Latent attractors: a model for context-dependent place representations in the hippocampus

Cells throughout the rodent hippocampal system show place-specific patterns of firing called place fields, creating a coarse-coded representation of location. The dependencies of this place code--or cognitive map--on sensory cues have been investigated extensively, and several computational models have been developed to explain them. However, place representations also exhibit strong dependence on spatial and behavioral context, and identical sensory environments can produce very different place codes in different situations. Several recent studies have proposed models for the computational basis of this phenomenon, but it is still not completely understood. In this article, we present a very simple connectionist model for producing context-dependent place representations in the hippocampus. We propose that context dependence arises in the dentate gyrus-hilus (DGH) system, which functions as a dynamic selector, disposing a small group of granule and pyramidal cells to fire in response to afferent stimulus while depressing the rest. It is hypothesized that the DGH system dynamics has "latent attractors," which are unmasked by the afferent input and channel system activity into subpopulations of cells in the DG, CA3, and other hippocampal regions as observed experimentally. The proposed model shows that a minimally structured hippocampus-like system can robustly produce context-dependent place codes with realistic attributes.     

网格细胞到位置细胞的竞争型神经网络模型

针对老鼠海马结构中网格细胞到位置细胞的信息传递问题, 构建网格细胞到位置细胞的竞争型神经网络模型. 在一维和二维环境中的仿真结果均符合生物学研究事实, 结果表明, 模型能够模拟齿状回和海马中位置细胞的放电特性, 可有效解释位置细胞位置野的形成机理.

     

Optimal population codes for space: grid cells outperform place cells

Rodents use two distinct neuronal coordinate systems to estimate their position: place fields in the hippocampus and grid fields in the entorhinal cortex. Whereas place cells spike at only one particular spatial location, grid cells fire at multiple sites that correspond to the points of an imaginary hexagonal lattice. We study how to best construct place and grid codes, taking the probabilistic nature of neural spiking into account. Which spatial encoding properties of individual neurons confer the highest resolution when decoding the animal's position from the neuronal population response? A priori, estimating a spatial position from a grid code could be ambiguous, as regular periodic lattices possess translational symmetry. The solution to this problem requires lattices for grid cells with different spacings; the spatial resolution crucially depends on choosing the right ratios of these spacings across the population. We compute the expected error in estimating the position in both the asymptotic limit, using Fisher information, and for low spike counts, using maximum likelihood estimation. Achieving high spatial resolution and covering a large range of space in a grid code leads to a trade-off: the best grid code for spatial resolution is built of nested modules with different spatial periods, one inside the other, whereas maximizing the spatial range requires distinct spatial periods that are pairwisely incommensurate. Optimizing the spatial resolution predicts two grid cell properties that have been experimentally observed. First, short lattice spacings should outnumber long lattice spacings. Second, the grid code should be self-similar across different lattice spacings, so that the grid field always covers a fixed fraction of the lattice period. If these conditions are satisfied and the spatial "tuning curves" for each neuron span the same range of firing rates, then the resolution of the grid code easily exceeds that of the best possible place code with the same number of neurons.     

Hebbian imprinting and retrieval in oscillatory neural networks

We introduce a model of generalized Hebbian learning and retrieval in oscillatory neural networks modeling cortical areas such as hippocampus and olfactory cortex. Recent experiments have shown that synaptic plasticity depends on spike timing, especially on synapses from excitatory pyramidal cells, in hippocampus, and in sensory and cerebellar cortex. Here we study how such plasticity can be used to form memories and input representations when the neural dynamics are oscillatory, as is common in the brain (particularly in the hippocampus and olfactory cortex). Learning is assumed to occur in a phase of neural plasticity, in which the network is clamped to external teaching signals. By suitable manipulation of the nonlinearity of the neurons or the oscillation frequencies during learning, the model can be made, in a retrieval phase, either to categorize new inputs or to map them, in a continuous fashion, onto the space spanned by the imprinted patterns. We identify the first of these possibilities with the function of olfactory cortex and the second with the observed response characteristics of place cells in hippocampus. We investigate both kinds of networks analytically and by computer simulations, and we link the models with experimental findings, exploring, in particular, how the spike timing dependence of the synaptic plasticity constrains the computational function of the network and vice versa.     

