Optimizing time histograms for non-Poissonian spike trains

The time histogram is a fundamental tool for representing the inhomogeneous density of event occurrences such as neuronal firings. The shape of a histogram critically depends on the size of the bins that partition the time axis. In most neurophysiological studies, however, researchers have arbitrarily selected the bin size when analyzing fluctuations in neuronal activity. A rigorous method for selecting the appropriate bin size was recently derived so that the mean integrated squared error between the time histogram and the unknown underlying rate is minimized (Shimazaki & Shinomoto, 2007 ). This derivation assumes that spikes are independently drawn from a given rate. However, in practice, biological neurons express non-Poissonian features in their firing patterns, such that the spike occurrence depends on the preceding spikes, which inevitably deteriorate the optimization. In this letter, we revise the method for selecting the bin size by considering the possible non-Poissonian features. Improvement in the goodness of fit of the time histogram is assessed and confirmed by numerically simulated non-Poissonian spike trains derived from the given fluctuating rate. For some experimental data, the revised algorithm transforms the shape of the time histogram from the Poissonian optimization method.     

Parameters of spike trains observed in a short time window

We study the estimation of statistical moments of interspike intervals based on observation of spike counts in many independent short time windows. This scenario corresponds to the situation in which a target neuron occurs. It receives information from many neurons and has to respond within a short time interval. The precision of the estimation procedures is examined. As the model for neuronal activity, two examples of stationary point processes are considered: renewal process and doubly stochastic Poisson process. Both moment and maximum likelihood estimators are investigated. Not only the mean but also the coefficient of variation is estimated. In accordance with our expectations, numerical studies confirm that the estimation of mean interspike interval is more reliable than the estimation of coefficient of variation. The error of estimation increases with increasing mean interspike interval, which is equivalent to decreasing the size of window (less events are observed in a window) and with decreasing the number of neurons (lower number of windows).     

Detection of syntonies between multiple spike trains using a coarse-grain binarization of spike count distributions

At very short timescales neuronal spike trains may be compared to binary streams where each neuron gives at most one spike per bin and therefore its state can be described by a binary variable. Time-averaged activity like the mean firing rate can be generally used on longer timescales to describe the dynamics; nevertheless, enlarging the space of the possible states up to the continuum may seriously bias the true statistics if the sampling is not accurate. We propose a simple transformation on binary variables which allows us to fix the dimensionality of the space to sample and to vary the temporal resolution of the analysis. For each time length interactions among simultaneously recorded neurons are evaluated using log-linear models. We illustrate how to use this method by analysing two different sets of data, recorded respectively in the temporal cortex of freely moving rats and in the inferotemporal cortex of behaving monkeys engaged in a visual fixation task. A detailed study of the interactions is provided for both samples. In both datasets we find that some assemblies share robust interactions, invariant at different time lengths, while others cooperate only at delimited time resolutions, yet the size of the samples is too small to allow an unbiased estimate of all possible interactions. We conclude that an extensive application of our method to larger samples of data, together with the development of techniques to correct the bias in the estimate of the coefficients, would provide significant information about the structure of the interactions in populations of neurons.     

Fuzzy characterization of spike synchrony in parallel spike trains

We present a framework for characterizing spike (and spike-train) synchrony in parallel neuronal spike trains that is based on the identification of spikes with what we call influence maps: real-valued functions that describe an influence region around the corresponding spike times within which possibly graded (i.e., fuzzy) synchrony with other spikes is defined. We formalize two models of synchrony in this framework: the bin-based model (the almost exclusively applied model in the field) and a novel, alternative model based on a continuous, graded notion of synchrony, aimed at overcoming the drawbacks of the bin-based model. We study the task of identifying frequent (and synchronous) neuronal patterns from parallel spike trains in our framework, formalized as an instance of what we call the fuzzy frequent pattern mining problem (a generalization of standard frequent pattern mining) and briefly evaluate our synchrony models on this task.     

