Designing losses for data-free training of normalizing flows on Boltzmann distributions.
Loris FelardosJérôme HéninGuillaume CharpiatPublished in: CoRR (2023)
Keyphrases
- data sets
- database
- raw data
- data analysis
- training examples
- databases
- data quality
- training data
- synthetic data
- data collection
- data processing
- standard deviation
- original data
- data distribution
- experimental data
- sensor data
- image data
- knowledge discovery
- probability distribution
- end users
- prior knowledge
- high dimensional
- data structure
- social networks
- machine learning
- neural network