Computational Neuroscience Models Research Articles

Mirrored STDP Implements Autoencoder Learning in a Network of Spiking Neurons.

The autoencoder algorithm is a simple but powerful unsupervised method for training neural networks. Autoencoder networks can learn sparse distributed codes similar to those seen in cortical sensory areas such as visual area V1, but they can also be stacked to learn increasingly abstract representations. Several computational neuroscience models of sensory areas, including Olshausen & Field’s Sparse Coding algorithm, can be seen as autoencoder variants, and autoencoders have seen extensive use in the machine learning community. Despite their power and versatility, autoencoders have been difficult to implement in a biologically realistic fashion. The challenges include their need to calculate differences between two neuronal activities and their requirement for learning rules which lead to identical changes at feedforward and feedback connections. Here, we study a biologically realistic network of integrate-and-fire neurons with anatomical connectivity and synaptic plasticity that closely matches that observed in cortical sensory areas. Our choice of synaptic plasticity rules is inspired by recent experimental and theoretical results suggesting that learning at feedback connections may have a different form from learning at feedforward connections, and our results depend critically on this novel choice of plasticity rules. Specifically, we propose that plasticity rules at feedforward versus feedback connections are temporally opposed versions of spike-timing dependent plasticity (STDP), leading to a symmetric combined rule we call Mirrored STDP (mSTDP). We show that with mSTDP, our network follows a learning rule that approximately minimizes an autoencoder loss function. When trained with whitened natural image patches, the learned synaptic weights resemble the receptive fields seen in V1. Our results use realistic synaptic plasticity rules to show that the powerful autoencoder learning algorithm could be within the reach of real biological networks.

Qualitative properties of solutions for the noisy integrate and fire model in computational neuroscience

The Noisy Integrate-and-Fire equation is a standard non-linear Fokker–Planck equation used to describe the activity of a homogeneous neural network characterized by its connectivity b (each neuron connected to all others through synaptic weights); b > 0 describes excitatory networks and b < 0 inhibitory networks. In the excitatory case, it was proved that, once the proportion of neurons that are close to their action potential is too high, solutions cannot exist for all times. In this paper, we show a priori uniform bounds in time on the firing rate to discard the scenario of blow-up, and, for small connectivity, we prove qualitative properties on the long time behavior of solutions. The methods are based on the one hand on relative entropy and Poincaré inequalities leading to L2 estimates and on the other hand, on the notion of ‘universal super-solution’ and parabolic regularizing effects to obtain bounds.

A reconfigurable on-line learning spiking neuromorphic processor comprising 256 neurons and 128K synapses.

Implementing compact, low-power artificial neural processing systems with real-time on-line learning abilities is still an open challenge. In this paper we present a full-custom mixed-signal VLSI device with neuromorphic learning circuits that emulate the biophysics of real spiking neurons and dynamic synapses for exploring the properties of computational neuroscience models and for building brain-inspired computing systems. The proposed architecture allows the on-chip configuration of a wide range of network connectivities, including recurrent and deep networks, with short-term and long-term plasticity. The device comprises 128 K analog synapse and 256 neuron circuits with biologically plausible dynamics and bi-stable spike-based plasticity mechanisms that endow it with on-line learning abilities. In addition to the analog circuits, the device comprises also asynchronous digital logic circuits for setting different synapse and neuron properties as well as different network configurations. This prototype device, fabricated using a 180 nm 1P6M CMOS process, occupies an area of 51.4 mm2, and consumes approximately 4 mW for typical experiments, for example involving attractor networks. Here we describe the details of the overall architecture and of the individual circuits and present experimental results that showcase its potential. By supporting a wide range of cortical-like computational modules comprising plasticity mechanisms, this device will enable the realization of intelligent autonomous systems with on-line learning capabilities.

