Hong-Wei Zeng
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Hong-Wei Zeng has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Huaikou Miao
Shanghai University
5 shared publications - Shengbo Chen
School of Computer Engineering and Science, Shanghai University, Shanghai 200444, China
2 shared publications - Pan Liu
School of Business Information, Shanghai Business School, Shanghai, China
1 shared publication - Pan Liu1 shared publication
- Xiaoli Du
School of Computer Engineering and Science, Shanghai University, Shanghai 200444, China
1 shared publication - Yan Ma
Nanjing University of Aeronautics and Astronautics
1 shared publication
Latest Publications
- Open Access
RNNCon: Contribution Coverage Testing for Stacked Recurrent Neural Networks
Recurrent Neural Networks (RNNs) are applied in safety-critical fields such as autonomous driving, aircraft collision detection, and smart credit. They are highly susceptible to input perturbations, but little research on RNN-oriented testing techniques has been conducted, leaving a threat to a large number of sequential application domains. To address these gaps, improve the test adequacy of RNNs, find more defects, and improve the performance of RNNs models and their robustness to input perturbations. We aim to propose a test coverage metric for the underlying structure of RNNs, which is used to guide the generation of test inputs to test RNNs. Although coverage metrics have been proposed for RNNs, such as the hidden state coverage in RNN-Test, they ignore the fact that the underlying structure of RNNs is still a fully connected neural network but with an additional “delayer” that records the network state at the time of data input. We use the contributions, i.e., the combination of the outputs of neurons and the weights they emit, as the minimum computational unit of RNNs to explore the finer-grained logical structure inside the recurrent cells. Compared to existing coverage metrics, our research covers the decision mechanism of RNNs in more detail and is more likely to generate more adversarial samples and discover more flaws in the model. In this paper, we redefine the contribution coverage metric applicable to Stacked LSTMs and Stacked GRUs by considering the joint effect of neurons and weights in the underlying structure of the neural network. We propose a new coverage metric, RNNCon, which can be used to guide the generation of adversarial test inputs. And we design and implement a test framework prototype RNNCon-Test. 2 datasets, 4 LSTM models, and 4 GRU models are used to verify the effectiveness of RNNCon-Test. Compared to the current state-of-the-art study RNN-Test, RNNCon can cover a deeper decision logic of RNNs. RNNCon-Test is not only effective in identifying defects in Deep Learning (DL) systems but also in improving the performance of the model if the adversarial inputs generated by RNNCon-Test are filtered and added to the training set to retrain the model. In the case where the accuracy of the model is already high, RNNCon-Test is still able to improve the accuracy of the model by up to 0.45%.
Open AccessRNNCon: Contribution Coverage Testing for Stacked Recurrent Neural Networks
Recurrent Neural Networks (RNNs) are applied in safety-critical fields such as autonomous driving, aircraft collision detection, and smart credit. They are highly susceptible to input perturbations, but little research on RNN-oriented testing techniques has been conducted, leaving a threat to a large number of sequential application domains. To address these gaps, improve the test adequacy of RNNs, find more defects, and improve the performance of RNNs models and their robustness to input perturbations. We aim to propose a test coverage metric for the underlying structure of RNNs, which is used to guide the generation of test inputs to test RNNs. Although coverage metrics have been proposed for RNNs, such as the hidden state coverage in RNN-Test, they ignore the fact that the underlying structure of RNNs is still a fully connected neural network but with an additional “delayer” that records the network state at the time of data input. We use the contributions, i.e., the combination of the outputs of neurons and the weights they emit, as the minimum computational unit of RNNs to explore the finer-grained logical structure inside the recurrent cells. Compared to existing coverage metrics, our research covers the decision mechanism of RNNs in more detail and is more likely to generate more adversarial samples and discover more flaws in the model. In this paper, we redefine the contribution coverage metric applicable to Stacked LSTMs and Stacked GRUs by considering the joint effect of neurons and weights in the underlying structure of the neural network. We propose a new coverage metric, RNNCon, which can be used to guide the generation of adversarial test inputs. And we design and implement a test framework prototype RNNCon-Test. 2 datasets, 4 LSTM models, and 4 GRU models are used to verify the effectiveness of RNNCon-Test. Compared to the current state-of-the-art study RNN-Test, RNNCon can cover a deeper decision logic of RNNs. RNNCon-Test is not only effective in identifying defects in Deep Learning (DL) systems but also in improving the performance of the model if the adversarial inputs generated by RNNCon-Test are filtered and added to the training set to retrain the model. In the case where the accuracy of the model is already high, RNNCon-Test is still able to improve the accuracy of the model by up to 0.45%.
