Leveraging Small Software Engineering Data Sets with Pre-Trained Neural Networks
Authors:
- 1 May 2019
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 29-32
Abstract
Many software engineering data sets, particularly those that demand manual labelling for classification, are necessarily small. As a consequence, several recent software engineering papers have cast doubt on the effectiveness of deep neural networks for classification tasks, when applied to these data sets. We provide initial evidence that recent advances in Natural Language Processing, that allow neural networks to leverage large amount of unlabelled data in a pre-training phase, can significantly improve performance.Keywords
This publication has 15 references indexed in Scilit:
- Keep it simple: Is deep learning good for linguistic smell detection?2018
- Are deep neural networks the best choice for modeling source code?2017
- Analyzing APIs Documentation and Code to Detect Directive Defects2017
- Release planning of mobile apps based on user reviews2016
- On the automatic classification of app reviewsRequirements Engineering, 2016
- Linking e-mails and source code artifacts2010
- A Survey on Transfer LearningIEEE Transactions on Knowledge and Data Engineering, 2009
- ImageNet: A large-scale hierarchical image database2009
- A Fast Learning Algorithm for Deep Belief NetsNeural Computation, 2006
- Long Short-Term MemoryNeural Computation, 1997