Wait, You Test Neural Networks Too?
summary
Covers testing machine learning models (neural networks) in production: how a model with <1% error in development can show 25% error on a client’s data. Describes concrete testing problems for ML systems and approaches to detect and prevent such mismatches between training and real-world data.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:13What is machine learning
- 4:00Key evaluation metrics
- 8:56ML vs software testing
- 14:30Voice and face verification
- 23:50Role of test data
- 24:11Speech synthesis metrics
- 30:18Summary
related talks
40mThe Applicability of Machine Learning and Neural Networks in Test Automation
31mTesting Machine Learning Models
53mUnleashing the Black Box: Testing Machine Learning Systems
45mHow to Test Artificial Intelligence and Machine Learning Solutions?
30mHow to Properly Test ML Models & Data
38mWhat Testing Strategy for Your Machine Learning Applications?
06mMachine learning is about your data and deployment
32mTesting Machine Learning Systems: What it is and why you should care