On the asymptotic equivalence between differential Hebbian and temporal difference learning using a local third factor
نویسندگان
چکیده
In this theoretical contribution we provide mathematical proof that two of the most important classes of network learning correlation-based differential Hebbian learning and reward-based temporal difference learning are asymptotically equivalent when timing the learning with a local modulatory signal. This opens the opportunity to consistently reformulate most of the abstract reinforcement learning framework from a correlation based perspective that is more closely related to the biophysics of neurons.
منابع مشابه
On the Asymptotic Equivalence Between Differential Hebbian and Temporal Difference Learning
In this theoretical contribution, we provide mathematical proof that two of the most important classes of network learning-correlation-based differential Hebbian learning and reward-based temporal difference learning-are asymptotically equivalent when timing the learning with a modulatory signal. This opens the opportunity to consistently reformulate most of the abstract reinforcement learning ...
متن کاملMathematical Description of Differential Hebbian Plasticity and its Relation to Reinforcement Learning
The human brain consists of more than a billion nerve cells, the neurons, each having several thousand connections, the synapses. These connections are not fixed but change all the time. In order to describe synaptic plasticity, different mathematical rules have been proposed most of which follow Hebb’s postulate. Donald Hebb suggested in 1949 that synapses only change if pre-synaptic activity,...
متن کاملNon-Local Thermo-Elastic Buckling Analysis of Multi-Layer Annular/Circular Nano-Plates Based on First and Third Order Shear Deformation Theories Using DQ Method
In present study, thermo-elastic buckling analysis of multi-layer orthotropic annular/circular graphene sheets is investigated based on Eringen’s theory. The moderately thick and also thick nano-plates are considered. Using the non-local first and third order shear deformation theories, the governing equations are derived. The van der Waals interaction between the layers is simulated for multi-...
متن کاملMulti-objective Differential Evolution for the Flow shop Scheduling Problem with a Modified Learning Effect
This paper proposes an effective multi-objective differential evolution algorithm (MDES) to solve a permutation flow shop scheduling problem (PFSSP) with modified Dejong's learning effect. The proposed algorithm combines the basic differential evolution (DE) with local search and borrows the selection operator from NSGA-II to improve the general performance. First the problem is encoded with a...
متن کاملAdaptive Agent Models Using Temporal Discounting, Memory Traces and Hebbian Learning with Inhibition, and their Rationality
In this paper three adaptive agent models incorporating triggered emotional responses are explored and evaluated on their rationality. One of the models is based on temporal discounting second on memory traces and the third one on hebbian learning with mutual inhibition. The models are assessed using a measure reflecting the environment’s behaviour and expressing the extent of rationality. Simu...
متن کامل