AHaH computing-from metastable switches to attractors to machine learning

BPSK Circuit Based on SDC Memristor

Digital communication based on memristors is a new field. The main principle is to construct a modulation and demodulation circuit by using the resistance variation characteristics of the memristor. Based on the establishment of the Knowm memristor simulation model, firstly, the modulation circuit is designed by using the polarity and symmetry of the memristor and combined with the commercial current feedback amplifier AD844. It is proved that the modulated signal based on the memristor is a strong function of phase, and the demodulation circuit is designed accordingly. All simulation circuits are based on the actual commercial physical device model. The analytical expression of the output signal of the modulation and demodulation circuit is deduced theoretically, and the communication performance of the whole system is simulated by LTSpice. At the same time, the influence of the parasitic capacitance of the memristor on the circuit performance is also considered. After the simulation verification, the hardware circuit experiment of the modulation and demodulation circuit is carried out. The waveforms of the modulated signal and the demodulated signal are measured by an oscilloscope. The experimental results are completely consistent with the simulation and theoretical results.

Electromagnetic Interference Effects of Continuous Waves on Memristors: A Simulation Study

As two-terminal passive fundamental circuit elements with memory characteristics, memristors are promising devices for applications such as neuromorphic systems, in-memory computing, and tunable RF/microwave circuits. The increasingly complex electromagnetic interference (EMI) environment threatens the reliability of memristor systems. However, various EMI signals' effects on memristors are still unclear. This paper selects continuous waves (CWs) as EMI signals. It provides a deeper insight into the interference effect of CWs on the memristor driven by a sinusoidal excitation voltage, as well as a method for investigating the EMI effect of memristors. The optimal memristor model is obtained by the exhaustive traversing of the possible model parameters, and the interference effect of CWs on memristors is quantified based on this model and the proposed evaluation metrics. Simulation results indicate that CW interference may affect the switching time, dynamic range, nonlinearity, symmetry, time to the boundary, and variation of memristance. The specific interference effect depends on the operating mode of the memristor, the amplitude, and the frequency of the CW. This research provides a foundation for evaluating EMI effects and designing electromagnetic protection for memristive neuromorphic systems.

Structural and Parametric Identification of Knowm Memristors

This paper proposes a novel identification method for memristive devices using Knowm memristors as an example. The suggested identification method is presented as a generalized process for a wide range of memristive elements. An experimental setup was created to obtain a set of intrinsic I-V curves for Knowm memristors. Using the acquired measurements data and proposed identification technique, we developed a new mathematical model that considers low-current effects and cycle-to-cycle variability. The process of parametric identification for the proposed model is described. The obtained memristor model represents the switching threshold as a function of the state variables vector, making it possible to account for snapforward or snapback effects, frequency properties, and switching variability. Several tools for the visual presentation of the identification results are considered, and some limitations of the proposed model are discussed.

Gradient Decomposition Methods for Training Neural Networks With Non-ideal Synaptic Devices

While promising for high-capacity machine learning accelerators, memristor devices have non-idealities that prevent software-equivalent accuracies when used for online training. This work uses a combination of Mini-Batch Gradient Descent (MBGD) to average gradients, stochastic rounding to avoid vanishing weight updates, and decomposition methods to keep the memory overhead low during mini-batch training. Since the weight update has to be transferred to the memristor matrices efficiently, we also investigate the impact of reconstructing the gradient matrixes both internally (rank-seq) and externally (rank-sum) to the memristor array. Our results show that streaming batch principal component analysis (streaming batch PCA) and non-negative matrix factorization (NMF) decomposition algorithms can achieve near MBGD accuracy in a memristor-based multi-layer perceptron trained on the MNIST (Modified National Institute of Standards and Technology) database with only 3 to 10 ranks at significant memory savings. Moreover, NMF rank-seq outperforms streaming batch PCA rank-seq at low-ranks making it more suitable for hardware implementation in future memristor-based accelerators.

Memristive GAN in Analog

Generative Adversarial Network (GAN) requires extensive computing resources making its implementation in edge devices with conventional microprocessor hardware a slow and difficult, if not impossible task. In this paper, we propose to accelerate these intensive neural computations using memristive neural networks in analog domain. The implementation of Analog Memristive Deep Convolutional GAN (AM-DCGAN) using Generator as deconvolutional and Discriminator as convolutional memristive neural network is presented. The system is simulated at circuit level with 1.7 million memristor devices taking into account memristor non-idealities, device and circuit parameters. The design is modular with crossbar arrays having a minimum average power consumption per neural computation of 47nW. The design exclusively uses the principles of neural network dropouts resulting in regularization and lowering the power consumption. The SPICE level simulation of GAN is performed with 0.18 μm CMOS technology and WO_x memristive devices with R_ON = 40 kΩ and R_OFF = 250 kΩ, threshold voltage 0.8 V and write voltage at 1.0 V.

Memristors for the Curious Outsiders

We present both an overview and a perspective of recent experimental advances and proposed new approaches to performing computation using memristors. A memristor is a 2-terminal passive component with a dynamic resistance depending on an internal parameter. We provide an brief historical introduction, as well as an overview over the physical mechanism that lead to memristive behavior. This review is meant to guide nonpractitioners in the field of memristive circuits and their connection to machine learning and neural computation.

Nano-Modeling and Computation in Bio and Brain Dynamics

The study of brain dynamics currently utilizes the new features of nanobiotechnology and bioengineering. New geometric and analytical approaches appear very promising in all scientific areas, particularly in the study of brain processes. Efforts to engage in deep comprehension lead to a change in the inner brain parameters, in order to mimic the external transformation by the proper use of sensors and effectors. This paper highlights some crossing research areas of natural computing, nanotechnology, and brain modeling and considers two interesting theoretical approaches related to brain dynamics: (a) the memory in neural network, not as a passive element for storing information, but integrated in the neural parameters as synaptic conductances; and (b) a new transport model based on analytical expressions of the most important transport parameters, which works from sub-pico-level to macro-level, able both to understand existing data and to give new predictions. Complex biological systems are highly dependent on the context, which suggests a “more nature-oriented” computational philosophy.

Open source libraries and frameworks for biological data visualisation: a guide for developers

Recent advances in high-throughput experimental techniques have led to an exponential increase in both the size and the complexity of the data sets commonly studied in biology. Data visualisation is increasingly used as the key to unlock this data, going from hypothesis generation to model evaluation and tool implementation. It is becoming more and more the heart of bioinformatics workflows, enabling scientists to reason and communicate more effectively. In parallel, there has been a corresponding trend towards the development of related software, which has triggered the maturation of different visualisation libraries and frameworks. For bioinformaticians, scientific programmers and software developers, the main challenge is to pick out the most fitting one(s) to create clear, meaningful and integrated data visualisation for their particular use cases. In this review, we introduce a collection of open source or free to use libraries and frameworks for creating data visualisation, covering the generation of a wide variety of charts and graphs. We will focus on software written in Java, JavaScript or Python. We truly believe this software offers the potential to turn tedious data into exciting visual stories.