Brain models typically focus either on low-level biological detail or on qualitative behavioral effects. In contrast, we present a biologically-plausible spiking-neuron model of associative learning and recognition that accounts for both human behavior and low-level brain activity across the whole task. Based on cognitive theories and insights from machine-learning analyses of M/EEG data, the model proceeds through five processing stages: stimulus encoding, familiarity judgement, associative retrieval, decision making, and motor response. The results matched human response times and source-localized MEG data in occipital, temporal, prefrontal, and precentral brain regions; as well as a classic fMRI effect in prefrontal cortex. This required two main conceptual advances: a basal-ganglia-thalamus action-selection system that relies on brief thalamic pulses to change the functional connectivity of the cortex, and a new unsupervised learning rule that causes very strong pattern separation in the hippocampus. The resulting model shows how low-level brain activity can result in goal-directed cognitive behavior in humans.

Low latency and low energy consumption are the indispensable characteristics of Edge Computing applications. With the fusion of Edge Computing and Artificial Intelligence (AI) into Edge Intelligence, this need is more than ever. Of late, Spiking Neural Networks have shown a promise for low latency and low power AI when deployed on a neuromorphic hardware e.g., Intel's Loihi. In this paper, we present a Spiking Reservoir Computing model, based on the Legendre Memory Units which processes temporal data on Loihi hardware. Such a model is greatly suitable for the battery-powered AI enabled edge devices which call for a prompt processing of the temporal sensor-signals with high energy efficiency. We experiment our model with the ECG5000 dataset on the Loihi boards to show its efficacy.

Many animal behaviors require orientation and steering with respect to the environment. For insects, a key brain area involved in spatial orientation and navigation is the central complex. Activity in this neural circuit has been shown to track the insect's current heading relative to its environment and has also been proposed to be the substrate of path integration. However, it remains unclear how the output of the central complex is integrated into motor commands. Central complex output neurons project to the lateral accessory lobes (LAL), from which descending neurons project to thoracic motor centers. Here, we present a computational model of a simple neural network that has been described anatomically and physiologically in the LALs of male silkworm moths, in the context of odor-mediated steering. We present and analyze two versions of this network, one rate based and one based on spiking neurons. The modeled network consists of an inhibitory local interneuron and a bistable descending neuron (flip-flop) that both receive input in the LAL. The flip-flop neuron projects onto neck motor neurons to induce steering. We show that this simple computational model not only replicates the basic parameters of male silkworm moth behavior in a simulated odor plume but can also take input from a computational model of path integration in the central complex and use it to steer back to a point of origin. Furthermore, we find that increasing the level of detail within the model improves the realism of the model's behavior, leading to the emergence of looping behavior as an orientation strategy. Our results suggest that descending neurons originating in the LALs, such as flip-flop neurons, are sufficient to mediate multiple steering behaviors. This study is therefore a first step to close the gap between orientation circuits in the central complex and downstream motor centers.

Differentiable neural computers (DNCs) extend artificial neural networks with an explicit memory without interference, thus enabling the model to perform classic computation tasks, such as graph traversal. However, such models are difficult to train, requiring long training times and large datasets. In this work, we achieve some of the computational capabilities of DNCs with a model that can be trained very efficiently, namely, an echo state network with an explicit memory without interference. This extension enables echo state networks to recognize all regular languages, including those that contractive echo state networks provably cannot recognize. Furthermore, we demonstrate experimentally that our model performs comparably to its fully trained deep version on several typical benchmark tasks for DNCs.

While neural networks are highly effective at learning task-relevant representations from data, they typically do not learn representations with the kind of symbolic structure that is hypothesized to support high-level cognitive processes, nor do they naturally model such structures within problem domains that are continuous in space and time. To fill these gaps, this work exploits a method for defining vector representations that bind discrete (symbol-like) entities to points in continuous topological spaces in order to simulate and predict the behavior of a range of dynamical systems. These vector representations are spatial semantic pointers (SSPs), and we demonstrate that they can (1) be used to model dynamical systems involving multiple objects represented in a symbol-like manner and (2) be integrated with deep neural networks to predict the future of physical trajectories. These results help unify what have traditionally appeared to be disparate approaches in machine learning.

