Embedding and classifying test execution traces using neural networks

Classifying test executions automatically as pass or fail remains a key challenge in software testing and is referred to the oracle problem. It being attempted solve this problem with supervised learning over execution traces. A programme instrumented gather traces sequences of method invocations. small fraction programme's labelled verdicts. Execution are then embedded fixed length vectors neural network (NN) component that uses line-by-line information classify designed. The classification accuracy approach evaluated using subject programs from different application domains—1. Module Ethereum Blockchain, 2. PyTorch deep framework, 3. Microsoft SEAL encryption library components, 4. Sed stream editor, 5. Nine protocols Linux packet identifier, L7-Filter 6. Utilities library, commons-lang for Java. For all programs, it was found had high precision, recall specificity, averaging 93%, 94% 96%, respectively, while only training an average 14% total Experiments show proposed NN-based promising classifying domains.