Magpie: Automatically Tuning Static Parameters for Distributed File Systems using Deep Reinforcement Learning.
Houkun ZhuDominik ScheinertLauritz ThamsenKain Kordian GontarskaOdej KaoPublished in: IC2E (2022)
Keyphrases
- file system
- reinforcement learning
- distributed data
- scalable distributed
- data transfer
- parameter settings
- fine tuning
- parameter tuning
- multi agent
- maximum likelihood
- distributed environment
- distributed systems
- cooperative
- storage systems
- computer networks
- tuning parameters
- markov decision processes
- learning algorithm
- parameter values
- reinforcement learning algorithms
- access patterns
- machine learning
- peer to peer
- database systems
- application specific
- storage and retrieval
- model free
- communication cost
- expectation maximization
- state space
- tunable parameters