Spatial variability of SEBAL estimated root-zone soil moisture across scales

This study investigated the spatial scaling behaviour of root-zone soil moisture obtained from optical/thermal remote-sensing observations. The data for this study were obtained from Landsat and Moderate Resolution Imaging Spectroradiometer (MODIS) satellites on five different dates between early spring (April) and fall (September) in the years from 2000 to 2004 in the semi-arid middle Rio Grande Valley of New Mexico. Soil moisture data were obtained using the Surface Energy Balance Algorithm for Land (SEBAL) algorithm. The data were spatially aggregated and checked for power-law behaviour over a range of scales from 30 m to 15 km for Landsat and from 1 to 28 km for MODIS images. Results of this study demonstrate that power-law scaling of soil moisture in the middle Rio Grande area holds up to 1 km² pixel size, but is no longer valid beyond that scale. Whereas previous studies have studied soil moisture in the top 5 cm of the soil using radar and point measurements, our study uses SEBAL to estimate root-zone soil moisture. Our study is consistent with these previous studies in showing that variation in root-zone soil follows an empirical power law for pixel sizes of up to about 10⁶ m² and that there is an apparent break in the scaling at larger scales.     

多时间尺度时间序列趋势预测

针对股票、基金等大量时间序列数据的趋势预测问题，提出一种基于新颖特征模型的多时间尺度时间序列趋势预测算法。首先，在原始时间序列中提取带有多时间尺度特征的特征树，其刻画了时间序列，不仅带有序列在各个层次的特征，同时表示了层次之间的关系。然后，利用聚类挖掘特征序列中的隐含状态。最后，应用隐马尔可夫模型（HMM）设计一个多时间尺度趋势预测算法（MTSTPA），同时对不同尺度下的趋势以及趋势的长度作出预测。在真实股票数据集上的实验中，在各个尺度上的预测准确率均在60%以上，与未使用特征树对比，使用特征树的模型预测效率更高，在某一尺度上准确率高出10个百分点以上。同时，与经典自回归滑动平均模型（ARMA）模型和PHMM（Pattern-based HMM）对比，MTSTPA表现更优，验证了其有效性。     

Detection of hidden structures in nonstationary spike trains

We propose an algorithm for simultaneously estimating state transitions among neural states and nonstationary firing rates using a switching state-space model (SSSM). This algorithm enables us to detect state transitions on the basis of not only discontinuous changes in mean firing rates but also discontinuous changes in the temporal profiles of firing rates (e.g., temporal correlation). We construct estimation and learning algorithms for a nongaussian SSSM, whose nongaussian property is caused by binary spike events. Local variational methods can transform the binary observation process into a quadratic form. The transformed observation process enables us to construct a variational Bayes algorithm that can determine the number of neural states based on automatic relevance determination. Additionally, our algorithm can estimate model parameters from single-trial data using a priori knowledge about state transitions and firing rates. Synthetic data analysis reveals that our algorithm has higher performance for estimating nonstationary firing rates than previous methods. The analysis also confirms that our algorithm can detect state transitions on the basis of discontinuous changes in temporal correlation, which are transitions that previous hidden Markov models could not detect. We also analyze neural data recorded from the medial temporal area. The statistically detected neural states probably coincide with transient and sustained states that have been detected heuristically. Estimated parameters suggest that our algorithm detects the state transitions on the basis of discontinuous changes in the temporal correlation of firing rates. These results suggest that our algorithm is advantageous in real-data analysis.     

Finding the event structure of neuronal spike trains

Neurons in sensory systems convey information about physical stimuli in their spike trains. In vitro, single neurons respond precisely and reliably to the repeated injection of the same fluctuating current, producing regions of elevated firing rate, termed events. Analysis of these spike trains reveals that multiple distinct spike patterns can be identified as trial-to-trial correlations between spike times (Fellous, Tiesinga, Thomas, & Sejnowski, 2004 ). Finding events in data with realistic spiking statistics is challenging because events belonging to different spike patterns may overlap. We propose a method for finding spiking events that uses contextual information to disambiguate which pattern a trial belongs to. The procedure can be applied to spike trains of the same neuron across multiple trials to detect and separate responses obtained during different brain states. The procedure can also be applied to spike trains from multiple simultaneously recorded neurons in order to identify volleys of near-synchronous activity or to distinguish between excitatory and inhibitory neurons. The procedure was tested using artificial data as well as recordings in vitro in response to fluctuating current waveforms.     

