Using test-time augmentation to investigate explainable AI: inconsistencies between method, model and human intuition.
Peter B. R. HartogFabian KrügerSamuel GenhedenIgor V. TetkoPublished in: J. Cheminformatics (2024)
Keyphrases
- test data
- probabilistic model
- objective function
- mathematical model
- cost function
- high order
- prediction model
- evaluation method
- high accuracy
- sensitivity analysis
- autoregressive
- detection method
- markov model
- gaussian distribution
- theoretical analysis
- bp neural network
- hybrid method
- statistical model
- monte carlo simulation
- input data
- prior knowledge
- optimization model
- modeling method
- artificial intelligence
- hybrid model
- parameter estimation
- linear model
- closed form
- em algorithm
- study proposes
- reconstruction method
- similarity measure
- regression analysis
- classification algorithm
- analytical model
- bayesian framework
- machine learning
- segmentation method
- case based reasoning
- significant improvement
- expert systems
- preprocessing
- image segmentation
- genetic algorithm