Artificial Neuron and Synapse Realized in an Antiferromagnet/Ferromagnet Heterostructure Using Dynamics of Spin–Orbit Torque Switching

Efficient information processing in the human brain is achieved by dynamics of neurons and synapses, motivating effective implementation of artificial spiking neural networks. Here, the dynamics of spin–orbit torque switching in antiferromagnet/ferromagnet heterostructures is studied to show the capability of the material system to form artificial neurons and synapses for asynchronous spiking neural networks. The magnetization switching, driven by a single current pulse or trains of pulses, is examined as a function of the pulse width (1 s to 1 ns), amplitude, number, and pulse‐to‐pulse interval. Based on this dynamics and the unique ability of the system to exhibit binary or analog behavior depending on the device size, key functionalities of a synapse (spike‐timing‐dependent plasticity) and a neuron (leaky integrate‐and‐fire) are reproduced in the same material and on the basis of the same working principle. These results open a way toward spintronics‐based neuromorphic hardware that executes cognitive tasks with the efficiency of the human brain.     

一种具有small world 特性的ESN 结构分析与设计

针对回声状态网络（ESN）结构设计复杂、参数选择难度大的问题,提出一种具有small world特性的ESN（SWESN）.首先采用神经元空间增长算法在平面区域生成small world拓扑网络;然后根据网络节点与基准点的Euclidean距离将网络节点进行重新排序,并将平面上的物理节点及其连接映射为SWESN的内部神经元连接矩阵,从而使动态神经元池具有small world特性.实验表明,SWESN动力学特性比常规ESN更为丰富,在鲁棒性、抗干扰能力等方面均优于常规的ESN.     

一种改进的复数BP神经网络算法研究

论文提出了复数BP神经网络的一种新结构和算法。算法的主要思想是将复值输人信号的实部和虚部分离,分别训练,使其达到稳定状态。其结构简单,易于实现,只需少量样本点却有很高收敛速度和精度。通过实验和仿真说明论文算法的有效性。     

Synchronization of multiple chaotic FitzHugh-Nagumo neurons with gap junctions under external electrical stimulation

This paper discusses the synchronization of three coupled chaotic FitzHugh-Nagumo (FHN) neurons with different gap junctions under external electrical stimulation. A nonlinear control law that guarantees the asymptotic synchronization of coupled neurons (with reduced computations) is proposed. The developed control law incorporates the synchronization error between two slave neurons in addition to the conventionally considered synchronization errors between the master and the slave neurons, which make the proposed scheme computationally more efficient. Further, a novel L₂ gain reduction criterion has been developed for multi-input multi-output systems with non-zero initial conditions, and is applied to robust synchronization of FHN neurons under L₂ norm bounded disturbance and uncertainties. Furthermore, a robust adaptive nonlinear control law is developed, which is capable of handling variations in nonlinear part of synchronization error dynamics, without using any neural-network-based training-oriented adaptive scheme. The proposed control schemes ensure global synchronization with computational simplicity, easy way of design and implementation and avoiding extra measurements. The results obtained with the proposed control laws are verified through numerical simulations.     

An architecture-adaptive neural network online control system

An architecture-adaptive intelligent self-tuning control system is presented. The system is composed of the supervisor module, the model refinement module, the process plant and the database. In the supervisor module, the user prescribes the desired curve for the plant dynamic process. The model refinement module is in parallel with the process plant, and consists of the self-tuning process model, which contains an architecture-adaptive neural network. The model refinement module could learn intelligently the real process plant by the prompt adjustments based on the difference of the outputs of the two modules, and its learned model is also refined gradually. This diagram is especially versatile in the complex nonlinear and time-variant systems in practice.     

