Flux.jacobian(::Chain, ::AbstractArray) gives the full jacobian of a neural network. Flux.hessian however throws the error “output is not scalar”. Let m = Chain(Dense(5,3,elu), Dense(3,3)) . What I can do is compute the gradient of each value in the jacobian matrix w.r. to each input using Flux.gra…

Is there an efficient way to compute the Hessian of a NN?

HenriDeh August 5, 2019, 2:26pm 8

Ha okay ! Great, thank you I’ll keep track of the progress then !

1 Like

Topic		Replies	Views
Efficiently computing Hessians of Neural Networks output with respect to inputs Performance question , flux , pde , hessian	1	261	July 16, 2023
Hessian matrix of a neural network vetor output Specific Domains physics , zygote , neural-network	3	439	July 16, 2022
Autograd in Flux General Usage	14	1286	March 5, 2021
How to compute hessian Machine Learning	1	866	August 26, 2019
Batched gradients and hessians with Flux Machine Learning question	10	200	January 25, 2025