Equilibrium Propagation: Bridging the Gap Between Energy-Based Models and Backpropagation

10 months ago by

View PDF

Benjamin Scellier, Yoshua Bengio

We introduce Equilibrium Propagation, a learning framework for energy-based models. It involves only one kind of neural computation, performed in both the first phase (when the prediction is made) and the second phase of training (after the target or prediction error is revealed). Although this algorithm computes the gradient of an objective function just like Backpropagation, it does not need a special computation or circuit for the second phase, where errors are implicitly propagated. Equilibrium Propagation shares similarities with Contrastive Hebbian Learning and Contrastive Divergence while solving the theoretical issues of both algorithms: our algorithm computes the gradient of a well defined objective function. Because the objective function is defined in terms of local perturbations, the second phase of Equilibrium Propagation corresponds to only nudging the prediction (fixed point, or stationary distribution) towards a configuration that reduces prediction error. In the case of a recurrent multi-layer supervised network, the output units are slightly nudged towards their target in the second phase, and the perturbation introduced at the output layer propagates backward in the hidden layers. We show that the signal 'back-propagated' during this second phase corresponds to the propagation of error derivatives and encodes the gradient of the objective function, when the synaptic update corresponds to a standard form of spike-timing dependent plasticity. This work makes it more plausible that a mechanism similar to Backpropagation could be implemented by brains, since leaky integrator neural computation performs both inference and error back-propagation in our model. The only local difference between the two phases is whether synaptic changes are allowed or not.

Community: paperstack

1 Answer

10 months ago by
Backpropagation was thought to be implausible to explain error propagation in the brain because the following two requirements have no clear supporting evidence:
  • A direct feedback connection would be needed for all forward-connected neurons.
  • A temporally separated forward and a backward phase would be needed.
This paper presents an alternative to backpropagation that is plausible without the second requirements: Within an energy-based model similar to a Hopfield network, with an extra term in the energy function that models closeness of the predictions in the last layer to the labels.

Experimental results yielded that the algorithm is slow (4 days vs. 5 minutes for backprop) and worse (3% error rate vs 1.6% for backprop), thoug this might improve with analog hardware.
Please login to add an answer/comment or follow this question.

Similar posts:
Search »