Feedforward Categorization on AER Motion Events Using Cortex-Like Features in a Spiking Neural Network

Exploiting Lightweight Statistical Learning for Event-Based Vision Processing

This paper presents a lightweight statistical learning framework potentially suitable for low-cost event-based vision systems, where visual information is captured by a dynamic vision sensor (DVS) and represented as an asynchronous stream of pixel addresses (events) indicating a relative intensity change on those locations. A simple random ferns classifier based on randomly selected patch-based binary features is employed to categorize pixel event flows. Our experimental results demonstrate that compared to existing event-based processing algorithms, such as spiking convolutional neural networks (SCNNs) and the state-of-the-art bag-of-events (BoE)-based statistical algorithms, our framework excels in high processing speed (2× faster than the BoE statistical methods and >100× faster than previous SCNNs in training speed) with extremely simple online learning process, and achieves state-of-the-art classification accuracy on four popular address-event representation data sets: MNIST-DVS, Poker-DVS, Posture-DVS, and CIFAR10-DVS. Hardware estimation shows that our algorithm will be preferable for low-cost embedded system implementations.

A Configurable Event-Driven Convolutional Node with Rate Saturation Mechanism for Modular ConvNet Systems Implementation

Convolutional Neural Networks (ConvNets) are a particular type of neural network often used for many applications like image recognition, video analysis or natural language processing. They are inspired by the human brain, following a specific organization of the connectivity pattern between layers of neurons known as receptive field. These networks have been traditionally implemented in software, but they are becoming more computationally expensive as they scale up, having limitations for real-time processing of high-speed stimuli. On the other hand, hardware implementations show difficulties to be used for different applications, due to their reduced flexibility. In this paper, we propose a fully configurable event-driven convolutional node with rate saturation mechanism that can be used to implement arbitrary ConvNets on FPGAs. This node includes a convolutional processing unit and a routing element which allows to build large 2D arrays where any multilayer structure can be implemented. The rate saturation mechanism emulates the refractory behavior in biological neurons, guaranteeing a minimum separation in time between consecutive events. A 4-layer ConvNet with 22 convolutional nodes trained for poker card symbol recognition has been implemented in a Spartan6 FPGA. This network has been tested with a stimulus where 40 poker cards were observed by a Dynamic Vision Sensor (DVS) in 1 s time. Different slow-down factors were applied to characterize the behavior of the system for high speed processing. For slow stimulus play-back, a 96% recognition rate is obtained with a power consumption of 0.85 mW. At maximum play-back speed, a traffic control mechanism downsamples the input stimulus, obtaining a recognition rate above 63% when less than 20% of the input events are processed, demonstrating the robustness of the network.

Feature Representations for Neuromorphic Audio Spike Streams

Event-driven neuromorphic spiking sensors such as the silicon retina and the silicon cochlea encode the external sensory stimuli as asynchronous streams of spikes across different channels or pixels. Combining state-of-art deep neural networks with the asynchronous outputs of these sensors has produced encouraging results on some datasets but remains challenging. While the lack of effective spiking networks to process the spike streams is one reason, the other reason is that the pre-processing methods required to convert the spike streams to frame-based features needed for the deep networks still require further investigation. This work investigates the effectiveness of synchronous and asynchronous frame-based features generated using spike count and constant event binning in combination with the use of a recurrent neural network for solving a classification task using N-TIDIGITS18 dataset. This spike-based dataset consists of recordings from the Dynamic Audio Sensor, a spiking silicon cochlea sensor, in response to the TIDIGITS audio dataset. We also propose a new pre-processing method which applies an exponential kernel on the output cochlea spikes so that the interspike timing information is better preserved. The results from the N-TIDIGITS18 dataset show that the exponential features perform better than the spike count features, with over 91% accuracy on the digit classification task. This accuracy corresponds to an improvement of at least 2.5% over the use of spike count features, establishing a new state of the art for this dataset.

