Design and Implementation of Neuromorphic Circuits and Neural Classifier

Continuous improvements in the VLSI domain have enabled the technology to mimic the neuro biological architectures present in nervous system. Traditionally, the concept of artificial limbs was merely an idea but now a days because of neuromorphic circuits this idea can be implemented successfully. Neuromorphic circuits are the electronic analog circuits which is used to imitate the neuro-biological architectures present in our nervous system. These circuits are implemented using VLSI technology and works as a data processing devices and intelligent sensors that has an upper hand than digital processors. In contrast to digital processors, their storage and processing capacity is distributed. They are robust or can use learning algorithm instead of being programmed. They are asynchronous and requires no clock signal. The term neuromorphic signifies the characterization of analog, digital, mixed-mode analog/digital VLSI, and software systems that are used to design the models of neural systems. The neuromorphic ciruits at the hardware level can be implemented by using oxide-based memristors, threshold switches, and transistors. These circuits are not only limited to the application of artificial limbs but can be used in all the applications that are rich in visual or auditory data and requires a machine to adjust its behaviour as it interacts with the world. In recent times, the neuromorphic circuits are implemented using voltage mode which requires a large power dissipation in the circuits due to which the implantation of these circuits inside the human body cannot be possible. But in this work, the circuits were implemented in current mode which enables the circuit to operate in sub threshold region. The current mode implementation of the circuits provides very less power dissipation as compared to the voltage mode implementation of the circuits. So these circuits requires very small size and can be easily implanted in an human brain.