Neural Information Processing Systems · 1989 · 14 citations · 9 references
ComputingComputer ScienceMathematicsAlgorithmsDynamical SystemsMachine LearningComputational DynamicsLearning In Neural SystemsNeural NetworksNonlinear DynamicsOptimization For Machine LearningAdjoint FormulationFast LearningLearningNeural Dynamical SystemsNonlinear Dynamical System ResponseNonlinear Dynamical SystemsOperatorsSelf-adjoint Operators
A methodology for faster supervised learning in dynamical nonlinear neural networks is presented. It exploits the concept of adjoint operators to enable computation of changes in the network's response due to perturbations in all system parameters, using the solution of a single set of appropriately constructed linear equations. The lower bound on speedup per learning iteration over conventional methods for calculating the neuromorphic energy gradient is O(N2), where N is the number of neurons in the network.
9
Terminal attractors in neural networks
Michail Zak · Neural Networks · 1989 · 233 citations
Dynamics and architecture for neural computation
Fernando J. Pineda · Journal of Complexity · 1988 · 190 citations
Learning state space trajectories in recurrent neural networks
Barak A. Pearlmutter · 1989 · 189 citations · Full text