Neuromorphic computing, which aims to replicate the computational structure and architecture of the brain in synthetic hardware, has typically focused on artificial intelligence applications. What is less explored is whether such brain-inspired hardware can provide value beyond cognitive tasks. Here we show that the high degree of parallelism and configurability of spiking neuromorphic architectures makes them well suited to implement random walks via discrete-time Markov chains. These random walks are useful in Monte Carlo methods, which represent a fundamental computational tool for solving a wide range of numerical computing tasks. Using IBM’s TrueNorth and Intel’s Loihi neuromorphic computing platforms, we show that our neuromorphic computing algorithm for generating random walk approximations of diffusion offers advantages in energy-efficient computation compared with conventional approaches. We also show that our neuromorphic computing algorithm can be extended to more sophisticated jump-diffusion processes that are useful in a range of applications, including financial economics, particle physics and machine learning.
Graph algorithms enable myriad large-scale applications including cybersecurity, social network analysis, resource allocation, and routing. The scalability of current graph algorithm implementations on conventional computing architectures are hampered by the demise of Moore’s law. We present a theoretical framework for designing and assessing the performance of graph algorithms executing in networks of spiking artificial neurons. Although spiking neural networks (SNNs) are capable of general-purpose computation, few algorithmic results with rigorous asymptotic performance analysis are known. SNNs are exceptionally well-motivated practically, as neuromorphic computing systems with 100 million spiking neurons are available, and systems with a billion neurons are anticipated in the next few years. Beyond massive parallelism and scalability, neuromorphic computing systems offer energy consumption orders of magnitude lower than conventional high-performance computing systems. We employ our framework to design and analyze new spiking algorithms for shortest path and dynamic programming problems. Our neuromorphic algorithms are message-passing algorithms relying critically on data movement for computation. For fair and rigorous comparison with conventional algorithms and architectures, which is challenging but paramount, we develop new models of data-movement in conventional computing architectures. This allows us to prove polynomial-factor advantages, even when we assume a SNN consisting of a simple grid-like network of neurons. To the best of our knowledge, this is one of the first examples of a rigorous asymptotic computational advantage for neuromorphic computing.
Automated vehicles (AV) hold great promise for improving safety, as well as reducing congestion and emissions. In order to make automated vehicles commercially viable, a reliable and highperformance vehicle-based computing platform that meets ever-increasing computational demands will be key. Given the state of existing digital computing technology, designers will face significant challenges in meeting the needs of highly automated vehicles without exceeding thermal constraints or consuming a large portion of the energy available on vehicles, thus reducing range between charges or refills. The accompanying increases in energy for AV use will place increased demand on energy production and distribution infrastructure, which also motivates increasing computational energy efficiency.
We present a theoretical framework for designing and assessing the performance of algorithms executing in networks consisting of spiking artificial neurons. Although spiking neural networks (SNNs) are capable of general-purpose computation, few algorithmic results with rigorous asymptotic performance analysis are known. SNNs are exceptionally well-motivated practically, as neuromorphic computing systems with 100 million spiking neurons are available, and systems with a billion neurons are anticipated in the next few years. Beyond massive parallelism and scalability, neuromorphic computing systems offer energy consumption orders of magnitude lower than conventional high-performance computing systems. We employ our framework to design and analyze neuromorphic graph algorithms, focusing on shortest path problems. Our neuromorphic algorithms are message-passing algorithms relying critically on data movement for computation, and we develop data-movement lower bounds for conventional algorithms. A fair and rigorous comparison with conventional algorithms and architectures is challenging but paramount. We prove a polynomial-factor advantage even when we assume an SNN consisting of a simple grid-like network of neurons. To the best of our knowledge, this is one of the first examples of a provable asymptotic computational advantage for neuromorphic computing.
n this presentation we will discuss recent results on using the SpiNNaker neuromorphic platform (48-chip model) for deep learning neural network inference. We use the Sandia Labs developed Whet stone spiking deep learning library to train deep multi-layer perceptrons and convolutional neural networks suitable for the spiking substrate on the neural hardware architecture. By using the massively parallel nature of SpiNNaker, we are able to achieve, under certain network topologies, substantial network tiling and consequentially impressive inference throughput. Such high-throughput systems may have eventual application in remote sensing applications where large images need to be chipped, scanned, and processed quickly. Additionally, we explore complex topologies that push the limits of the SpiNNaker routing hardware and investigate how that impacts mapping software-implemented networks to on-hardware instantiations.
