Automating test results analysis using neural networks
summary
Neural network classifies embedded functional test results into pass, known bug, setup issue, test issue or new regression by combining 1024 tokens from tf‑idf vectorized test logs with execution history (last 10 runs) encoded into 1024 values for a 2048‑input neural network with three failure labels. Heuristic prelabeling trains the model (75/25 split), web UI shows buckets, clustering groups setup failures, reclassification entries retrain the model, and deployment already cuts false positive regression alerts for thousands of daily tests.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 0:39Embedded test results
- 1:42Embedded test setup
- 4:30Multiple failure reasons
- 6:04The problem
- 6:25The proposed model
- 12:28Outputs
- 13:18Results classification
- 18:10Preliminary Results
- 19:17Results Pages
- 21:01Regressions
- 22:45Conclusions
- 24:00Future work
- 25:18References
- 26:50Tokenization
related talks
40mMachine Learning helps to work with your test automation
50mPractical Applications of ML in Software Testing
46mAI-Driven Techniques for Noise Filtration in Software Validation Logs
40mThe Applicability of Machine Learning and Neural Networks in Test Automation
18mPredictive Network: How AI Forecasts System Failures
50mEffective test scrubbing
17mAnnotating Log Data For Machine Learning
31mWait, You Test Neural Networks Too?