HIGHLIGHTS
SUMMARY
The problem of data saturation in neural_network based resource management methods has been addressed to some extent using coupled simulators, also referred to as co-simulators in the literature6. Other methods combine analytical methods with neural_networks to simulate a physical environment9,10. To find optimal φt at each timestep, the authors leverage a DNN based model to develop a low-fidelity surrogate model of the simulator that takes in the simulator inputs, its parameters and outputs another set of QoS estimates t=f (Wt, φt, Gt; θ ), Q where θ denotes the parameters of the . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.