- 1 January 2013
- Published by Springer Nature in Journal of Computer Science and Technology
- Vol. 28 (1) , 203-216
Nondeterministic Probabilistic Petri Net — A New Method to Study Qualitative and Quantitative Behaviors of System
- 1 January 2013
- Published by Springer Nature in Journal of Computer Science and Technology
- Vol. 28 (1) , 203-216
Nondeterministic Probabilistic Petri Net — A New Method to Study Qualitative and Quantitative Behaviors of System
- 20 June 2011
- Published by China Science Publishing & Media Ltd. in Chinese Journal of Computers
- Vol. 34 (6) , 965-984
FSM-Based Testing: Theory, Method and Evaluation
To build the bridge between theoretical studies and industrial application towards FSM-based testing,the authors study some FSM-based testing theories and methods,and present the theories for building distinguishable sequence and reducing the redundancies among test set derived from FSM-based testing methods.Next some practical algorithms for realizing some FSM-based testing methods are designed in the paper.Finally,the authors evaluate some FSM-based testing methods by using five suggested evaluation indicators on some experiments.According to experimental results,the authors give some experiential suggestions for the implementation of FSM-based testing methods.It is intended to help with understanding benefits and limitations of FSM-based testing,selecting the approach used in a particular FSM-based testing method and popularizing FSM-based testing in the industry.
- 20 June 2011
- Published by China Science Publishing & Media Ltd. in Chinese Journal of Computers
- Vol. 34 (6) , 965-984
FSM-Based Testing: Theory, Method and Evaluation
To build the bridge between theoretical studies and industrial application towards FSM-based testing,the authors study some FSM-based testing theories and methods,and present the theories for building distinguishable sequence and reducing the redundancies among test set derived from FSM-based testing methods.Next some practical algorithms for realizing some FSM-based testing methods are designed in the paper.Finally,the authors evaluate some FSM-based testing methods by using five suggested evaluation indicators on some experiments.According to experimental results,the authors give some experiential suggestions for the implementation of FSM-based testing methods.It is intended to help with understanding benefits and limitations of FSM-based testing,selecting the approach used in a particular FSM-based testing method and popularizing FSM-based testing in the industry.
- 20 June 2011
- Published by China Science Publishing & Media Ltd. in Chinese Journal of Computers
- Vol. 34 (6) , 1012-1028
Model-Based Testing for Web Applications
- 20 June 2011
- Published by China Science Publishing & Media Ltd. in Chinese Journal of Computers
- Vol. 34 (6) , 1012-1028
Model-Based Testing for Web Applications
- 1 February 2010
- Published by Springer Nature in Journal of Shanghai University (English Edition)
- Vol. 14 (1) , 1-5
Deadlock detection using abstraction refinement
- 1 February 2010
- Published by Springer Nature in Journal of Shanghai University (English Edition)
- Vol. 14 (1) , 1-5
Deadlock detection using abstraction refinement
- 1 January 2010
- Published by Springer Nature
- p. 23-34
Deadlock Detection for Parallel Composition of Components
- 1 January 2010
- Published by Springer Nature
- p. 23-34
Deadlock Detection for Parallel Composition of Components
- 1 September 2007
- Published by EDP Sciences in Wuhan University Journal of Natural Sciences
- Vol. 12 (5) , 922-926
Model checking-based testing of Web applications
A formal model representing the navigation behavior of a Web application as the Kripke structure is proposed and an approach that applies model checking to test case generation is presented. The Object Relation Diagram as the object model is employed to describe the object structure of a Web application design and can be translated into the behavior model. A key problem of model checking-based test generation for a Web application is how to construct a set of trap properties that intend to cause the violations of model checking against the behavior model and output of counterexamples used to construct the test sequences. We give an algorithm that derives trap properties from the object model with respect to node and edge coverage criteria.
- 1 September 2007
- Published by EDP Sciences in Wuhan University Journal of Natural Sciences
- Vol. 12 (5) , 922-926
Model checking-based testing of Web applications
A formal model representing the navigation behavior of a Web application as the Kripke structure is proposed and an approach that applies model checking to test case generation is presented. The Object Relation Diagram as the object model is employed to describe the object structure of a Web application design and can be translated into the behavior model. A key problem of model checking-based test generation for a Web application is how to construct a set of trap properties that intend to cause the violations of model checking against the behavior model and output of counterexamples used to construct the test sequences. We give an algorithm that derives trap properties from the object model with respect to node and edge coverage criteria.