InTune: Reinforcement Learning-based Data Pipeline Optimization for Deep Recommendation Models.
Kabir NagrechaLingyi LiuPablo DelgadoPrasanna PadmanabhanPublished in: RecSys (2023)
Keyphrases
- reinforcement learning
- data sources
- data collection
- data sets
- data structure
- synthetic data
- probabilistic model
- database
- accurate models
- historical data
- experimental data
- models built
- statistical analysis
- computer systems
- optimization problems
- prior knowledge
- data analysis
- high quality
- data processing
- image data
- model selection
- probability distribution
- high dimensional data
- statistical models
- markov decision processes
- raw data
- training data
- data mining