Predicting future behavior of other traffic participants is an essential task that needs to be solved by automated vehicles and human drivers alike to achieve safe and situation-aware driving. Modern approaches to vehicles trajectory prediction typically rely on data-driven models like neural networks, in particular LSTMs (Long Short-Term Memorys), achieving promising results. However, the question of optimal composition of the underlying training data has received less attention. In this paper, we expand on previous work on vehicle trajectory prediction based on neural network models employing distributed representations to encode automotive scenes in a semantic vector substrate. We analyze the influence of variations in the training data on the performance of our prediction models. Thereby, we show that the models employing our semantic vector representation outperform the numerical model when trained on an adequate data set and thereby, that the composition of training data in vehicle trajectory prediction is crucial for successful training. We conduct our analysis on challenging real-world driving data.

Vector Symbolic Architectures belong to a family of related cognitive modeling approaches that encode symbols and structures in high-dimensional vectors. Similar to human subjects, whose capacity to process and store information or concepts in short-term memory is subject to numerical restrictions, the capacity of information that can be encoded in such vector representations is limited and one way of modeling the numerical restrictions to cognition. In this paper, we analyze these limits regarding information capacity of distributed representations. We focus our analysis on simple superposition and more complex, structured representations involving convolutive powers to encode spatial information. In two experiments, we find upper bounds for the number of concepts that can effectively be stored in a single vector only depending on the dimensionality of the underlying vector space.

This paper presents a neural-network based adaptive feedback control structure to regulate the velocity of 3D bipedal robots under dynamics uncertainties. Existing Hybrid Zero Dynamics (HZD)-based controllers regulate velocity through the implementation of heuristic regulators that do not consider model and environmental uncertainties, which may significantly affect the tracking performance of the controllers. In this paper, we address the uncertainties in the robot dynamics from the perspective of the reduced dimensional representation of virtual constraints and propose the integration of an adaptive neural network-based controller to regulate the robot velocity in the presence of model parameter uncertainties. The proposed approach yields improved tracking performance under dynamics uncertainties. The shallow adaptive neural network used in this paper does not require training a priori and has the potential to be implemented on the real-time robotic controller. A comparative simulation study of a 3D Cassie robot is presented to illustrate the performance of the proposed approach under various scenarios.

In this paper, we present a functional spiking-neuron model of human working memory (WM).This model combines neural firing for encoding of information with activity-silent maintenance.While it used to be widely assumed that information in WM is maintained through persistent recurrent activity, recent studies have shown that information can be maintained without persistent firing; instead, information can be stored in activity-silent states.A candidate mechanism underlying this type of storage is short-term synaptic plasticity (STSP), by which the strength of connections between neurons rapidly changes to encode new information.To demonstrate that STSP can lead to functional behavior, we integrated STSP by means of calcium-mediated synaptic facilitation in a large-scale spikingneuron model and added a decision mechanism.The model was used to simulate a recent study that measured behavior and EEG activity of participants in three delayed-response tasks.In these tasks, one or two visual gratings had to be maintained in WM, and compared to subsequent probes.The original study demonstrated that WM contents and its priority status could be decoded from neural activity elicited by a task-irrelevant stimulus displayed during the activity-silent maintenance period.In support of our model, we show that it can perform these tasks, and that both its behavior as well as its neural representations are in agreement with the human data.We conclude that information in WM can be effectively maintained in activity-silent states by means of calcium-mediated STSP. Author summaryMentally maintaining information for short periods of time in working memory is crucial for human adaptive behavior.It was recently shown that the human brain does not only store information through neural firing-as was widely believed-but also maintains information in activity-silent states.Here, we present a detailed neural model of how this could

Braindrop is the first neuromorphic system designed to be programmed at a high level of abstraction. Previous neuromorphic systems were programmed at the neurosynaptic level and required expert knowledge of the hardware to use. In stark contrast, Braindrop's computations are specified as coupled nonlinear dynamical systems and synthesized to the hardware by an automated procedure. This procedure not only leverages Braindrop's fabric of subthreshold analog circuits as dynamic computational primitives but also compensates for their mismatched and temperature-sensitive responses at the network level. Thus, a clean abstraction is presented to the user. Fabricated in a 28-nm FDSOI process, Braindrop integrates 4096 neurons in 0.65 mm2. Two innovations-sparse encoding through analog spatial convolution and weighted spike-rate summation though digital accumulative thinning-cut digital traffic drastically, reducing the energy Braindrop consumes per equivalent synaptic operation to 381 fJ for typical network configurations.