ModelView for ModelDB: Online Presentation of Model Structure

ModelDB ( modeldb.yale.edu ), a searchable repository of source code of more than 950 published computational neuroscience models, seeks to promote model reuse and reproducibility. Code sharing is a first step; however, model source code is often large and not easily understood. To aid users, we have developed ModelView, a web application for ModelDB that presents a graphical view of model structure augmented with contextual information for NEURON and NEURON-runnable (e.g. NeuroML, PyNN) models. Web presentation provides a rich, simulator-independent environment for interacting with graphs. The necessary data is generated by combining manual curation, text-mining the source code, querying ModelDB, and simulator introspection. Key features of the user interface along with the data analysis, storage, and visualization algorithms are explained. With this tool, researchers can examine and assess the structure of hundreds of models in ModelDB in a standardized presentation without installing any software, downloading the model, or reading model source code.

Creating the brain and interacting with the brain: an integrated approach to understanding the brain.

In the past two decades, brain science and robotics have made gigantic advances in their own fields, and their interactions have generated several interdisciplinary research fields. First, in the 'understanding the brain by creating the brain' approach, computational neuroscience models have been applied to many robotics problems. Second, such brain-motivated fields as cognitive robotics and developmental robotics have emerged as interdisciplinary areas among robotics, neuroscience and cognitive science with special emphasis on humanoid robots. Third, in brain-machine interface research, a brain and a robot are mutually connected within a closed loop. In this paper, we review the theoretical backgrounds of these three interdisciplinary fields and their recent progress. Then, we introduce recent efforts to reintegrate these research fields into a coherent perspective and propose a new direction that integrates brain science and robotics where the decoding of information from the brain, robot control based on the decoded information and multimodal feedback to the brain from the robot are carried out in real time and in a closed loop.

A neuromorphic system for video object recognition.

Automated video object recognition is a topic of emerging importance in both defense and civilian applications. This work describes an accurate and low-power neuromorphic architecture and system for real-time automated video object recognition. Our system, Neuormorphic Visual Understanding of Scenes (NEOVUS), is inspired by computational neuroscience models of feed-forward object detection and classification pipelines for processing visual data. The NEOVUS architecture is inspired by the ventral (what) and dorsal (where) streams of the mammalian visual pathway and integrates retinal processing, object detection based on form and motion modeling, and object classification based on convolutional neural networks. The object recognition performance and energy use of the NEOVUS was evaluated by the Defense Advanced Research Projects Agency (DARPA) under the Neovision2 program using three urban area video datasets collected from a mix of stationary and moving platforms. These datasets are challenging and include a large number of objects of different types in cluttered scenes, with varying illumination and occlusion conditions. In a systematic evaluation of five different teams by DARPA on these datasets, the NEOVUS demonstrated the best performance with high object recognition accuracy and the lowest energy consumption. Its energy use was three orders of magnitude lower than two independent state of the art baseline computer vision systems. The dynamic power requirement for the complete system mapped to commercial off-the-shelf (COTS) hardware that includes a 5.6 Megapixel color camera processed by object detection and classification algorithms at 30 frames per second was measured at 21.7 Watts (W), for an effective energy consumption of 5.45 nanoJoules (nJ) per bit of incoming video. These unprecedented results show that the NEOVUS has the potential to revolutionize automated video object recognition toward enabling practical low-power and mobile video processing applications.

A neuroscience inspired gated learning action selection mechanism

Abstract This paper presents an algorithm for action selection, in the context of intelligent agents, capable of learning from rewards which are sparse in time. Inspiration for the proposed algorithm was drawn from computational neuroscience models of how the human prefrontal cortex (PFC) works. We have observed that this abstraction provides some advantages, such as the representation of solutions as trees, making it human-readable, and turning the learning process into a combinatorial optimization problem. Results for it solving the 1-2-AX working memory task are presented and discussed. We also argue the pros and cons of the proposed algorithm and, finally, address potential future work.

Racing to learn: statistical inference and learning in a single spiking neuron with adaptive kernels.