A Scalable Multicore Architecture With Heterogeneous Memory Structures for Dynamic Neuromorphic Asynchronous Processors (DYNAPs)

Neuromorphic computing systems comprise networks of neurons that use asynchronous events for both computation and communication. This type of representation offers several advantages in terms of bandwidth and power consumption in neuromorphic electronic systems. However, managing the traffic of asynchronous events in large scale systems is a daunting task, both in terms of circuit complexity and memory requirements. Here, we present a novel routing methodology that employs both hierarchical and mesh routing strategies and combines heterogeneous memory structures for minimizing both memory requirements and latency, while maximizing programming flexibility to support a wide range of event-based neural network architectures, through parameter configuration. We validated the proposed scheme in a prototype multicore neuromorphic processor chip that employs hybrid analog/digital circuits for emulating synapse and neuron dynamics together with asynchronous digital circuits for managing the address-event traffic. We present a theoretical analysis of the proposed connectivity scheme, describe the methods and circuits used to implement such scheme, and characterize the prototype chip. Finally, we demonstrate the use of the neuromorphic processor with a convolutional neural network for the real-time classification of visual symbols being flashed to a dynamic vision sensor (DVS) at high speed.

Efficient Training of Supervised Spiking Neural Network via Accurate Synaptic-Efficiency Adjustment Method

The spiking neural network (SNN) is the third generation of neural networks and performs remarkably well in cognitive tasks, such as pattern recognition. The temporal neural encode mechanism found in biological hippocampus enables SNN to possess more powerful computation capability than networks with other encoding schemes. However, this temporal encoding approach requires neurons to process information serially on time, which reduces learning efficiency significantly. To keep the powerful computation capability of the temporal encoding mechanism and to overcome its low efficiency in the training of SNNs, a new training algorithm, the accurate synaptic-efficiency adjustment method is proposed in this paper. Inspired by the selective attention mechanism of the primate visual system, our algorithm selects only the target spike time as attention areas, and ignores voltage states of the untarget ones, resulting in a significant reduction of training time. Besides, our algorithm employs a cost function based on the voltage difference between the potential of the output neuron and the firing threshold of the SNN, instead of the traditional precise firing time distance. A normalized spike-timing-dependent-plasticity learning window is applied to assigning this error to different synapses for instructing their training. Comprehensive simulations are conducted to investigate the learning properties of our algorithm, with input neurons emitting both single spike and multiple spikes. Simulation results indicate that our algorithm possesses higher learning performance than the existing other methods and achieves the state-of-the-art efficiency in the training of SNN.

CIFAR10-DVS: An Event-Stream Dataset for Object Classification

Neuromorphic vision research requires high-quality and appropriately challenging event-stream datasets to support continuous improvement of algorithms and methods. However, creating event-stream datasets is a time-consuming task, which needs to be recorded using the neuromorphic cameras. Currently, there are limited event-stream datasets available. In this work, by utilizing the popular computer vision dataset CIFAR-10, we converted 10,000 frame-based images into 10,000 event streams using a dynamic vision sensor (DVS), providing an event-stream dataset of intermediate difficulty in 10 different classes, named as "CIFAR10-DVS." The conversion of event-stream dataset was implemented by a repeated closed-loop smooth (RCLS) movement of frame-based images. Unlike the conversion of frame-based images by moving the camera, the image movement is more realistic in respect of its practical applications. The repeated closed-loop image movement generates rich local intensity changes in continuous time which are quantized by each pixel of the DVS camera to generate events. Furthermore, a performance benchmark in event-driven object classification is provided based on state-of-the-art classification algorithms. This work provides a large event-stream dataset and an initial benchmark for comparison, which may boost algorithm developments in even-driven pattern recognition and object classification.