Low-power, high-speed neural networks are critical for providing deployable embedded AI applications at the edge. We describe an FPGA implementation of Neural Engineering Framework (NEF) networks with online learning that outperforms mobile GPU implementations by an order of magnitude or more. Specifically, we provide an embedded Python-capable PYNQ FPGA implementation supported with a High-Level Synthesis (HLS) workflow that allows sub-millisecond implementation of adaptive neural networks with low-latency, direct I/O access to the physical world. We tune the precision of the different intermediate variables in the code to achieve competitive absolute accuracy against slower and larger floating-point reference designs. The online learning component of the neural network exploits immediate feedback to adjust the network weights to best support a given arithmetic precision. As the space of possible design configurations of such networks is vast and is subject to a target accuracy constraint, we use the Hyperopt hyper-parameter tuning tool instead of manual search to find Pareto optimal designs. Specifically, we are able to generate the optimized designs in under 500 iterations of Vivado HLS before running the complete Vivado place-and-route phase on that subset. For neural network populations of 64-4096 neurons and 1-8 representational dimensions our optimized FPGA implementation generated by Hyperopt has a speedup of 10-484× over a competing cuBLAS implementation on the Jetson TX1 GPU while using 2.4-9.5× less power. Our speedups are a result of HLS-specific reformulation (15× improvement), precision adaptation (4× improvement), and low-latency direct I/O access (1000× improvement).

Demonstration Setup: We will bring Braindrop, a mixed-signal neuromorphic chip that is configured to perform arbitrary computations using the Neural Engineering Framework (NEF). Fabricated in a 28-nm FDSOI process, Braindrop has 4,096 silicon neurons whose design we optimized for nonlinear function approximation (Fig. 1). Our optimization procedure consists of a pre-fabrication phase and a run-time phase. In the pre-fabrication phase, transistors are sized to introduce an intermediate amount of heterogeneity into the neurons' tuning curves: Not so little that spiking-thresholds bunch up in the middle of the function's domain and not so much that spiking-thresholds mostly fall outside the function's domain. In the run-time phase, the outliers-neurons that never spike or always spike-are rescued by adjusting programmable bias currents appropriately. We explored various choices of the number of programmable bias-current levels and the amount of transistor-mismatch during the design phase to determine the combination that yielded the highest number of good neurons. To facilitate design-space exploration, we developed a SPICE-derived compact model of the dependence of tuning-curve heterogeneity on transistor-mismatch.

Silicon neurons designed using subthreshold analog-circuit techniques offer low power and compact area but are exponentially sensitive to threshold-voltage mismatch in transistors. The resulting heterogeneity in the neurons' responses, however, provides a diverse set of basis functions for smooth nonlinear function approximation. For low-order polynomials, neuron spiking thresholds ought to be distributed uniformly across the function's domain. This uniform distribution is difficult to achieve solely by sizing transistors to titrate mismatch. With too much mismatch, many neuron's thresholds fall outside the domain (i.e. they either always spike or remain silent). With too little mismatch, all their thresholds bunch up in the middle of the domain. Here, we present a silicon-neuron design methodology that minimizes overall area by optimizing transistor sizes in concert with a few locally-stored programmable bits to adjust each neuron's offset (and gain). We validated this methodology in a 28-nm mixed analog-digital CMOS process. Compared to relying on mismatch alone, augmentation with digital correction effectively reduced silicon area by 38%.

