神经系统中随机和混沌感知信号的联想记忆与分割

为了模拟人与动物感知信息的真实环境，以脉动神经元节点组成神经元网络，研究在随机刺激和混沌刺激等极端条件下的记忆模式存储与时间分割问题.研究表明：网络对于若干种模式的叠加输入，能够以一部分神经元同步发放的形式在时间域上分割出每一模式. 如果输入模式是缺损的，系统能够把它们恢复到原型，即具有联想记忆功能.通过调节耦合强度和噪声强度等参数使得网络在中等强度噪声达到最优的时间分割，与广泛讨论的随机共振现象一致. 关键词： 神经网络 空时模式 联想记忆 随机共振     

Synchronization of map-based neurons with memory and synaptic delay

Synchronization of two synaptically coupled neurons with memory and synaptic delay is studied using the Rulkov map, one of the simplest neuron models which displays specific features inherent to bursting dynamics. We demonstrate a transition from lag to anticipated synchronization as the relationship between the memory duration and the synaptic delay time changes. The neuron maps synchronize either with anticipation, if the memory is longer than the synaptic delay time, or with lag otherwise. The mean anticipation time is equal to the difference between the memory and synaptic delay independently of the coupling strength. Frequency entrainment and phase-locking phenomena as well as a transition from regular spikes to chaos are demonstrated with respect to the coupling strength.     

Attractive target wave patterns in complex networks consisting of excitable nodes

This review describes the investigations of oscillatory complex networks consisting of excitable nodes,focusing on the target wave patterns or say the target wave attractors.A method of dominant phase advanced driving(DPAD) is introduced to reveal the dynamic structures in the networks supporting oscillations,such as the oscillation sources and the main excitation propagation paths from the sources to the whole networks.The target center nodes and their drivers are regarded as the key nodes which can completely determine the corresponding target wave patterns.Therefore,the center(say node A) and its driver(say node B) of a target wave can be used as a label,(A,B),of the given target pattern.The label can give a clue to conveniently retrieve,suppress,and control the target waves.Statistical investigations,both theoretically from the label analysis and numerically from direct simulations of network dynamics,show that there exist huge numbers of target wave attractors in excitable complex networks if the system size is large,and all these attractors can be labeled and easily controlled based on the information given by the labels.The possible applications of the physical ideas and the mathematical methods about multiplicity and labelability of attractors to memory problems of neural networks are briefly discussed.     

Analog-symbolic memory that tracks via reconsolidation

A fundamental part of a computational system is its memory, which is used to store and retrieve data. Classical computer memories rely on the static approach and are very different from human memories. Neural network memories are based on auto-associative attractor dynamics and thus provide a high level of pattern completion. However, they are not used in general computation since there are practically no algorithms to load an arbitrary landscape of attractors into them. In this sense neural network memory models cannot communicate well with symbolic and prior knowledge.We propose the design of a new memory based on localist attractor dynamics with reconsolidation called Reconsolidation Attractor Network (RAN). RAN combines symbolic and subsymbolic features in a very attractive way: it is based on attractors; enables pattern classification under missing data; and demonstrates dynamic reconsolidation, which is very useful for tracking changing concepts. The perception RAN enables is somewhat reminiscent of human perception due to its context sensitivity. Furthermore, it enables an immediate and clear interface with symbolic memories, including loading of attractors by means of trivial wiring, updating attractors, and retrieving them faster without waiting for full convergence. It also scales to any number of concepts. This provides a useful counterpoint to more conventional memory systems, such as random access memory and auto-associative neural networks.     