Hierarchical aggregation of linear systems with multiple time scales

In this paper we carry out a detailed analysis of the multiple time scale behavior of singularly perturbed linear systems of the formdot{x}^{epsilon}(t) = A(epsilon)x^{epsilon}(t)whereA(epsilon)is analytic in the small parameter ε. Our basic result is a uniform asymptotic approximation toexp A(epsilon)tthat we obtain under a certain multiple semistability condition. This asymptotic approximation gives a complete multiple time scale decomposition of the above system and specifies a set of reduced order models valid at each time scale. Our contribution is threefold. 1) We do not require that the state variables be chosen so as to display the time scale structure of the system. 2) Our formulation can handle systems with multiple ( > 2) time scales and we obtain uniform asymptotic expansions for their behavior on [0, infty]. 3) We give an aggregation method to produce increasingly simplified models valid at progressively slower time scales.     

A decomposition approach to modelling high service time variability

The predictive ability of queueing network models can be greatly enhanced if these models include the effects of system characteristics such as high service time variability and simultaneous resource possession, which violate the assumptions required for their efficient exact solution. In this paper we present a new approximate solution technique for queueing networks that include Coxian servers to represent resources at which customers have high service time variability. Our approach is unique in several respects: it is based directly on the theory of near-complete decomposability, it is non-iterative (performance measures for the queueing network of interest are expressed as linear combinations of the performance measures of a set of separable queueing networks), and it is conceptually and computationally simple.     

Improved similarity measures for small sets of spike trains

Multiple measures have been developed to quantify the similarity between two spike trains. These measures have been used for the quantification of the mismatch between neuron models and experiments as well as for the classification of neuronal responses in neuroprosthetic devices and electrophysiological experiments. Frequently only a few spike trains are available in each class. We derive analytical expressions for the small-sample bias present when comparing estimators of the time-dependent firing intensity. We then exploit analogies between the comparison of firing intensities and previously used spike train metrics and show that improved spike train measures can be successfully used for fitting neuron models to experimental data, for comparisons of spike trains, and classification of spike train data. In classification tasks, the improved similarity measures can increase the recovered information. We demonstrate that when similarity measures are used for fitting mathematical models, all previous methods systematically underestimate the noise. Finally, we show a striking implication of this deterministic bias by reevaluating the results of the single-neuron prediction challenge.     

Bayesian estimation of stimulus responses in Poisson spike trains

A Bayesian method is developed for estimating neural responses to stimuli, using likelihood functions incorporating the assumption that spike trains follow either pure Poisson statistics or Poisson statistics with a refractory period. The Bayesian and standard estimates of the mean and variance of responses are similar and asymptotically converge as the size of the data sample increases. However, the Bayesian estimate of the variance of the variance is much lower. This allows the Bayesian method to provide more precise interval estimates of responses. Sensitivity of the Bayesian method to the Poisson assumption was tested by conducting simulations perturbing the Poisson spike trains with noise. This did not affect Bayesian estimates of mean and variance to a significant degree, indicating that the Bayesian method is robust. The Bayesian estimates were less affected by the presence of noise than estimates provided by the standard method.     

Generation of synthetic spike trains with defined pairwise correlations

Recent technological advances as well as progress in theoretical understanding of neural systems have created a need for synthetic spike trains with controlled mean rate and pairwise cross-correlation. This report introduces and analyzes a novel algorithm for the generation of discretized spike trains with arbitrary mean rates and controlled cross correlation. Pairs of spike trains with any pairwise correlation can be generated, and higher-order correlations are compatible with common synaptic input. Relations between allowable mean rates and correlations within a population are discussed. The algorithm is highly efficient, its complexity increasing linearly with the number of spike trains generated and therefore inversely with the number of cross-correlated pairs.     

Frequency domain results for systems with multiple time scales

The mathematical treatment of systems with multiple time scales has traditionally employed the use of singular perturbation theory for differential equations. This paper treats such systems from a frequency-domain viewpoint. Previously existing two-frequency-scale results, involving approximation and algebraic characterization, are extended. In addition, an exact frequency-scale decomposition is presented.     

