UDAMA: Unsupervised Domain Adaptation through Multi-discriminator Adversarial Training with Noisy Labels Improves Cardio-fitness Prediction.
Yu WuDimitris SpathisHong JiaIgnacio Perez-PozueloTomas I. GonzalesSøren BrageNicholas J. WarehamCecilia MascoloPublished in: MLHC (2023)
Keyphrases
- training examples
- training set
- prediction accuracy
- label noise
- genetic algorithm
- test set
- training process
- pairwise
- prediction error
- training algorithm
- multi agent
- labelled data
- annotation effort
- body mass index
- noisy data
- prediction model
- prediction algorithm
- fitness landscape
- multi layer perceptron
- training data
- class labels
- supervised learning
- evolutionary algorithm
- multi label
- fitness function
- radial basis function network
- training samples
- genetic programming
- hidden markov models
- labeling effort
- active learning