دانلود رایگان مقاله انگلیسی مدلی برای پیش بینی قابلیت اطمینان نرم افزار با استفاده از شبکه حافظه طولانی کوتاه مدت - IEEE 2017

Abstract

With the development of software reliability research and machine learning, many machine learning models have been used in software reliability prediction. A long short term memory network (LSTM) modeling approach for software reliability prediction is proposed. Profit from its particular data flow control structure, the model overcomes the vanishing and exploding sensitivity of simple recursive neural network for software reliability prediction. Proposed approach also combines with layer normalization and truncate back propagation. To some extent, these two methods promote the effect of the proposed model. Compared with the simple recursive neural network, numerical results show that our proposed approach has a better performance and robustness with respect to software reliability prediction.

IV. CONCLUSIONS

This paper proposed a software reliability prediction model based on long short term memory network and truncated back propagation and layer normalization were added in our model to improve the performance. Compared with other neural network models, proposed approach have better predictive performance and robustness. Our further step will mainly concentrate in how to further enhance the software reliability prediction accuracy and compare with traditional software reliability prediction methods.