Recent advances in neuromorphic algorithm development have shown that neural inspired architectures can efficiently solve scientific computing problems including graph decision problems and partial-integro differential equations (PIDEs). The latter requires the generation of a large number of samples from a stochastic process. While the Monte Carlo approximation of the solution of the PIDEs converges with an increasing number of sampled neuromorphic trajectories, the fidelity of samples from a given stochastic process using neuromorphic hardware requires verification. Such an exercise increases our trust in this emerging hardware and works toward unlocking its energy and scaling efficiency for scientific purposes such as synthetic data generation and stochastic simulation. In this paper, we focus our verification efforts on a one-dimensional Ornstein- Uhlenbeck stochastic differential equation. Using a discrete-time Markov chain approximation, we sample trajectories of the stochastic process across a variety of parameters on an Intel 8- Loihi chip Nahuku neuromorphic platform. Using relative entropy as a verification measure, we demonstrate that the random samples generated on Loihi are, in an average sense, acceptable. Finally, we demonstrate how Loihi's fidelity to the distribution changes as a function of the parameters of the Ornstein- Uhlenbeck equation, highlighting a trade-off between the lower-precision random number generation of the neuromorphic platform and our algorithm's ability to represent a discrete-time Markov chain.
Boolean functions and binary arithmetic operations are central to standard computing paradigms. Accordingly, many advances in computing have focused upon how to make these operations more efficient as well as exploring what they can compute. To best leverage the advantages of novel computing paradigms it is important to consider what unique computing approaches they offer. However, for any special-purpose co-processor, Boolean functions and binary arithmetic operations are useful for, among other things, avoiding unnecessary I/O on-and-off the co-processor by pre- and post-processing data on-device. This is especially true for spiking neuromorphic architectures where these basic operations are not fundamental low-level operations. Instead, these functions require specific implementation. Here we discuss the implications of an advantageous streaming binary encoding method as well as a handful of circuits designed to exactly compute elementary Boolean and binary operations.
The widely parallel, spiking neural networks of neuromorphic processors can enable computationally powerful formulations. While recent interest has focused on primarily machine learning tasks, the space of appropriate applications is wide and continually expanding. Here, we leverage the parallel and event-driven structure to solve a steady state heat equation using a random walk method. The random walk can be executed fully within a spiking neural network using stochastic neuron behavior, and we provide results from both IBM TrueNorth and Intel Loihi implementations. Additionally, we position this algorithm as a potential scalable benchmark for neuromorphic systems.
Deep learning networks have become a vital tool for image and data processing tasks for deployed and edge applications. Resource constraints, particularly low power budgets, have motivated methods and devices for efficient on-edge inference. Two promising methods are reduced precision communication networks (e.g. binary activation spiking neural networks) and weight pruning. In this paper, we provide a preliminary exploration for combining these two methods, specifically in-training weight pruning of whetstone networks, to achieve deep networks with both sparse weights and binary activations.
Parsa, Maryam P.; Schuman, Catherine D.; Date, Prasanna D.; Rose, Derek R.; Kay, Bill K.; Mitchell, J.P.; Young, Steven Y.; Dellana, Ryan A.; Severa, William M.; Potok, Thomas P.; Roy, Kaushik R.
Remote sensing (RS) data collection capabilities are rapidly evolving hyper-spectrally (sensing more spectral bands), hyper-temporally (faster sampling rates) and hyper-spatially (increasing number of smaller pixels). Accordingly, sensor technologies have outpaced transmission capa- bilities introducing a need to process more data at the sensor. While many sophisticated data processing capabilities are emerging, power and other hardware requirements for these approaches on conventional electronic systems place them out of context for resource constrained operational environments. To address these limitations, in this research effort we have investigated and char- acterized neural-inspired architectures to determine suitability for implementing RS algorithms In doing so, we have been able to highlight a 100x performance per watt improvement using neu- romorphic computing as well as developed an algorithmic architecture co-design and exploration capability.