We present a novel approach to achieving temperature-robust behavior in neuromorphic systems that operates at the population level, trading an increase in silicon-neuron count for robustness across temperature. Our silicon neurons' tuning curves were highly sensitive to temperature, which could be decoded from a 400-neuron population with a precision of 0.07° C. We overcame this temperature-sensitivity by combining methods from robust optimization theory with the Neural Engineering Framework. We developed two algorithms and compared their temperature-robustness across a range of 2° C by decoding one period of a sinusoid-like function from populations with 25 to 800 neurons. We find that 560 neurons are required to achieve the same precision across this temperature range as 35 neurons achieved at a single temperature.

The Neural Engineering Framework (NEF) is a theory for mapping computations onto biologically plausible networks of spiking neurons. This theory has been applied to a number of neuromorphic chips. However, within both silicon and real biological systems, synapses exhibit higher-order dynamics and heterogeneity. To date, the NEF has not explicitly addressed how to account for either feature. Here, we analytically extend the NEF to directly harness the dynamics provided by heterogeneous mixed-analog-digital synapses. This theory is successfully validated by simulating two fundamental dynamical systems in Nengo using circuit models validated in SPICE. Thus, our work reveals the potential to engineer robust neuromorphic systems with well-defined high-level behaviour that harness the low-level heterogeneous properties of their physical primitives with millisecond resolution.

By building and simulating neural systems we hope to understand how the brain may work and use this knowledge to build neural and cognitive systems to tackle engineering problems. The Neural Engineering Framework (NEF) is a hypothesis about how such systems may be constructed and has recently been used to build the world's first functional brain model, Spaun. However, while the NEF simplifies the design of neural networks, simulating them using standard computer hardware is still computationally expensive - often running far slower than biological real-time and scaling very poorly: problems the SpiNNaker neuromorphic simulator was designed to solve. In this paper we (1) argue that employing the same model of computation used for simulating general purpose spiking neural networks on SpiNNaker for NEF models results in suboptimal use of the architecture, and (2) provide and evaluate an alternative simulation scheme which overcomes the memory and compute challenges posed by the NEF. This proposed method uses factored weight matrices to reduce memory usage by around 90% and, in some cases, simulate 2000 neurons on a processing core - double the SpiNNaker architectural target.

Living organisms are capable of autonomously adapting to dynamically changing environments by receiving inputs from highly specialized sensory organs and elaborating them on the same parallel, power-efficient neural substrate. In this paper we present a prototype for a comprehensive integrated platform that allows replicating principles of neural information processing in real-time. Our system consists of (a) an autonomous mobile robotic platform, (b) on-board actuators and multiple (neuromorphic) sensors, and (c) the SpiNNaker computing system, a configurable neural architecture for exploration of parallel, brain-inspired models. The simulation of neurally inspired perception and reasoning algorithms is performed in real-time by distributed, low-power, low-latency event-driven computing nodes, which can be flexibly configured using C or specialized neural languages such as PyNN and Nengo. We conclude by demonstrating the platform in two experimental scenarios, exhibiting real-world closed loop behavior consisting of environmental perception, reasoning and execution of adequate motor actions.

In this paper, we review the theoretical and software tools used to construct Spaun, the first (and so far only) brain model capable of performing cognitive tasks. This tool set allowed us to configure 2.5 million simple nonlinear components (neurons) with 60 billion connections between them (synapses) such that the resulting model can perform eight different perceptual, motor, and cognitive tasks. To reverse-engineer the brain in this way, a method is needed that shows how large numbers of simple components, each of which receives thousands of inputs from other components, can be organized to perform the desired computations. We achieve this through the neural engineering framework (NEF), a mathematical theory that provides methods for systematically generating biologically plausible spiking networks to implement nonlinear and linear dynamical systems. On top of this, we propose the semantic pointer architecture (SPA), a hypothesis regarding some aspects of the organization, function, and representational resources used in the mammalian brain. We conclude by discussing Spaun, which is an example model that uses the SPA and is implemented using the NEF. Throughout, we discuss the software tool Neural ENGineering Objects (Nengo), which allows for the synthesis and simulation of neural models efficiently on the scale of Spaun, and provides support for constructing models using the NEF and the SPA. The resulting NEF/SPA/Nengo combination is a general tool set for both evaluating hypotheses about how the brain works, and for building systems that compute particular functions using neuron-like components.