Bag of Events: An Efficient Probability-Based Feature Extraction Method for AER Image Sensors

Address event representation (AER) image sensors represent the visual information as a sequence of events that denotes the luminance changes of the scene. In this paper, we introduce a feature extraction method for AER image sensors based on the probability theory, namely, bag of events (BOE). The proposed approach represents each object as the joint probability distribution of the concurrent events, and each event corresponds to a unique activated pixel of the AER sensor. The advantages of BOE include: 1) it is a statistical learning method and has a good interpretability in mathematics; 2) BOE can significantly reduce the effort to tune parameters for different data sets, because it only has one hyperparameter and is robust to the value of the parameter; 3) BOE is an online learning algorithm, which does not require the training data to be collected in advance; 4) BOE can achieve competitive results in real time for feature extraction (>275 frames/s and >120,000 events/s); and 5) the implementation complexity of BOE only involves some basic operations, e.g., addition and multiplication. This guarantees the hardware friendliness of our method. The experimental results on three popular AER databases (i.e., MNIST-dynamic vision sensor, Poker Card, and Posture) show that our method is remarkably faster than two recently proposed AER categorization systems while preserving a good classification accuracy.

An Event-Based Neurobiological Recognition System with Orientation Detector for Objects in Multiple Orientations

A new multiple orientation event-based neurobiological recognition system is proposed by integrating recognition and tracking function in this paper, which is used for asynchronous address-event representation (AER) image sensors. The characteristic of this system has been enriched to recognize the objects in multiple orientations with only training samples moving in a single orientation. The system extracts multi-scale and multi-orientation line features inspired by models of the primate visual cortex. An orientation detector based on modified Gaussian blob tracking algorithm is introduced for object tracking and orientation detection. The orientation detector and feature extraction block work in simultaneous mode, without any increase in categorization time. An addresses lookup table (addresses LUT) is also presented to adjust the feature maps by addresses mapping and reordering, and they are categorized in the trained spiking neural network. This recognition system is evaluated with the MNIST dataset which have played important roles in the development of computer vision, and the accuracy is increased owing to the use of both ON and OFF events. AER data acquired by a dynamic vision senses (DVS) are also tested on the system, such as moving digits, pokers, and vehicles. The experimental results show that the proposed system can realize event-based multi-orientation recognition. The work presented in this paper makes a number of contributions to the event-based vision processing system for multi-orientation object recognition. It develops a new tracking-recognition architecture to feedforward categorization system and an address reorder approach to classify multi-orientation objects using event-based data. It provides a new way to recognize multiple orientation objects with only samples in single orientation.

Benchmarking Spike-Based Visual Recognition: A Dataset and Evaluation

Today, increasing attention is being paid to research into spike-based neural computation both to gain a better understanding of the brain and to explore biologically-inspired computation. Within this field, the primate visual pathway and its hierarchical organization have been extensively studied. Spiking Neural Networks (SNNs), inspired by the understanding of observed biological structure and function, have been successfully applied to visual recognition and classification tasks. In addition, implementations on neuromorphic hardware have enabled large-scale networks to run in (or even faster than) real time, making spike-based neural vision processing accessible on mobile robots. Neuromorphic sensors such as silicon retinas are able to feed such mobile systems with real-time visual stimuli. A new set of vision benchmarks for spike-based neural processing are now needed to measure progress quantitatively within this rapidly advancing field. We propose that a large dataset of spike-based visual stimuli is needed to provide meaningful comparisons between different systems, and a corresponding evaluation methodology is also required to measure the performance of SNN models and their hardware implementations. In this paper we first propose an initial NE (Neuromorphic Engineering) dataset based on standard computer vision benchmarksand that uses digits from the MNIST database. This dataset is compatible with the state of current research on spike-based image recognition. The corresponding spike trains are produced using a range of techniques: rate-based Poisson spike generation, rank order encoding, and recorded output from a silicon retina with both flashing and oscillating input stimuli. In addition, a complementary evaluation methodology is presented to assess both model-level and hardware-level performance. Finally, we demonstrate the use of the dataset and the evaluation methodology using two SNN models to validate the performance of the models and their hardware implementations. With this dataset we hope to (1) promote meaningful comparison between algorithms in the field of neural computation, (2) allow comparison with conventional image recognition methods, (3) provide an assessment of the state of the art in spike-based visual recognition, and (4) help researchers identify future directions and advance the field.

