Scaling Equilibrium Propagation to Deep ConvNets by Drastically Reducing its Gradient Estimator Bias.
Axel LaborieuxMaxence ErnoultBenjamin ScellierYoshua BengioJulie GrollierDamien QuerliozPublished in: CoRR (2021)
Keyphrases
- feed forward
- variance reduction
- gradient estimation
- least squares
- maximum likelihood
- bias variance
- nash equilibrium
- market equilibrium
- variance estimator
- variational inequalities
- neural network
- maximum a posteriori
- game theory
- error estimation
- image gradient
- nash equilibria
- gradient information
- resource allocation
- maximum likelihood estimator
- back propagation
- edge detection
- trade off
- image processing