This paper describes the Synapto-dendritic Kernel Adapting Neuron (SKAN), a simple spiking neuron model that performs statistical inference and unsupervised learning of spatiotemporal spike patterns. SKAN is the first proposed neuron model to investigate the effects of dynamic synapto-dendritic kernels and demonstrate their computational power even at the single neuron scale. The rule-set defining the neuron is simple: there are no complex mathematical operations such as normalization, exponentiation or even multiplication. The functionalities of SKAN emerge from the real-time interaction of simple additive and binary processes. Like a biological neuron, SKAN is robust to signal and parameter noise, and can utilize both in its operations. At the network scale neurons are locked in a race with each other with the fastest neuron to spike effectively “hiding” its learnt pattern from its neighbors. The robustness to noise, high speed, and simple building blocks not only make SKAN an interesting neuron model in computational neuroscience, but also make it ideal for implementation in digital and analog neuromorphic systems which is demonstrated through an implementation in a Field Programmable Gate Array (FPGA). Matlab, Python, and Verilog implementations of SKAN are available at: http://www.uws.edu.au/bioelectronics_neuroscience/bens/reproducible_research.

Towards a Neuromorphic Vestibular System

The vestibular system plays a crucial role in the sense of balance and spatial orientation in mammals. It is a sensory system that detects both rotational and translational motion of the head, via its semicircular canals and otoliths respectively. In this work, we propose a real-time hardware model of an artificial vestibular system, implemented using a custom neuromorphic Very Large Scale Integration (VLSI) multi-neuron chip interfaced to a commercial Inertial Measurement Unit (IMU). The artificial vestibular system is realized with spiking neurons that reproduce the responses of biological hair cells present in the real semicircular canals and otholitic organs. We demonstrate the real-time performance of the hybrid analog-digital system and characterize its response properties, presenting measurements of a successful encoding of angular velocities as well as linear accelerations. As an application, we realized a novel implementation of a recurrent integrator network capable of keeping track of the current angular position. The experimental results provided validate the hardware implementation via comparisons with a detailed computational neuroscience model. In addition to being an ideal tool for developing bio-inspired robotic technologies, this work provides a basis for developing a complete low-power neuromorphic vestibular system which integrates the hardware model of the neural signal processing pathway described with custom bio-mimetic gyroscopic sensors, exploiting neuromorphic principles in both mechanical and electronic aspects.

Accurate approximation and MPI parallelization of spatial stochastic reaction-diffusion in STEPS

Spatial stochastic reaction-diffusion simulations have become an important component of molecular modeling in Computational Neuroscience, as shown in a growing number of recent studies including our previous work in which we show that stochastic effects, in particular stochastic calcium dynamics, contribute to Purkinje cell calcium burst variability [1]. However, the computational cost of exact, serial algorithms such as Gillespie’s direct method as implemented in STEPS [2] in which every reaction event and cross-subvolume diffusion event is simulated, impose severe restrictions on model complexity. Serial solutions limit the number of molecular species and their reaction channels that can be represented, as well as imposing restrictive upper-limits on the number of subvolumes in the discretized space. It is desirable to provide a solution that allows investigation of stochastic effects in larger systems, with complex meshes of many hundreds of thousands or millions of subvolumes representing complex neuronal morphology, in projects such as the Human Brain project [3], which is not possible with serial solutions. Therefore, parallel solutions for stochastic reaction-diffusion is vital for the future of this field. There are common challenges of parallelization for reaction-diffusion simulators as well as some unique challenges for STEPS, in which space is discretized into irregular tetrahedrons and not cubes that are commonly used by other reaction-diffusion simulators. Exact solutions are not beneficial, where stochastic diffusion across geometry partitions result in frequent conflict between nodes, requiring regular costly rollbacks. Therefore it is essential that an approximation algorithm is applied. Such an approximation must minimise loss of accuracy whilst maximising performance gain. Existing solutions such as the Gillespie-Multi-Particle method (GMP) [4,5] have been designed for cubic subvolumes, which have the advantage of regularity (yet restrict morphological accuracy [2]). We review such approaches from a theoretical standpoint and find that a systematic diffusion slowing error exists with naive application to irregular subvolumes, which have local scaled diffusion coefficients and therefore localized expected diffusion times. We present a method that is tailored for irregular subvolumes, with a defined theoretical upper limit to the communication time-step. The method could, however, be generalised for different subvolume configurations where adaptable step-sizes could be beneficial. We further expand our method to allow multiple particle leaps, which for some systems this can greatly increase the minimum acceptable step-size and thus benefit performance whilst still ensuring acceptable accuracy. We demonstrate accuracy of our approximation by comparing to analytical solutions on a number of test cases. Our method is implemented on an MPI framework, and we present performance gains under two scenariosglobal application of the method, and application only at partition boundaries. We discuss the potential of this method for future large-scale parallel simulations in STEPS.