Self-organizing continuous attractor networks and path integration: two-dimensional models of place cells

Single-neuron recording studies have demonstrated the existence of neurons in the hippocampus which appear to encode information about the place where a rat is located, and about the place at which a macaque is looking. We describe 'continuous attractor' neural network models of place cells with Gaussian spatial fields in which the recurrent collateral synaptic connections between the neurons reflect the distance between two places. The networks maintain a localized packet of neuronal activity that represents the place where the animal is located. We show for two related models how the representation of the two-dimensional space in the continuous attractor network of place cells could self-organize by modifying the synaptic connections between the neurons, and also how the place being represented can be updated by idiothetic (self-motion) signals in a neural implementation of path integration.     

Theta-modulated feedforward network generates rate and phase coded firing in the entorhino-hippocampal system

Principal cells of the hippocampus and of its only cortical input region, the entorhinal cortex exhibit place specific activity in the freely moving rat. While entorhinal cells have widely tuned place fields, hippocampal place fields are more localized and determine not only the rate but also the timing of place cell spikes. Several models have successfully attempted to explain this fine tuning making use of intrahippocampal attractor network dynamics provided by the recurrent collaterals of hippocampal area CA3. Recent experimental evidence shows that CA1 place cells preserve their tuning curves even in the absence of input from CA3. We propose a model in which entorhinal and hippocampal pyramidal cell populations are only connected via feedforward connections. Synaptic transmission in our system is gated by a class of interneurons inhibiting specifically the entorhino-hippocampal pathway. Theta rhythm modulates the activity of each component. Our results show that rhythmic shunting inhibition endows entorhinal cells with a novel type of temporal code conveyed by the phase jitter of individual spikes. This converts coarsely tuned place-specific activity in the entorhinal cortex to velocity-dependent postsynaptic excitation and, thus, provides hippocampal place cells with an input that has recently been proposed to account for their rate and phase coded firing. Hippocampal place fields are generated through this mechanism and also shown to be robust against variations in the level of tonic inhibition.     

A mobile robot mapping model inspired from the place cells functionality of hippocampus based on dimension reduction technique

ABSTRACT

In this paper, a new mobile robot mapping algorithm inspired from the functionality of hippocampus cells is presented. Place cells in hippocampus can store a map of the environment. This model fuses odometry and vision data based on dimensionality reduction technique, hierarchically. These two types of data are first fused and then considered as inputs to the place cell model. Place cells do the clustering of places. The proposed Place cell model has two types of inputs: Grid cells input and input from the lateral entorhinal cortex (LEC). The LEC is modelled based on the dimension reduction technique. Therefore, the data that causes locations different to be inserted into the place cell from this layer. Another contribution is proposing a new unsupervised dimension reduction method based on k-means. The method can find perpendicular independent dimensions. Also, the distance of cluster centres found in these dimensions is maximised. The method was compared with LDA and PCA in standard functions. Although LDA is a supervised method, the result showed that the proposed unsupervised method outperformed. To evaluate the place cells model, sequences of images collected by a mobile robot was used and similar results to real place cells achieved.     

成年海马神经再生改善模式分离鲁棒性

成年海马神经再生(adult hippocampal neurogenesis, AHN)被认为能有效参与齿状回(dentate gyrus, DG)网络来加强模式分离功能.目前虽然神经再生在模式分离中的潜在作用已在理论上得到了研究,但产生的新生颗粒细胞在信息处理和网络调节中的具体作用仍存在争议.针对上述问题,本文引入4-6周新生颗粒细胞作为独立的信息处理单元,提出了一种具有神经再生的DG网络计算模型.重点研究了不同输入刺激下新生颗粒细胞对模式分离的贡献.通过模拟实验,本文发现在不同强度的刺激下,新生颗粒细胞在齿状回网络中扮演着不同的角色.在低强度刺激下,新生颗粒细胞利用其易激活的神经元特性,可以恢复网络的信息表达能力,避免模式分离失败.在高强度刺激下,新生颗粒细胞作为一种中间神经元,能有效增强局部回路的反馈抑制作用,提高成熟颗粒细胞的稀疏性,最终增强模式分离功能.因此,该模型预测了在更精细和更广泛的输入下,成年海马神经再生在模式分离鲁棒性中的关键作用.     