Structural Plasticity Denoises Responses and Improves Learning Speed

Despite an abundance of computational models for learning of synaptic weights, there has been relatively little research on structural plasticity, i.e., the creation and elimination of synapses. Especially, it is not clear how structural plasticity works in concert with spike-timing-dependent plasticity (STDP) and what advantages their combination offers. Here we present a fairly large-scale functional model that uses leaky integrate-and-fire neurons, STDP, homeostasis, recurrent connections, and structural plasticity to learn the input encoding, the relation between inputs, and to infer missing inputs. Using this model, we compare the error and the amount of noise in the network's responses with and without structural plasticity and the influence of structural plasticity on the learning speed of the network. Using structural plasticity during learning shows good results for learning the representation of input values, i.e., structural plasticity strongly reduces the noise of the response by preventing spikes with a high error. For inferring missing inputs we see similar results, with responses having less noise if the network was trained using structural plasticity. Additionally, using structural plasticity with pruning significantly decreased the time to learn weights suitable for inference. Presumably, this is due to the clearer signal containing less spikes that misrepresent the desired value. Therefore, this work shows that structural plasticity is not only able to improve upon the performance using STDP without structural plasticity but also speeds up learning. Additionally, it addresses the practical problem of limited resources for connectivity that is not only apparent in the mammalian neocortex but also in computer hardware or neuromorphic (brain-inspired) hardware by efficiently pruning synapses without losing performance.

Multiclass Classification by Adaptive Network of Dendritic Neurons with Binary Synapses Using Structural Plasticity

The development of power-efficient neuromorphic devices presents the challenge of designing spike pattern classification algorithms which can be implemented on low-precision hardware and can also achieve state-of-the-art performance. In our pursuit of meeting this challenge, we present a pattern classification model which uses a sparse connection matrix and exploits the mechanism of nonlinear dendritic processing to achieve high classification accuracy. A rate-based structural learning rule for multiclass classification is proposed which modifies a connectivity matrix of binary synaptic connections by choosing the best "k" out of "d" inputs to make connections on every dendritic branch (k < < d). Because learning only modifies connectivity, the model is well suited for implementation in neuromorphic systems using address-event representation (AER). We develop an ensemble method which combines several dendritic classifiers to achieve enhanced generalization over individual classifiers. We have two major findings: (1) Our results demonstrate that an ensemble created with classifiers comprising moderate number of dendrites performs better than both ensembles of perceptrons and of complex dendritic trees. (2) In order to determine the moderate number of dendrites required for a specific classification problem, a two-step solution is proposed. First, an adaptive approach is proposed which scales the relative size of the dendritic trees of neurons for each class. It works by progressively adding dendrites with fixed number of synapses to the network, thereby allocating synaptic resources as per the complexity of the given problem. As a second step, theoretical capacity calculations are used to convert each neuronal dendritic tree to its optimal topology where dendrites of each class are assigned different number of synapses. The performance of the model is evaluated on classification of handwritten digits from the benchmark MNIST dataset and compared with other spike classifiers. We show that our system can achieve classification accuracy within 1 - 2% of other reported spike-based classifiers while using much less synaptic resources (only 7%) compared to that used by other methods. Further, an ensemble classifier created with adaptively learned sizes can attain accuracy of 96.4% which is at par with the best reported performance of spike-based classifiers. Moreover, the proposed method achieves this by using about 20% of the synapses used by other spike algorithms. We also present results of applying our algorithm to classify the MNIST-DVS dataset collected from a real spike-based image sensor and show results comparable to the best reported ones (88.1% accuracy). For VLSI implementations, we show that the reduced synaptic memory can save upto 4X area compared to conventional crossbar topologies. Finally, we also present a biologically realistic spike-based version for calculating the correlations required by the structural learning rule and demonstrate the correspondence between the rate-based and spike-based methods of learning.