Spike-Based Synaptic Plasticity in Silicon: Design, Implementation, Application, and Challenges

The ability to carry out signal processing, classification, recognition, and computation in artificial spiking neural networks (SNNs) is mediated by their synapses. In particular, through activity-dependent alteration of their efficacies, synapses play a fundamental role in learning. The mathematical prescriptions under which synapses modify their weights are termed synaptic plasticity rules. These learning rules can be based on abstract computational neuroscience models or on detailed biophysical ones. As these rules are being proposed and developed by experimental and computational neuroscientists, engineers strive to design and implement them in silicon and en masse in order to employ them in complex real-world applications. In this paper, we describe analog very large-scale integration (VLSI) circuit implementations of multiple synaptic plasticity rules, ranging from phenomenological ones (e.g., based on spike timing, mean firing rates, or both) to biophysically realistic ones (e.g., calcium-dependent models). We discuss the application domains, weaknesses, and strengths of various representative approaches proposed in the literature, and provide insight into the challenges that engineers face when designing and implementing synaptic plasticity rules in VLSI technology for utilizing them in real-world applications.

A neuron-inspired computational architecture for spatiotemporal visual processing

In this article, we present a neurologically motivated computational architecture for visual information processing. The computational architecture's focus lies in multiple strategies: hierarchical processing, parallel and concurrent processing, and modularity. The architecture is modular and expandable in both hardware and software, so that it can also cope with multisensory integrations - making it an ideal tool for validating and applying computational neuroscience models in real time under real-world conditions. We apply our architecture in real time to validate a long-standing biologically inspired visual object recognition model, HMAX. In this context, the overall aim is to supply a humanoid robot with the ability to perceive and understand its environment with a focus on the active aspect of real-time spatiotemporal visual processing. We show that our approach is capable of simulating information processing in the visual cortex in real time and that our entropy-adaptive modification of HMAX has a higher efficiency and classification performance than the standard model (up to ∼+6%).

N2A: a computational tool for modeling from neurons to algorithms

The exponential increase in available neural data has combined with the exponential growth in computing (“Moore's law”) to create new opportunities to understand neural systems at large scale and high detail. The ability to produce large and sophisticated simulations has introduced unique challenges to neuroscientists. Computational models in neuroscience are increasingly broad efforts, often involving the collaboration of experts in different domains. Furthermore, the size and detail of models have grown to levels for which understanding the implications of variability and assumptions is no longer trivial. Here, we introduce the model design platform N2A which aims to facilitate the design and validation of biologically realistic models. N2A uses a hierarchical representation of neural information to enable the integration of models from different users. N2A streamlines computational validation of a model by natively implementing standard tools in sensitivity analysis and uncertainty quantification. The part-relationship representation allows both network-level analysis and dynamical simulations. We will demonstrate how N2A can be used in a range of examples, including a simple Hodgkin-Huxley cable model, basic parameter sensitivity of an 80/20 network, and the expression of the structural plasticity of a growing dendrite and stem cell proliferation and differentiation.

On the Relationships between Generative Encodings, Regularity, and Learning Abilities when Evolving Plastic Artificial Neural Networks

A major goal of bio-inspired artificial intelligence is to design artificial neural networks with abilities that resemble those of animal nervous systems. It is commonly believed that two keys for evolving nature-like artificial neural networks are (1) the developmental process that links genes to nervous systems, which enables the evolution of large, regular neural networks, and (2) synaptic plasticity, which allows neural networks to change during their lifetime. So far, these two topics have been mainly studied separately. The present paper shows that they are actually deeply connected. Using a simple operant conditioning task and a classic evolutionary algorithm, we compare three ways to encode plastic neural networks: a direct encoding, a developmental encoding inspired by computational neuroscience models, and a developmental encoding inspired by morphogen gradients (similar to HyperNEAT). Our results suggest that using a developmental encoding could improve the learning abilities of evolved, plastic neural networks. Complementary experiments reveal that this result is likely the consequence of the bias of developmental encodings towards regular structures: (1) in our experimental setup, encodings that tend to produce more regular networks yield networks with better general learning abilities; (2) whatever the encoding is, networks that are the more regular are statistically those that have the best learning abilities.