Replicating receptive fields of simple and complex cells in primary visual cortex in a neuronal network model with temporal and population sparseness and reliability

We propose a new principle for replicating receptive field properties of neurons in the primary visual cortex. We derive a learning rule for a feedforward network, which maintains a low firing rate for the output neurons (resulting in temporal sparseness) and allows only a small subset of the neurons in the network to fire at any given time (resulting in population sparseness). Our learning rule also sets the firing rates of the output neurons at each time step to near-maximum or near-minimum levels, resulting in neuronal reliability. The learning rule is simple enough to be written in spatially and temporally local forms. After the learning stage is performed using input image patches of natural scenes, output neurons in the model network are found to exhibit simple-cell-like receptive field properties. When the output of these simple-cell-like neurons are input to another model layer using the same learning rule, the second-layer output neurons after learning become less sensitive to the phase of gratings than the simple-cell-like input neurons. In particular, some of the second-layer output neurons become completely phase invariant, owing to the convergence of the connections from first-layer neurons with similar orientation selectivity to second-layer neurons in the model network. We examine the parameter dependencies of the receptive field properties of the model neurons after learning and discuss their biological implications. We also show that the localized learning rule is consistent with experimental results concerning neuronal plasticity and can replicate the receptive fields of simple and complex cells.     

Incremental supervised learning for mobile robot reactive control

Reactive control for a mobile robot can be defined as a mapping from a perceptual space to a command space. This mapping can be hard-coded by the user (potential fields, fuzzy logic), and can also be learnt. This paper is concerned with supervised learning for perception to action mapping for a mobile robot. Among the existing neural approaches for supervised learning of a function, we have selected the grow and learn network for its properties adapted to robotic problems: incrementality and flexible structure. We will present the results we have obtained with this network using first raw sensor data and then pre-processed measures with the automatic construction of virtual sensors.     

Low-dimensional maps encoding dynamics in entorhinal cortex and hippocampus

Cells that produce intrinsic theta oscillations often contain the hyperpolarization-activated current I(h). In this article, we use models and dynamic clamp experiments to investigate the synchronization properties of two such cells (stellate cells of the entorhinal cortex and O-LM cells of the hippocampus) in networks with fast-spiking (FS) interneurons. The model we use for stellate cells and O-LM cells is the same, but the stellate cells are excitatory and the O-LM cells are inhibitory, with inhibitory postsynaptic potential considerably longer than those from FS interneurons. We use spike time response curve methods (STRC), expanding that technique to three-cell networks and giving two different ways in which the analysis of the three-cell network reduces to that of a two-cell network. We show that adding FS cells to a network of stellate cells can desynchronize the stellate cells, while adding them to a network of O-LM cells can synchronize the O-LM cells. These synchronization and desynchronization properties critically depend on I(h). The analysis of the deterministic system allows us to understand some effects of noise on the phase relationships in the stellate networks. The dynamic clamp experiments use biophysical stellate cells and in silico FS cells, with connections that mimic excitation or inhibition, the latter with decay times associated with FS cells or O-LM cells. The results obtained in the dynamic clamp experiments are in a good agreement with the analytical framework.     

Asynchronous inputs and NMDA conductances predict excitatory responses in the cortical-cA1 pathway of the hippocampus

In the hippocampus, CA1 place cells are driven by a substantial input from CA3. There is a second pathway to CA1 from the entorhinal cortex. The mode of action of cortex on CA1 through this pathway is not known. The pathway supports CA1 place field activity after CA3 has been lesioned, yet stimulation of the pathway in rat slices results in strong feedforward inhibition that prevents pyramidal cell action potentials. We use a detailed conductance-based model of this pathway to simulate the response to cortical stimulation in slice experiments and in vivo spatial exploration. We find that the presence of NMDA conductances enable CA1 pyramidal cells to integrate cortical inputs over a time scale longer than that which is effective in recruiting the inhibitory response that can suppress action potentials. We then show that this asynchronous response mode supports place field formation in response to experimentally constrained spatially modulated cortical activity. Within this model, the inclusion of GABAB conductances and the hyperpolarisation activated current I(h) reduces the strength of the GABAA inputs required to balance the excitatory inputs, and this facilitates place field formation by reducing variability in the inhibitory inputs.     