Poker-DVS and MNIST-DVS. Their History, How They Were Made, and Other Details

This article reports on two databases for event-driven object recognition using a Dynamic Vision Sensor (DVS). The first, which we call Poker-DVS and is being released together with this article, was obtained by browsing specially made poker card decks in front of a DVS camera for 2–4 s. Each card appeared on the screen for about 20–30 ms. The poker pips were tracked and isolated off-line to constitute the 131-recording Poker-DVS database. The second database, which we call MNIST-DVS and which was released in December 2013, consists of a set of 30,000 DVS camera recordings obtained by displaying 10,000 moving symbols from the standard MNIST 70,000-picture database on an LCD monitor for about 2–3 s each. Each of the 10,000 symbols was displayed at three different scales, so that event-driven object recognition algorithms could easily be tested for different object sizes. This article tells the story behind both databases, covering, among other aspects, details of how they work and the reasons for their creation. We provide not only the databases with corresponding scripts, but also the scripts and data used to generate the figures shown in this article (as Supplementary Material).

Converting Static Image Datasets to Spiking Neuromorphic Datasets Using Saccades

Creating datasets for Neuromorphic Vision is a challenging task. A lack of available recordings from Neuromorphic Vision sensors means that data must typically be recorded specifically for dataset creation rather than collecting and labeling existing data. The task is further complicated by a desire to simultaneously provide traditional frame-based recordings to allow for direct comparison with traditional Computer Vision algorithms. Here we propose a method for converting existing Computer Vision static image datasets into Neuromorphic Vision datasets using an actuated pan-tilt camera platform. Moving the sensor rather than the scene or image is a more biologically realistic approach to sensing and eliminates timing artifacts introduced by monitor updates when simulating motion on a computer monitor. We present conversion of two popular image datasets (MNIST and Caltech101) which have played important roles in the development of Computer Vision, and we provide performance metrics on these datasets using spike-based recognition algorithms. This work contributes datasets for future use in the field, as well as results from spike-based algorithms against which future works can compare. Furthermore, by converting datasets already popular in Computer Vision, we enable more direct comparison with frame-based approaches.

Unsupervised learning of digit recognition using spike-timing-dependent plasticity

In order to understand how the mammalian neocortex is performing computations, two things are necessary; we need to have a good understanding of the available neuronal processing units and mechanisms, and we need to gain a better understanding of how those mechanisms are combined to build functioning systems. Therefore, in recent years there is an increasing interest in how spiking neural networks (SNN) can be used to perform complex computations or solve pattern recognition tasks. However, it remains a challenging task to design SNNs which use biologically plausible mechanisms (especially for learning new patterns), since most such SNN architectures rely on training in a rate-based network and subsequent conversion to a SNN. We present a SNN for digit recognition which is based on mechanisms with increased biological plausibility, i.e., conductance-based instead of current-based synapses, spike-timing-dependent plasticity with time-dependent weight change, lateral inhibition, and an adaptive spiking threshold. Unlike most other systems, we do not use a teaching signal and do not present any class labels to the network. Using this unsupervised learning scheme, our architecture achieves 95% accuracy on the MNIST benchmark, which is better than previous SNN implementations without supervision. The fact that we used no domain-specific knowledge points toward the general applicability of our network design. Also, the performance of our network scales well with the number of neurons used and shows similar performance for four different learning rules, indicating robustness of the full combination of mechanisms, which suggests applicability in heterogeneous biological neural networks.

Turn Down That Noise: Synaptic Encoding of Afferent SNR in a Single Spiking Neuron

We have added a simplified neuromorphic model of Spike Time Dependent Plasticity (STDP) to the previously described Synapto-dendritic Kernel Adapting Neuron (SKAN), a hardware efficient neuron model capable of learning spatio-temporal spike patterns. The resulting neuron model is the first to perform synaptic encoding of afferent signal-to-noise ratio in addition to the unsupervised learning of spatio-temporal spike patterns. The neuron model is particularly suitable for implementation in digital neuromorphic hardware as it does not use any complex mathematical operations and uses a novel shift-based normalization approach to achieve synaptic homeostasis. The neuron's noise compensation properties are characterized and tested on random spatio-temporal spike patterns as well as a noise corrupted subset of the zero images of the MNIST handwritten digit dataset. Results show the simultaneously learning common patterns in its input data while dynamically weighing individual afferents based on their signal to noise ratio. Despite its simplicity the interesting behaviors of the neuron model and the resulting computational power may also offer insights into biological systems.