Improving the Stochastic Gradient Descent's Test Accuracy by Manipulating the ℓ∞ Norm of its Gradient Approximation.

Published in: ICASSP (2023)

Keyphrases

stochastic gradient descent
loss function
least squares
prediction accuracy
machine learning
classification accuracy
matrix factorization
step size
objective function
lower bound
active learning
markov chain
random forests