Parameter-dependent synchronization transition of coupled neurons with co-existing spiking and bursting

A firing pattern transition is simulated in the Leech neuron model, firstly from bursting to co-existence of spiking and bursting and then to spiking. The attraction domain of spiking and bursting for three different parameter values are calculated. Synchronization transition processes of two coupled Leech neurons, one is bursting and the other the co-existing spiking, are simulated for the three parameters. The three synchronization processes appear similar as the coupling strength increases, beginning from non-synchronization to complete synchronization through a complex dynamical procedure, but their detailed processes are different depending on the parameter values. The transition procedure is complex and the complete synchronization is in bursting for larger parameter values, while the process is simple with complete synchronization of spiking for smaller values. The potential relationship between complete synchronization and the attraction domain is also discussed. The results are instructive to understanding the synchronization behaviors of the coupled neuronal system with co-existing attractors.     

FitzHugh-Nagumo神经元网络的联想记忆与分割

以广泛讨论的Fitz Hugh-Nagumo神经元节点组成脉动神经元网络,从神经系统空时模式编码理论研究网络的记忆(或模式)存储与时间分割问题.给定一个输入模式,它是几种模式的叠加,网络能够以一部分神经元同步发放的形式一个接一个地分割出每一种模式.如果输入的模式有缺损,系统能够把它们恢复成原型,即神经网络的联想记忆功能.模拟需要调节耦合强度和噪声强度等参数使得网络在特定的参数值和中等强度噪声达到最优的时间分割,与广泛讨论的随机共振现象一致.     

耦合不连续系统同步转换过程中的多吸引子共存

本文研究了耦合不连续系统的同步转换过程中的动力学行为, 发现由混沌非同步到混沌同步的转换过程中特殊的多吸引子共存现象. 通过计算耦合不连续系统的同步序参量和最大李雅普诺夫指数随耦合强度的变化, 发现了较复杂的同步转换过程: 临界耦合强度之后出现周期非同步态(周期性窗口); 分析了系统周期态的迭代轨道,发现其具有两类不同的迭代轨道: 对称周期轨道和非对称周期轨道, 这两类周期吸引子和同步吸引子同时存在, 系统表现出对初值敏感的多吸引子共存现象. 分析表明, 耦合不连续系统中的周期轨道是由于局部动力学的不连续特性和耦合动力学相互作用的结果. 最后, 对耦合不连续系统的同步转换过程进行了详细的分析, 结果表明其同步呈现出较复杂的转换过程.     

Two-Layer Feedback Neural Networks with Associative Memories

We construct a two-layer feedback neural network by a Monte Carlo based algorithm to store memories as fixed-point attractors or as limit-cycle attractors. Special attention is focused on comparing the dynamics of the network with limit-cycle attractors and with fixed-point attractors. It is found that the former has better retrieval property than the latter. Particularly, spurious memories may be suppressed completely when the memories are stored as a long-limit cycle. Potential application of limit-cycle-attractor networks is discussed briefly.     

Unstable attractors induce perpetual synchronization and desynchronization

Common experience suggests that attracting invariant sets in nonlinear dynamical systems are generally stable. Contrary to this intuition, we present a dynamical system, a network of pulse-coupled oscillators, in which unstable attractors arise naturally. From random initial conditions, groups of synchronized oscillators (clusters) are formed that send pulses alternately, resulting in a periodic dynamics of the network. Under the influence of arbitrarily weak noise, this synchronization is followed by a desynchronization of clusters, a phenomenon induced by attractors that are unstable. Perpetual synchronization and desynchronization lead to a switching among attractors. This is explained by the geometrical fact, that these unstable attractors are surrounded by basins of attraction of other attractors, whereas the full measure of their own basin is located remote from the attractor. Unstable attractors do not only exist in these systems, but moreover dominate the dynamics for large networks and a wide range of parameters.     

