Adversarial Attacks on Uncertainty Enable Active Learning for Neural Network Potentials.
Daniel Schwalbe-KodaAik Rui TanRafael Gómez-BombarelliPublished in: CoRR (2021)
Keyphrases
- active learning
- neural network
- pattern recognition
- back propagation
- label noise
- higher order
- feed forward
- countermeasures
- learning strategies
- random sampling
- uncertain data
- genetic algorithm
- selective sampling
- artificial neural networks
- decision theory
- high order
- learning algorithm
- relevance feedback
- semi supervised
- recurrent neural networks
- multi agent
- self organizing maps
- music emotion classification
- belief functions
- experimental design
- imbalanced data classification
- security mechanisms
- pool based active learning
- active learning strategies
- prediction model
- multilayer perceptron
- neural network model
- radial basis function
- fault diagnosis
- intrusion detection
- anomaly detection
- labeled data
- supervised learning
- fuzzy logic
- learning process
- machine learning