Learning to stop while learning to predict
ICML2020 9-25-2020
PreviousLarge-Scale Long-Tailed Recognition in an Open WorldNextAdaptive Risk Minimization: A Meta-Learning Approach for Tackling Group Shift
Last updated
Was this helpful?
ICML2020 9-25-2020
Last updated
Was this helpful?
predictive model : transforms the input x to generate a path of states
stop policy : sequentially observes the states and determines the probability of stop at layer
variational stop time distribution : stop time distribution induced by stopping policy