Test data reuse for evaluation of adaptive machine learning algorithms: over-fitting to a fixed 'test' dataset and a potential solution.
Alexej GossmannAria PezeshkBerkman SahinerPublished in: Medical Imaging: Image Perception, Observer Performance, and Technology Assessment (2018)
Keyphrases
- test data
- machine learning algorithms
- test cases
- benchmark data sets
- learning algorithm
- machine learning
- training data
- decision trees
- test set
- predictive accuracy
- machine learning methods
- training and test data
- learning problems
- training set
- random forests
- learning tasks
- machine learning approaches
- data sets
- machine learning models
- learning models
- training samples
- testing process
- standard machine learning algorithms
- base classifiers
- transfer learning
- supervised learning
- learning process
- feature extraction
- database systems