一种基于海马认知机理的仿生机器人认知地图构建方法

海马结构空间细胞的放电活动被认为能够形成对环境内在地图的表达,即所谓的认知地图.先前的仿生环境认知地图构建方法（例如RatSLAM）以及传统的SLAM方法均缺乏足够的生理学依据,不能准确地体现出生物在导航中的生理学现象和认知功能实现过程.本文模仿海马结构空间细胞的认知机理提出了一种构建精确的环境认知地图的方法,其特点在于通过构建统一的空间细胞吸引子计算模型对自运动线索进行路径积分;网格细胞和位置细胞对环境的表达来源于条纹细胞的前向驱动作用;通过环境的颜色深度图像进行闭环检测,对空间细胞路径积分进行误差修正,最终生成精确的环境认知地图.该认知地图是一种拓扑度量地图,包含了环境特征点坐标、视觉线索以及特定位点的拓扑关系.本文通过仿真实验和机器人平台物理实验验证了方法的有效性,研究成果为仿海马认知机理的机器人导航方法研究奠定了基础.     

Learning optimized features for hierarchical models of invariant object recognition

There is an ongoing debate over the capabilities of hierarchical neural feedforward architectures for performing real-world invariant object recognition. Although a variety of hierarchical models exists, appropriate supervised and unsupervised learning methods are still an issue of intense research. We propose a feedforward model for recognition that shares components like weight sharing, pooling stages, and competitive nonlinearities with earlier approaches but focuses on new methods for learning optimal feature-detecting cells in intermediate stages of the hierarchical network. We show that principles of sparse coding, which were previously mostly applied to the initial feature detection stages, can also be employed to obtain optimized intermediate complex features. We suggest a new approach to optimize the learning of sparse features under the constraints of a weight-sharing or convolutional architecture that uses pooling operations to achieve gradual invariance in the feature hierarchy. The approach explicitly enforces symmetry constraints like translation invariance on the feature set. This leads to a dimension reduction in the search space of optimal features and allows determining more efficiently the basis representatives, which achieve a sparse decomposition of the input. We analyze the quality of the learned feature representation by investigating the recognition performance of the resulting hierarchical network on object and face databases. We show that a hierarchy with features learned on a single object data set can also be applied to face recognition without parameter changes and is competitive with other recent machine learning recognition approaches. To investigate the effect of the interplay between sparse coding and processing nonlinearities, we also consider alternative feedforward pooling nonlinearities such as presynaptic maximum selection and sum-of-squares integration. The comparison shows that a combination of strong competitive nonlinearities with sparse coding offers the best recognition performance in the difficult scenario of segmentation-free recognition in cluttered surround. We demonstrate that for both learning and recognition, a precise segmentation of the objects is not necessary.     

Learning sensory maps with real-world stimuli in real time using abiophysically realistic learning rule

We present a real-time model of learning in the auditory cortex that is trained using real-world stimuli. The system consists of a peripheral and a central cortical network of spiking neurons. The synapses formed by peripheral neurons on the central ones are subject to synaptic plasticity. We implemented a biophysically realistic learning rule that depends on the precise temporal relation of pre- and postsynaptic action potentials. We demonstrate that this biologically realistic real-time neuronal system forms stable receptive fields that accurately reflect the spectral content of the input signals and that the size of these representations can be biased by global signals acting on the local learning mechanism. In addition, we show that this learning mechanism shows fast acquisition and is robust in the presence of large imbalances in the probability of occurrence of individual stimuli and noise.     