Computing optical flow across multiple scales: An adaptive coarse-to-fine strategy

Single-scale approaches to the determination of the optical flow field from the time-varying brightness pattern assume that spatio-temporal discretization is adequate for representing the patterns and motions in a scene. However, the choice of an appropriate spatial resolution is subject to conflicting, scene-dependent, constraints. In intensity-base methods for recovering optical flow, derivative estimation is more accurate for long wavelengths and slow velocities (with respect to the spatial and temporal discretization steps). On the contrary, short wavelengths and fast motions are required in order to reduce the errors caused by noise in the image acquisition and quantization process.Estimating motion across different spatial scales should ameliorate this problem. However, homogeneous multiscale approaches, such as the standard multigrid algorithm, do not improve this situation, because an optimal velocity estimate at a given spatial scale is likely to be corrupted at a finer scale. We propose an adaptive multiscale method, where the discretization scale is chosen locally according to an estimate of the relative error in the velocity estimation, based on image properties.Results for synthetic and video-acquired images show that our coarse-to-fine method, fully parallel at each scale, provides substantially better estimates of optical flow than do conventional algorithms, while adding little computational cost.     

Multiparameter singular perturbations of linear systems with multiple time scales

A joint multitime scale-multiparameter singular perturbation is formulated and resolved in the context of linear time-varying systems. An interesting feature of the solution procedure is a hierarchial scheme of aggregating and arranging of the groups of small parameters according to their order. The scheme provides a suitable framework for establishing qualitative properties of multitime scale systems.     

融合多时间尺度和特征加强的知识图谱推荐模型

针对现有的知识图谱推荐模型没有考虑到用户的周期特征以及待推荐项目会对用户近期兴趣产生影响的问题,提出一种融合多时间尺度和特征加强的知识图谱推荐模型（MTFE）。首先,采用长短期记忆（LSTM）网络在不同时间尺度上挖掘用户的周期特征并融入到用户表示中;然后,通过注意力机制挖掘待推荐项目中与用户近期特征相关性较强的特征,将其加强后融入项目表示中;最后,通过评分函数计算用户对待推荐项目的评分。在真实数据集Last.FM、MovieLens-1M和MovieLens-20M上把所提模型和个性化实体推荐（PER）、协同知识嵌入（CKE）、LibFM、RippleNet、知识图卷积网络（KGCN）、协同知识感知注意网络（CKAN）等知识图谱推荐模型进行对比。实验结果表明,在三个数据集上MTFE相较于表现最优的对比模型的F1性能分别提升了0.78、1.63和1.92个百分点,AUC指标在三个数据集上分别提升了3.94、2.73和1.15个百分点。可见,所提模型相较于对比图谱推荐模型有更好的推荐效果。     

Estimating the entropy rate of spike trains via Lempel-Ziv complexity

Normalized Lempel-Ziv complexity, which measures the generation rate of new patterns along a digital sequence, is closely related to such important source properties as entropy and compression ratio, but, in contrast to these, it is a property of individual sequences. In this article, we propose to exploit this concept to estimate (or, at least, to bound from below) the entropy of neural discharges (spike trains). The main advantages of this method include fast convergence of the estimator (as supported by numerical simulation) and the fact that there is no need to know the probability law of the process generating the signal. Furthermore, we present numerical and experimental comparisons of the new method against the standard method based on word frequencies, providing evidence that this new approach is an alternative entropy estimator for binned spike trains.     

Testing for and estimating latency effects for poisson and non-poisson spike trains

Determining the variations in response latency of one or several neurons to a stimulus is of interest in different contexts. Two common problems concern correlating latency with a particular behavior, for example, the reaction time to a stimulus, and adjusting tools for detecting synchronization between two neurons. We use two such problems to illustrate the latency testing and estimation methods developed in this article. Our test for latencies is a formal statistical test that produces a p-value. It is applicable for Poisson and non-Poisson spike trains via use of the bootstrap. Our estimation method is model free, it is fast and easy to implement, and its performance compares favorably to other methods currently available.