Cognitive approach to Memristor which is able to associative learning
18th International Conference on Neurology and Neurological Disorders
August 23-24, 2018 | Paris, France
Hakki Halil Babacan, Salih Erkan and Gulcin Arkis
Istanbul University, Turkey
Posters & Accepted Abstracts : J Neurol Neurorehabil Res
Abstract:
Objective: We are at a time when electronic systems are
structured in a manner similar to the human brain. The
memristor, a neuromorphic circuit designed by Chua in 1971,
is a modeling of synaptic learning and associative learning.
Neuromorphic circuit elements and memristor can be used in
artificial brain formation in the later periods and in the treatment
of various lesions, psychiatric and neurological diseases.
Scientific publications of memristor related neuroscientists,
behavioral scientists, cognitive scientists and psychologists are
scarce. The aim of this review is to examine the learning models
built on the memristor by cognitive perspective.
Methods: In this study, the learning experiments on the
memristor were investigated in the literature and the results
were compared.
Results: In conditional learning experiments on the memristor,
which is its own memory, the unconditional stimulus and the
neutral stimulus represent different types of signals. Before the
learning, the signals which are denoted as neutral stimuli cannot
give output from the electronic angle. But just like PavlovâÃÂÃÂs dog
experiment, when the signal representing the unconditioned
stimulus was presented before learning, the output is taken.
When both stimuli were presented in the order of the Pavlov
experiment, the output was taken from the neutral stimulus.
And after learning, the output can be taken when the neutral
stimulus given alone. In this way, the memristors were able to
learn conditionally and to achieve synaptic modeling.
Conclusion: It has been found that learning procedures can be
applied to hardware devices other than algorithmic devices. The
learning experiments on the memristor successfully support
the synaptic learning and Pavlov type conditional learning
procedures. In some experiments, however, the conditional
responses in the memristor do not decrease over time. This
can be described as a pathological learning and may reduce the
efficiency of the memristor.
Biography:
E-mail:
PDF HTML