Adding learning to cellular genetic algorithms for trainingrecurrent neural networks

This paper proposes a hybrid optimization algorithm which combines the efforts of local search (individual learning) and cellular genetic algorithms (GA) for training recurrent neural nets (RNN). Each RNN weight is encoded as a floating point number, and a concatenation of numbers forms a chromosome. Reproduction takes place locally in a square grid, each grid point representing a chromosome. Lamarckian and Baldwinian (1896) mechanisms for combining cellular GA and learning are compared. Different hill-climbing algorithms are incorporated into the cellular GA. These include the real-time recurrent learning (RTRL) and its simplified versions, and the delta rule. RTRL has been successively simplified by freezing some of the weights to form simplified versions. The delta rule, the simplest form of learning, has been implemented by considering the RNN as feedforward networks. The hybrid algorithms are used to train the RNN to solve a long-term dependency problem. The results show that Baldwinian learning is inefficient in assisting the cellular GA. It is conjectured that the more difficult it is for genetic operations to produce the genotypic changes that match the phenotypic changes due to learning, the poorer is the convergence of Baldwinian learning. Most of the combinations using the Lamarckian mechanism show an improvement in reducing the number of generations for an optimum network; however, only a few can reduce the actual time taken. Embedding the delta rule in the cellular GA is the fastest method. Learning should not be too extensive.     

使用模糊竞争Hopfield网络进行图像分割

针对传统自组织竞争学习方法的不足,将模糊竞争学习引入竞争Hopfield网络中,由此设计了一个用于图像分割的模糊竞争Hopfield网络,通过将图像空间映射到灰度特征空间,实现灰度特征集的模糊聚类,进而实现图像分割.实验结果表明:对于二值分割,与Ostu方法相比,此算法在分割效果和对噪声的自适应能力方面具有明显的优点.对于多类分割,此算法比目前的FCM(fuzzy C mean)算法的处理速度要快.     

基于循环神经网络的正交网格的自动化生成算法

随着计算机图形学、工业设计、自然科学等领域的飞速发展,对高质量的科学计算方法的需求随之增大,而这些科学计算的方法离不开高质量的网格生成算法。对于常用的正交网格生成算法,是否能减少计算量以及是否能降低的人工干预等问题仍是它们所面临的主要挑战。针对这些挑战,对于单连通的目标区域,提出了基于循环神经网络之一的长短期记忆网络（LSTM）和Schwarz-Christoffel共形映射（SC映射）的正交网格自动化生成算法。首先,利用基于SC映射的Gridgen-c工具的基本条件将网格生成问题转换为一个带线性限制条件的整数规划问题。接着,利用预处理后的GADM数据集和LSTM训练获得能计算目标多边形区域每个顶点转角类型的概率的分类器。该分类器可以大幅度降低整数规划问题的时间复杂度,使该问题能被自动化且快速地求解。最后以简单图形区域、动画图形区域、地理边界区域为样例,进行网格生成实验。结果表明：对于简单图形区域,所提算法均能达到最优解;而对于具有复杂边界的动画图形区域和地理边界区域,实例网格结果表明,所提算法能使这些目标区域的计算量分别降低88.42%和91.16%,且能自动化地生成较好的正交网格。     

语义网格中基于多策略学习的Ontology匹配方法

近些年来,语义Web和网格计算这两个方向在各自的研究社区分别发展着,这两方面的交叉即语义网格(semantic grid)则是最近一段时间兴起的研究领域．通过给网格附加语义层,能够促进网格自组织的形成．现有的Gnd社区都是使用集中式的、一致性的、可扩充的Ontology库．超越集中式的语义存储是语义网格发展面临的最大挑战之一．针对网格社区间的Ontology异构性这个问题,提出了一种多策略的Ontology匹配学习方法．它使用多种分类方法来学习Ontology之间的匹配：使用一般的基于统计的分类方法来发现数据实例内部的分类特征;或者使用基于一阶逻辑的学习算法FOIL来发现数据实例之间的语义联系．在单个方法预测的基础上,匹配系统使用称之为最突出的冠军的匹配委员会方法来集成分类结果．实验表明在现实的知识领域中,系统能达到很高的匹配精度．