Universal Approximation and QoS Violation Application of Extreme Learning Machine

Neural networks have been successfully applied to many applications due to their approximation capability. However, complicated network structures and algorithms will lead to computational and time-consuming burdens. In order to satisfy demanding real-time requirements, many fast learning algorithms were explored in the past. Recently, a fast algorithm, Extreme Learning Machine (ELM) (Huang et al. 70:489–501, 2006) was proposed. Unlike conventional algorithms whose neurons need to be tuned, the input-to-hidden neurons of ELM are randomly generated. Though a large number of experimental results have shown that input-to-hidden neurons need not be tuned, there lacks a rigorous proof whether ELM possesses the universal approximation capability. In this paper, based on the universal approximation property of an orthonormal method, we firstly illustrate the equivalent relationship between ELM and the orthonormal method, and further prove that neural networks with ELM are also universal approximations. We also successfully apply ELM to the identification of QoS violation in the multimedia transmission.     

Some new neural network architectures with improved learning schemes

 Here, we present two new neuron model architectures and one modified form of existing standard feedforward architecture (MSTD). Both the new models use self-scaling scaled conjugate gradient algorithm (SSCGA) and lambda–gamma (L–G) algorithm and entail the properties of basic as well as higher order neurons (i.e., multiplication and the aggregation functions). Of these two, compensatory neural network architecture (CNNA) requires relatively smaller number of inter-neuronal connections, cuts down on the computational budget by almost 50% and speeds up convergence, besides, gives better training and prediction accuracy. The second model sigma–pi–sigma (SPS) ensures faster convergence, better training and prediction accuracy. The third model (MSTD) performs much better than the standard feedforward architecture (STD). The effect of normalizing the outputs for training also studied here shows virtually no improvement, at low iteration level, say ∼500, with increasing range of scaling. Increasing the number of neurons beyond a point also shows to have little effect in the case of higher order neuron.The numerous simulation runs for the problem of satellite orbit determination and the complex XOR problems establishes the robustness of the proposed neuron models architectures.     

跨越连接多层神经网络的稳定性

研究了跨越连接多层神经网络的结构和稳定性。提出了有跨越连接的误差反向传播算法,通过仿真实验说明了跨越连接多层神经网络收敛性和稳定性优于传统网络。     

基于复数权神经元的多值整形器稳健设计

复数权神经元是由于引入多阈值逻辑而使得权值为复数的神经元,被认为具有更强的性能。该文根据该神经元数学模型,结合POST代数系统中多值逻辑取值及运算的概念,实现了基于单个复数权神经元的多值整形运算,设计的多值逻辑整形运算具有稳健性能,所得结果表明用复数权神经元实现多值逻辑的有效性和可行性,说明了其强大的信息处理能力。     

A potential kairomone stimulates pheromone-responsive receptor neurons inUtetheisa ornatrix (Lepidoptera: Arctiidae)

The olfactory receptor neurons in basiconic sensilla on the antennae ofUtetheisa ornatrix, which, in females, had earlier been shown to be responsive to stimulation with hydroxydanaidal (HD), are here shown to be responsive to volatile substances in samples of pyrrolizidine alkaloids (PAs). These latter substances are secondary plant metabolites present in the host plant of the larvae. Their sequestration during larval life serves to protect all life stages from predation. In males, the PAs also provide precursors for the production of the male pheromone, HD. In females, basiconic receptor neurons begin to respond to stimulus cartridges containing 1 ng of (R)-(–)-hydroxydanaidal, 100 ng of its isomer, (S)-(+)-hydroxydanaidal, and to volatiles emanating from 10g of the alkaloids monocrotaline and heliotrine. Receptor neurons in males are generally responsive to the same array of substances but with reduced sensitivity. The dietary background of the adult moths with respect to prior ingestion of PAs does not influence the response capabilities of basiconic receptor neurons to HD or to the volatile components of PAs. Earlier studies in another arctiid (Rhodogastria), had indicated that trace amounts of HD are present in PAs, presumably as their hydrolysis product. Thus we assume that, under natural conditions, HD may be an active component of the volatiles from PAs and may serve as both a male pheromone and a kairomone. Sensitivity to HD could thereby provide information about the location and PA content of potential mates and food plants. Mechanisms that may have resulted in the evolution of a signaling system with these properties are discussed.