Anti-Backdoor Learning: Training Clean Models on Poisoned Data.
Yige LiXixiang LyuNodens KorenLingjuan LyuBo LiXingjun MaPublished in: NeurIPS (2021)
Keyphrases
- prior knowledge
- historical data
- learning models
- data collection
- data sets
- experimental data
- database
- hidden variables
- data analysis
- accurate models
- learning process
- knowledge discovery
- bayesian methods
- data processing
- online learning
- data points
- raw data
- machine learning
- learned models
- training data
- data structure
- learning problems
- image data
- background knowledge
- supervised learning
- input data
- missing data
- learning algorithm
- learning systems
- predictive model
- feedforward neural networks
- probability distribution