Corticonic models of brain mechanisms underlying cognition and intelligence

The concern of this review is brain theory or more specifically, in its first part, a model of the cerebral cortex and the way it: (a) interacts with subcortical regions like the thalamus and the hippocampus to provide higher-level-brain functions that underlie cognition and intelligence, (b) handles and represents dynamical sensory patterns imposed by a constantly changing environment, (c) copes with the enormous number of such patterns encountered in a lifetime by means of dynamic memory that offers an immense number of stimulus-specific attractors for input patterns (stimuli) to select from, (d) selects an attractor through a process of “conjugation” of the input pattern with the dynamics of the thalamo–cortical loop, (e) distinguishes between redundant (structured) and non-redundant (random) inputs that are void of information, (f) can do categorical perception when there is access to vast associative memory laid out in the association cortex with the help of the hippocampus, and (g) makes use of “computation” at the edge of chaos and information driven annealing to achieve all this. Other features and implications of the concepts presented for the design of computational algorithms and machines with brain-like intelligence are also discussed. The material and results presented suggest, that a Parametrically Coupled Logistic Map network (PCLMN) is a minimal model of the thalamo–cortical complex and that marrying such a network to a suitable associative memory with re-entry or feedback forms a useful, albeit, abstract model of a cortical module of the brain that could facilitate building a simple artificial brain.

In the second part of the review, the results of numerical simulations and drawn conclusions in the first part are linked to the most directly relevant works and views of other workers. What emerges is a picture of brain dynamics on the mesoscopic and macroscopic scales that gives a glimpse of the nature of the long sought after brain code underlying intelligence and other higher level brain functions.     

Phase synchronization in two coupled chaotic neurons

Chaotically-spiking dynamics of Hindmarsh–Rose neurons are discussed based on a flexible definition of phase for chaotic flow. The phase synchronization of two coupled chaotic neurons is in fact the spike synchronization. As a multiple time-scale model, the coupled HR neurons have quite different behaviors from the Rössler oscillators only having single time-scale mechanism. Using such a multiple time-scale model, the phase function can detect synchronization dynamics that cannot be distinguished by cross-correlation. Moreover, simulation results show that the Lyapunov exponents cannot be used as a definite criterion for the occurrence of chaotic phase synchronization for such a system. Evaluation of the phase function shows its utility in analyzing nonlinear neural systems.     

Efficient Hopfield pattern recognition on a scale-free neural network

Neural networks are supposed to recognise blurred images (or patterns) of N pixels (bits) each. Application of the network to an initial blurred version of one of P pre-assigned patterns should converge to the correct pattern. In the “standard" Hopfield model, the N “neurons” are connected to each other via N² bonds which contain the information on the stored patterns. Thus computer time and memory in general grow with N². The Hebb rule assigns synaptic coupling strengths proportional to the overlap of the stored patterns at the two coupled neurons. Here we simulate the Hopfield model on the Barabási-Albert scale-free network, in which each newly added neuron is connected to only m other neurons, and at the end the number of neurons with q neighbours decays as 1/q ³. Although the quality of retrieval decreases for small m, we find good associative memory for 1 ≪ m ≪ N. Hence, these networks gain a factor N/m ≫ 1 in the computer memory and time. Received 12 January 2003 Published online 11 April 2003 RID="a" ID="a"e-mail: stauffer@thp.uni-koeln.de     

Dynamics of a neural network model with finite connectivity and cycle stored patterns

The spatiotemporal evolution and memory retrieval properties of a Hopfield-like neural network with cycle-stored patterns and finite connectivity are studied. The analytical studies on a mean-field version show that, given the number of stored patterns p, there is a critical connectivity k_c such that the retrieval states are stable fixed points if and only if k > k_c. The dependence of k_c on the number of stored patterns is also present. The numerical simulations are applied to the short-ranged model with local interaction. It is revealed that, given p, the memory retrieval function is kept if the connectivity is high enough while the dynamics of the system is in the frozen phase. However when the connectivity k is less than a critical value k_c the system is in the chaotic phase and loses its memory retrieval ability. The critical points of both the dynamical phase transition and memory-loss phase transition are obtained by simulation data.     