Charles Bonnet Syndrome: Evidence for a Generative Model in the Cortex?

Several theories propose that the cortex implements an internal model to explain, predict, and learn about sensory data, but the nature of this model is unclear. One condition that could be highly informative here is Charles Bonnet syndrome (CBS), where loss of vision leads to complex, vivid visual hallucinations of objects, people, and whole scenes. CBS could be taken as indication that there is a generative model in the brain, specifically one that can synthesise rich, consistent visual representations even in the absence of actual visual input. The processes that lead to CBS are poorly understood. Here, we argue that a model recently introduced in machine learning, the deep Boltzmann machine (DBM), could capture the relevant aspects of (hypothetical) generative processing in the cortex. The DBM carries both the semantics of a probabilistic generative model and of a neural network. The latter allows us to model a concrete neural mechanism that could underlie CBS, namely, homeostatic regulation of neuronal activity. We show that homeostatic plasticity could serve to make the learnt internal model robust against e.g. degradation of sensory input, but overcompensate in the case of CBS, leading to hallucinations. We demonstrate how a wide range of features of CBS can be explained in the model and suggest a potential role for the neuromodulator acetylcholine. This work constitutes the first concrete computational model of CBS and the first application of the DBM as a model in computational neuroscience. Our results lend further credence to the hypothesis of a generative model in the brain.

A Healthy Fear of the Unknown: Perspectives on the Interpretation of Parameter Fits from Computational Models in Neuroscience

Fitting models to behavior is commonly used to infer the latent computational factors responsible for generating behavior. However, the complexity of many behaviors can handicap the interpretation of such models. Here we provide perspectives on problems that can arise when interpreting parameter fits from models that provide incomplete descriptions of behavior. We illustrate these problems by fitting commonly used and neurophysiologically motivated reinforcement-learning models to simulated behavioral data sets from learning tasks. These model fits can pass a host of standard goodness-of-fit tests and other model-selection diagnostics even when the models do not provide a complete description of the behavioral data. We show that such incomplete models can be misleading by yielding biased estimates of the parameters explicitly included in the models. This problem is particularly pernicious when the neglected factors are unknown and therefore not easily identified by model comparisons and similar methods. An obvious conclusion is that a parsimonious description of behavioral data does not necessarily imply an accurate description of the underlying computations. Moreover, general goodness-of-fit measures are not a strong basis to support claims that a particular model can provide a generalized understanding of the computations that govern behavior. To help overcome these challenges, we advocate the design of tasks that provide direct reports of the computational variables of interest. Such direct reports complement model-fitting approaches by providing a more complete, albeit possibly more task-specific, representation of the factors that drive behavior. Computational models then provide a means to connect such task-specific results to a more general algorithmic understanding of the brain.

Exploring the cognitive and motor functions of the basal ganglia: an integrative review of computational cognitive neuroscience models

Many computational models of the basal ganglia (BG) have been proposed over the past twenty-five years. While computational neuroscience models have focused on closely matching the neurobiology of the BG, computational cognitive neuroscience (CCN) models have focused on how the BG can be used to implement cognitive and motor functions. This review article focuses on CCN models of the BG and how they use the neuroanatomy of the BG to account for cognitive and motor functions such as categorization, instrumental conditioning, probabilistic learning, working memory, sequence learning, automaticity, reaching, handwriting, and eye saccades. A total of 19 BG models accounting for one or more of these functions are reviewed and compared. The review concludes with a discussion of the limitations of existing CCN models of the BG and prescriptions for future modeling, including the need for computational models of the BG that can simultaneously account for cognitive and motor functions, and the need for a more complete specification of the role of the BG in behavioral functions.

