DDI-PULearn: a positive-unlabeled learning method for large-scale prediction of drug-drug interactions.
Yi ZhengHui PengXiaocai ZhangZhixun ZhaoXiaoying GaoJinyan LiPublished in: BMC Bioinform. (2019)
Keyphrases
- prior knowledge
- high accuracy
- learning algorithm
- unsupervised learning
- pairwise
- significant improvement
- positive and negative
- supervised learning
- prediction error
- learning process
- computational cost
- segmentation method
- detection method
- positive and unlabeled examples
- neural network
- prediction model
- semi supervised learning
- support vector machine
- cost function
- objective function
- bayesian networks