Fast Adversarial Training with Adaptive Step Size.
Zhichao HuangYanbo FanChen LiuWeizhong ZhangYong ZhangMathieu SalzmannSabine SüsstrunkJue WangPublished in: CoRR (2022)
Keyphrases
- step size
- variable step size
- adaptive filter
- stochastic gradient descent
- convergence rate
- cost function
- convergence speed
- training set size
- faster convergence
- evolutionary programming
- least mean square
- hessian matrix
- training set
- line search
- neural network
- gradient method
- steepest descent method
- wavelet coefficients
- multiscale
- training algorithm
- recommender systems
- face recognition
- image processing