rockpool.training.adversarial_jax

Functions to implement adversarial training approaches using Jax

See also

👹 Adversarial training illustrates how to use the functions in this module to implement adversarial attacks on the parameters of a network during training.

Functions

adversarial_loss(parameters, net, inputs, ...)	Implement a hybrid task / adversarial robustness loss
pga_attack(params_flattened, net, rng_key, ...)	Performs the PGA (projected gradient ascent) based attack on the parameters of the network given inputs.