Chimeras in leaky integrate-and-fire neural networks: effects of reflecting connectivities

The effects of attracting-nonlocal and reflecting connectivity are investigated in coupled Leaky Integrate-and-Fire (LIF) elements, which model the exchange of electrical signals between neurons. Earlier investigations have demonstrated that repulsive-nonlocal and hierarchical network connectivity can induce complex synchronization patterns and chimera states in systems of coupled oscillators. In the LIF system we show that if the elements are nonlocally linked with positive diffusive coupling on a ring network, the system splits into a number of alternating domains. Half of these domains contain elements whose potential stays near the threshold and they are interrupted by active domains where the elements perform regular LIF oscillations. The active domains travel along the ring with constant velocity, depending on the system parameters. When we introduce reflecting coupling in LIF networks unexpected complex spatio-temporal structures arise. For relatively extensive ranges of parameter values, the system splits into two coexisting domains: one where all elements stay near the threshold and one where incoherent states develop, characterized by multi-leveled mean phase velocity profiles.     

Stability of discrete memory states to stochastic fluctuations in neuronal systems

Noise can degrade memories by causing transitions from one memory state to another. For any biological memory system to be useful, the time scale of such noise-induced transitions must be much longer than the required duration for memory retention. Using biophysically-realistic modeling, we consider two types of memory in the brain: short-term memories maintained by reverberating neuronal activity for a few seconds, and long-term memories maintained by a molecular switch for years. Both systems require persistence of (neuronal or molecular) activity self-sustained by an autocatalytic process and, we argue, that both have limited memory lifetimes because of significant fluctuations. We will first discuss a strongly recurrent cortical network model endowed with feedback loops, for short-term memory. Fluctuations are due to highly irregular spike firing, a salient characteristic of cortical neurons. Then, we will analyze a model for long-term memory, based on an autophosphorylation mechanism of calcium/calmodulin-dependent protein kinase II (CaMKII) molecules. There, fluctuations arise from the fact that there are only a small number of CaMKII molecules at each postsynaptic density (putative synaptic memory unit). Our results are twofold. First, we demonstrate analytically and computationally the exponential dependence of stability on the number of neurons in a self-excitatory network, and on the number of CaMKII proteins in a molecular switch. Second, for each of the two systems, we implement graded memory consisting of a group of bistable switches. For the neuronal network we report interesting ramping temporal dynamics as a result of sequentially switching an increasing number of discrete, bistable, units. The general observation of an exponential increase in memory stability with the system size leads to a trade-off between the robustness of memories (which increases with the size of each bistable unit) and the total amount of information storage (which decreases with increasing unit size), which may be optimized in the brain through biological evolution.     

一个全局耦合不连续映像格子中的冻结化随机图案模式

本文研究了一类既不连续又不可逆分段线性映像构成的全局耦合映像格子系统中的一类典型集体动力学行为, 即冻结化随机图案模式. 计算了平均同步序参量和最大李雅普诺夫指数随耦合强度的变化. 结果显示, 当耦合强度超过某个阈值后, 在给定动力学变量的初始下, 系统几乎都能达到完全或部分同步状态, 出现冻结化随机图案. 这些现象表明, 耦合映像格子系统中存在着多个共存的吸引子. 因此, 其冻结化图案的结构和分布敏感地依赖于格点动力学变量初始值的选取. 感兴趣地是, 即使当单映像处于混沌状态时, 格点间的耦合仍能将系统调制到规则的运动状态, 这种特征对于混沌控制具有重要的利用价值. 上述丰富动力学行为的出现是由于单映像中不连续性和不可逆性相互作用的结果.     

Noise with memory as a model of lemming cycles

Population cycles in small rodents of the north are modeled by noise with memory. Multiannual lemming density fluctuations are presented as a pulse sequence. These pulses correspond to the peaks of lemming density. The memory is presented as some delay time after each pulse. During this time the next pulse is forbidden. Parameter of periodicity, average period, correlation function and parameter of synchronization are calculated for different places of North America. Examples of equations modeling population dynamics of lemmings (or their predators) are considered. The model of connected oscillators gives the qualitative explanation of synchronization effects and relation between synchronization and periodicity. These results have implication for the testing of hypotheses regarding lemming cycles.