wedeling / easysurrogate Goto Github PK

Hi, I have read with a lot if interest you report on DAS. I can see that it ends with some important questions. Did you investigate those further since the report? For instance, how does the inductive bias in DAS compared to just a standard feedforward NN, where presumably the first layers would, like in DAS, learn a convenient representation of the data space (albeit, unlike DAS, this is unconstrained for a standard NN)?

Thanks
Lorenzo

We discussed how to make surrogates for Flee, and converged on two possibilities:

1. We apply mulitlevel monte carlo, to do runs with reduced # of agents, and reproduce sensitivity analysis at an original resolution,
2. Apply HDMR (Hi-dimensional model representations) to perform SC/Sobol sensitivity analysis with a vastly reduced number of runs.

Hi, there is an error keyError:('run_id', 0) in line 136 in campaign.py when I run the train_DAS_surrogate.py of tests/deep_active_subspaces