audio signal types and has received intensive attention due to its wide applications. Generally speaking, the raw signal can be transformed into various representations (such as Short Time Fourier Transform and Mel Frequency Cepstral Coefficients), and information implied in different representations can be complementary. Ensembling the models trained on different representations can greatly boost the classification performance, however, making inference using a large number of models is cumbersome and computationally expensive. In this paper, we propose a novel end-to-end collaborative learning framework for the audio classification task. The framework takes multiple representations as the input to train the models in parallel. The complementary information provided by different representations is shared by knowledge distillation. Consequently, the performance of each model can be significantly promoted without increasing the computational overhead in the inference stage. Extensive experimental results demonstrate that the proposed approach can improve the classification performance and achieve state-of-the-art results on both acoustic scene classification tasks and general audio tagging tasks. Copyright © 2020, The Authors. All rights reserved.

关键词： Convolutional neural networks

A distributed class-incremental learning method based on neural network parameter fusion 21

学校读者我要写书评

暂无评论

A distributed class-incremental learning method based on neu...

21st IEEE International Conference on High Performance Computing and Communications, 17th IEEE International Conference on Smart City and 5th IEEE International Conference on Data science and Systems, HPCC/SmartCity/DSS 2019

作者： Guan, Hongtao Wang, Yijie Ma, Xingkong Li, Yongmou National Laboratory for Parallel and Distributed Processing College of Computer Science and Technology National University of Defense Technology Changsha Hunan China

ISBN: (纸本)9781728120584

Class-incremental learning has received considerable attention due to the better adaptability to constantly changing characteristic of online learning. The neural network is suitable for class-incremental learning because of batch training style and fine-tuning. However, as the rapid growth of data centers and data category, data isolated island problem presents a new challenge to the current machine learning model: how to provide an accurate training method in a distributed and incremental manner. Existing solutions cannot adapt to incremental classes without aggregating data. In this paper, we propose a distributed class-incremental learning framework, called DCIL, which iteratively updates new class parameters and fixes existing class parameters to achieve high training accuracy. Especially, we propose MF, a parameter fusion method to integrate the parameters of multi-nodes into a new global parameter. In order to match the requirements of identical columns corresponding to the known classes, we propose a single node class-incremental learning, called SCIL. To evaluate the performance of DCIL, we conduct experiments on 5 datasets under various parameter settings on both 2 and 4 nodes distributed scenarios. Extensive experiments confirm that DCIL outperform the general baselines and achieve classification accuracy closes to the method of data aggregating. © 2019 IEEE.

关键词： Iterative methods