Multiscale modeling with GENESIS 3, using the G-shell and Python

The CBI architecture [1] being used as the basis of 3 (G-3) allows a single model-container to be used to describe a model spanning many levels of scale. This feature allows a user to transparently run multi-scale simulations. As will be described in further detail during the workshop “Multi-Scale Modeling in Computational Neuroscience II: Challenges and Opportunities, the CBI architecture contains a communication component to upscale and downscale numerical variables when moving across different levels of scale. These new capabilities and advances in G-3 usability also allow interfacing with many Python graphical tools (e.g. wxPython, matplotlib), potential web interfaces (e.g. Django), and other independent modules (e.g. Chemesis-3) for use in that cover multiple levels of scale. Progress in developing Python interfaces to G-3 [2], combined with recent implementation of network and biochemical modeling capabilities in G-3 have allowed us to construct a new series of self-guided hands-on modeling tutorials. These are being introduced at the Introduction to Genesis 3 Workshop held in Luebeck, Germany 30 April – 5 May 2012 (https://www.gradschool.uni-luebeck.de/index.php?id=366). This poster provides an introduction to these new modeling capabilities, and to the new instructional material. Additions to the existing G-3 tutorials on use of the G-shell cover network creation commands and the use of the Chemesis-3 module. The rewritten version of the tutorial Creating large networks with GENESIS demonstrates the use of Python scripting to create cortical network models in G-3. The tutorial Adding a GUI to G-3 simulations shows users how to leverage the Python programming interface to construct visual tools. As an example, Figure ​Figure11 illustrates the use of the new G-3 Netview visualization application to display and replay an animation of the spreading excitation in the RSnet2 simulation that is the basis of the network modeling tutorial. Figure 1 A short injection pulse applied to a 32 x 32 cell network of coupled excitatory cells starts a wave of excitation that can be replayed and examined in detail during any time window, using the G-3 Network Viewer.

The layer-oriented approach to declarative languages for biological modeling.

We present a new approach to modeling languages for computational biology, which we call the layer-oriented approach. The approach stems from the observation that many diverse biological phenomena are described using a small set of mathematical formalisms (e.g. differential equations), while at the same time different domains and subdomains of computational biology require that models are structured according to the accepted terminology and classification of that domain. Our approach uses distinct semantic layers to represent the domain-specific biological concepts and the underlying mathematical formalisms. Additional functionality can be transparently added to the language by adding more layers. This approach is specifically concerned with declarative languages, and throughout the paper we note some of the limitations inherent to declarative approaches. The layer-oriented approach is a way to specify explicitly how high-level biological modeling concepts are mapped to a computational representation, while abstracting away details of particular programming languages and simulation environments. To illustrate this process, we define an example language for describing models of ionic currents, and use a general mathematical notation for semantic transformations to show how to generate model simulation code for various simulation environments. We use the example language to describe a Purkinje neuron model and demonstrate how the layer-oriented approach can be used for solving several practical issues of computational neuroscience model development. We discuss the advantages and limitations of the approach in comparison with other modeling language efforts in the domain of computational biology and outline some principles for extensible, flexible modeling language design. We conclude by describing in detail the semantic transformations defined for our language.

Artificial Neural Expert Computing Models for Determining Shelf Life of Processed Cheese

Time-delay single and multi layer models were developed for predicting shelf life of processed cheese stored at 30 o C. Processed cheese is very nutritious dairy product, rich in milk proteins and milk fat. For developing computational neuroscience models,experimental data relating to body & texture, aroma & flavour, moisture, free fatty acids were taken as input variables, while sensory score as output variable. Mean Square Error, Root Mean Square Error, Coefficient of determination and Nash - Sutcliffo Coefficient were applied in order to compare the prediction performance of the developed computational models. The results of the study established excellent correlation between experimental data and the predicted values, with a high determination coefficient. From the study it was concluded that artificial neural expert time-delay models are good for predicting the shelf life of processed cheese. DOI: http://dx.doi.org/10.11591/ijece.v2i3.353