Training-Free Neural Active Learning with Initialization-Robustness Guarantees.
Apivich HemachandraZhongxiang DaiJasraj SinghSee-Kiong NgBryan Kian Hsiang LowPublished in: ICML (2023)
Keyphrases
- active learning
- training examples
- training set
- annotation effort
- supervised learning
- neural network
- batch mode
- online learning
- network architecture
- training algorithm
- test set
- learning strategies
- recurrent networks
- machine learning
- training phase
- sample selection
- random sampling
- training process
- semi supervised learning
- genetic algorithm
- labeled data
- bio inspired
- experimental design
- learning process
- feed forward neural networks
- image retrieval
- selective sampling
- random selection
- training data
- learning algorithm
- crowd sourced
- active learning strategies
- imbalanced data classification