I understand what you mean. This is indeed a way to use quantity to overcome high-dimensional output.
The input data is a variable similar to wind speed, and the output data is the wind pressure value of 400 measuring points on a building. This means that when determining the two input values, the wind pressure values of the 400 measuring points must be obtained at the same time.
According to your suggestion, it is equivalent to I established a gp model for each measuring point, a total of 400 gp models. But after I build the gp model, I will use it on MCMC to calculate the posterior distribution of the input variables in the gp model. This means that every step of MCMC requires the output results of the 400 gp models established before. But this is indeed a method, I think I might try it, thank you very much for your suggestions.
1 Like