Shyam Prasad Adhikari

Three Fingerprints of Memristor

This paper illustrates that for a device to be a memristor it should exhibit three characteristic fingerprints: 1) When driven by a bipolar periodic signal the device must exhibit a “pinched hysteresis loop” in the voltage-current plane, assuming the response is periodic. 2) Starting from some critical frequency, the hysteresis lobe area should decrease monotonically as the excitation frequency increases, and 3) the pinched hysteresis loop should shrink to a single-valued function when the frequency tends to infinity. Examples of memristors exhibiting these three fingerprints, along with non-memristors exhibiting only a subset of these fingerprints are also presented. In addition, two different types of pinched hysteresis loops; the transversal (self-crossing) and the non-transversal (tangential) loops exhibited by memristors are also discussed with its identification criterion.

Memristor Bridge Synapse-Based Neural Network and Its Learning

Analog hardware architecture of a memristor bridge synapse-based multilayer neural network and its learning scheme is proposed. The use of memristor bridge synapse in the proposed architecture solves one of the major problems, regarding nonvolatile weight storage in analog neural network implementations. To compensate for the spatial nonuniformity and nonideal response of the memristor bridge synapse, a modified chip-in-the-loop learning scheme suitable for the proposed neural network architecture is also proposed. In the proposed method, the initial learning is conducted in software, and the behavior of the software-trained network is learned by the hardware network by learning each of the single-layered neurons of the network independently. The forward calculation of the single-layered neuron learning is implemented on circuit hardware, and followed by a weight updating phase assisted by a host computer. Unlike conventional chip-in-the-loop learning, the need for the readout of synaptic weights for calculating weight updates in each epoch is eliminated by virtue of the memristor bridge synapse and the proposed learning scheme. The hardware architecture along with the successful implementation of proposed learning on a three-bit parity network, and on a car detection network is also presented.

A Circuit-Based Learning Architecture for Multilayer Neural Networks With Memristor Bridge Synapses

Memristor-based circuit architecture for multilayer neural networks is proposed. It is a first of its kind demonstrating successful circuit-based learning for multilayer neural network built with memristors. Though back-propagation algorithm is a powerful learning scheme for multilayer neural networks, its hardware implementation is very difficult due to complexities of the neural synapses and the operations involved in the learning algorithm. In this paper, the circuit of a multilayer neural network is designed with memristor bridge synapses and the learning is realized with a simple learning algorithm called Random Weight Change (RWC). Though RWC algorithm requires more iterations than back-propagation algorithm, we show that a circuit-based learning using RWC is two orders faster than its software counterpart. The method to build a multilayer neural network using memristor bridge synapses and a circuit-based learning architecture of RWC algorithm is proposed. Comparison between software-based and memristor circuit-based learning are presented via simulations.

Composite Behavior of Multiple Memristor Circuits

Composite characteristics of the parallel and serial connections of memristors are investigated. The memristor is one of the fundamental electrical elements, which has recently been successfully built. However, its electrical characteristics are not yet fully understood. When multiple memristors are connected to each other, the composite behavior of the devices becomes complicated and is difficult to predict, due to the polarity dependent nonlinear variation in the memristance of individual memristors. In this work, we investigate the relationships among flux, charge, and memristance of diverse composite memristors, using both linear and nonlinear memristor models, and analyze the characteristics of complex memristor circuits.

Memistor Is Not Memristor [Express Letters]

This note clarifies the circuit-theoretic differences between a memristor and a memistor.

HIGHLY ACCURATE DOUBLET GENERATOR FOR MEMRISTOR-BASED ANALOG MEMORY

A novel doublet generator circuit for memristor-based analog memories or artificial synapses is presented. In memristor-based analog memories or artificial synapses, the read-out pulses cause a drifting problem in the programmed resistance of the memristor. Use of doublet pulse is known to be an efficient solution for preventing resistance variation in memristors. Switching speed and area similarity between the positive and the negative cycles of doublet are two extremely important factors in designing a doublet generator. In this paper, a fast and highly accurate doublet generator where the difference between the positive and negative cycles of pulses is less than 0.012% is proposed.

Why Are Memristor and Memistor Different Devices

This paper presents a circuit-theoretic foundation of the “memristor,” and clarifies why it is fundamentally different from a 3-terminal device with a similarly-sounding name called the “memistor.” Here we show that while the memristor is a basic 2-terminal circuit element based on classic nonlinear circuit theory, the memistor is an ad hoc 3-terminal gadget devised for one specific application, and does not qualify as a 3-terminal circuit element because it is impossible to predict its behavior when connected with other circuit elements.

A Circuit-Based Neural Network with Hybrid Learning of Backpropagation and Random Weight Change Algorithms

A hybrid learning method of a software-based backpropagation learning and a hardware-based RWC learning is proposed for the development of circuit-based neural networks. The backpropagation is known as one of the most efficient learning algorithms. A weak point is that its hardware implementation is extremely difficult. The RWC algorithm, which is very easy to implement with respect to its hardware circuits, takes too many iterations for learning. The proposed learning algorithm is a hybrid one of these two. The main learning is performed with a software version of the BP algorithm, firstly, and then, learned weights are transplanted on a hardware version of a neural circuit. At the time of the weight transplantation, a significant amount of output error would occur due to the characteristic difference between the software and the hardware. In the proposed method, such error is reduced via a complementary learning of the RWC algorithm, which is implemented in a simple hardware. The usefulness of the proposed hybrid learning system is verified via simulations upon several classical learning problems.

Learning with memristor bridge synapse-based neural networks

A learning architecture for memristor-based multilayer neural networks is proposed in this paper. A multilayer neural network is implemented based on memristor bridge synapses and its learning is performed with Random Weight Change architecture. The memristor bridge synapses are composed of bridge type architectures of back-to-back connected 4 memristors and the Random Weight Change (RWC) algorithm is based on a simple trial-and-error learning. Though the RWC algorithm requires more iterations than backpropagation, learning time is two orders faster than that of a software counterpart due to the benefit of circuit-based learning.

Composite memristance of parallel and serial memristor circuits

When multiple memristors are connected to each other, the composite behavior of the devices becomes complicated and is difficult to predict, due to the polarity dependent nonlinear variation in the memristance of individual memristor. In this paper, we investigate the relationships among flux, charge and memristance of diverse composite memristors, using the HP TiO2 model, and analyze the characteristics of complex memristor circuits. It is assumed that all memristor circuits operate at a stable composite memristance state, in which the composite flux curve does not vary and the memristor circuits act as a single memristive system, regardless of input current or voltage. Such study will be conducted for serial and parallel memristor circuits.