An empirical investigation of challenges of specifying training data and runtime monitors for critical software with machine learning and their relation to architectural decisions.
Hans-Martin HeynEric KnaussIswarya MalleswaranShruthi DinakaranPublished in: Requir. Eng. (2024)
Keyphrases
- machine learning
- training data
- learning algorithm
- software architecture
- decision trees
- supervised learning
- decision making
- training set
- support vector machine
- test data
- decision makers
- lessons learned
- active learning
- prior knowledge
- artificial intelligence
- test set
- software systems
- semi supervised learning
- commercial off the shelf
- natural language processing
- source code
- software development
- machine learning algorithms
- training examples
- inductive learning
- machine learning methods
- text classification
- information extraction
- maintenance activities
- domain knowledge
- user interface
- data sets
- classification models
- training process
- software maintenance
- training dataset
- learning tasks
- data mining
- learning systems
- labeled data
- computer systems
- text mining
- software engineering
- classification accuracy
- test cases
- development process
- computational intelligence
- generalization error
- knowledge discovery
- technical issues
- feature selection
- interdisciplinary field