Visuospatial attention produces myriad effects on the activity and selectivity of cortical neurons. Spiking neuron models capable of reproducing a wide variety of these effects remain elusive. We present a model called the Attentional Routing Circuit (ARC) that provides a mechanistic description of selective attentional processing in cortex. The model is described mathematically and implemented at the level of individual spiking neurons, with the computations for performing selective attentional processing being mapped to specific neuron types and laminar circuitry. The model is used to simulate three studies of attention in macaque, and is shown to quantitatively match several observed forms of attentional modulation. Specifically, ARC demonstrates that with shifts of spatial attention, neurons may exhibit shifting and shrinking of receptive fields; increases in responses without changes in selectivity for non-spatial features (i.e. response gain), and; that the effect on contrast-response functions is better explained as a response-gain effect than as contrast-gain. Unlike past models, ARC embodies a single mechanism that unifies the above forms of attentional modulation, is consistent with a wide array of available data, and makes several specific and quantifiable predictions.

Recent papers have shown the possibility to implement large scale neural network models that perform complex algorithms in a biologically realistic way. However, such models have been simulated on architectures unable to perform real-time simulations. In previous work we presented the possibility to simulate simple models in real-time on the SpiNNaker neuromimetic architecture. However, such models were “static”: the algorithm performed was defined at design-time. In this paper we present a novel learning rule, that exploits the peculiarities of the SpiNNaker system, enabling models designed with the Neural Engineering Framework (NEF) to learn transfer functions using a supervised framework. We show that the proposed learning rule, belonging to the Prescribed Error Sensitivity (PES) class, is able to learn, effectively, both linear and non-linear functions.

Simulation of large-scale networks of spiking neurons has become appealing for understanding the computational principles of the nervous system by producing models based on biological evidence. In particular, networks that can assume a variety of (dynamically) stable states have been proposed as the basis for different behavioural and cognitive functions. This work focuses on implementing the Neural Engineering Framework (NEF), a formal method for mapping attractor networks and control-theoretic algorithms to biologically plausible networks of spiking neurons, on the SpiNNaker system, a massive programmable parallel architecture oriented to the simulation of networks of spiking neurons. We describe how to encode and decode analog values to patterns of neural spikes directly on chip. These methods take advantage of the full programmability of the ARM968 cores constituting the processing base of a SpiNNaker node, and exploit the fast Network-on-chip for spike communication. In this paper we focus on the fundamentals of representing, transforming and implementing dynamics in spiking networks. We show real time simulation results demonstrating the NEF principles and discuss advantages, precision and scalability. More generally, the present approach can be used to state and test hypotheses with large-scale spiking neural network models for a range of different cognitive functions and behaviours.

Motor prostheses aim to restore function to disabled patients. Despite compelling proof of concept systems, barriers to clinical translation remain. One challenge is to develop a low-power, fully-implantable system that dissipates only minimal power so as not to damage tissue. To this end, we implemented a Kalman-filter based decoder via a spiking neural network (SNN) and tested it in brain-machine interface (BMI) experiments with a rhesus monkey. The Kalman filter was trained to predict the arm’s velocity and mapped on to the SNN using the Neural Engineer- ing Framework (NEF). A 2,000-neuron embedded Matlab SNN implementation runs in real-time and its closed-loop performance is quite comparable to that of the standard Kalman filter. The success of this closed-loop decoder holds promise for hardware SNN implementations of statistical signal processing algorithms on neuromorphic chips, which may offer power savings necessary to overcome a major obstacle to the successful clinical translation of neural motor prostheses.

This paper re-examines the question of localist vs. distributed neural representations using a biologically realistic framework based on the central notion of neurons having a preferred direction vector. A preferred direction vector captures the general observation that neurons fire most vigorously when the stimulus lies in a particular direction in a represented vector space. This framework has been successful in capturing a wide variety of detailed neural data, although here we focus on cognitive representation. In particular, we describe methods for constructing spiking networks that can represent and manipulate structured, symbol-like representations. In the context of such networks, neuron activities can seem both localist and distributed, depending on the space of inputs being considered. This analysis suggests that claims of a set of neurons being localist or distributed cannot be made sense of without specifying the particular stimulus set used to examine the neurons.