Neuromorphic hardware architectures represent a growing family of potential post-Moore's Law Era platforms. Largely due to event-driving processing inspired by the human brain, these computer platforms can offer significant energy benefits compared to traditional von Neumann processors. Unfortunately there still remains considerable difficulty in successfully programming, configuring and deploying neuromorphic systems. We present the Fugu framework as an answer to this need. Rather than necessitating a developer attain intricate knowledge of how to program and exploit spiking neural dynamics to utilize the potential benefits of neuromorphic computing, Fugu is designed to provide a higher level abstraction as a hardware-independent mechanism for linking a variety of scalable spiking neural algorithms from a variety of sources. Individual kernels linked together provide sophisticated processing through compositionality. Fugu is intended to be suitable for a wide-range of neuromorphic applications, including machine learning, scientific computing, and more brain-inspired neural algorithms. Ultimately, we hope the community adopts this and other open standardization attempts allowing for free exchange and easy implementations of the ever-growing list of spiking neural algorithms.
With the advent of large-scale neuromorphic platforms, we seek to better understand the applications of neuromorphic computing to more general-purpose computing domains. Graph analysis problems have grown increasingly relevant in the wake of readily available massive data. We demonstrate that a broad class of combinatorial and graph problems known as dynamic programs enjoy simple and efficient neuromorphic implementations, by developing a general technique to convert dynamic programs to spiking neuromorphic algorithms. Dynamic programs have been studied for over 50 years and have dozens of applications across many fields.
Technological advances have enabled exponential growth in both sensor data collection, as well as computational processing. However, as a limiting factor, the transmission bandwidth in between a space-based sensor and a ground station processing center has not seen the same growth. A resolution to this bandwidth limitation is to move the processing to the sensor, but doing so faces size, weight, and power operational constraints. Different physical constraints on processor manufacturing are spurring a resurgence in neuromorphic approaches amenable to the space-based operational environment. Here we describe historical trends in computer architecture and the implications for neuromorphic computing, as well as give an overview of how remote sensing applications may be impacted by this emerging direction for computing.
The insect brain is a great model system for low power electronics: Insects carry out multisensory integration and are able to change the way the process information, learn, and adapt to changes in their environment with a very limited power budget. This context-dependent processing allows them to implement multiple functionalities within the same network, as well as to minimize power consumption by having context-dependent gains in their first layers of input processing. The combination of low power consumption, adaptability and online learning, and robustness makes them particularly appealing for a number of space applications, from rovers and probes to satellites, all having to deal with the progressive degradation of their capabilities in remote environments. In this work, we explore architectures inspired in the insect brain capable of context-dependent processing and learning. Starting from algorithms, we have explored three different implementations: A spiking implementation in a neuromorphic chip, a custom implementation in an FPGA, and finally hybrid analog/digital implementations based on cross-bar arrays. For the latter, we found that the development of novel resistive materials is crucial in order to enhance the energy efficiency of analog devices while maintaining an adequate footprint. Metal-oxide nanocomposite materials, fabricated using ALD with processes compatible with semiconductor processing, are promising candidates to fill in that role.
With the successes deep neural networks have achieved across a range of applications, researchers have been exploring computational architectures to more efficiently execute their operation. In addition to the prevalent role of graphics processing units (GPUs), many accelerator architectures have emerged. Neuromorphic is one such particular approach which takes inspiration from the brain to guide the computational principles of the architecture including varying levels of biological realism. In this paper we present results on using the SpiNNaker neuromorphic platform (48-chip model) for deep learning neural network inference. We use the Sandia National Laboratories developed Whetstone spiking deep learning library to train deep multi-layer perceptrons and convolutional neural networks suitable for the spiking substrate on the neural hardware architecture. By using the massively parallel nature of SpiNNaker, we are able to achieve, under certain network topologies, substantial network tiling and consequentially impressive inference throughput. Such high-throughput systems may have eventual application in remote sensing applications where large images need to be chipped, scanned, and processed quickly. Additionally, we explore complex topologies that push the limits of the SpiNNaker routing hardware and investigate how that impacts mapping software-implemented networks to on-hardware instantiations.
Krichmar, Jeffrey L.; Severa, William M.; Khan, Muhammad S.; Olds, James L.
The Artificial Intelligence (AI) revolution foretold of during the 1960s is well underway in the second decade of the twenty first century. Its period of phenomenal growth likely lies ahead. AI-operated machines and technologies will extend the reach of Homo sapiens far beyond the biological constraints imposed by evolution: outwards further into deep space, as well as inwards into the nano-world of DNA sequences and relevant medical applications. And yet, we believe, there are crucial lessons that biology can offer that will enable a prosperous future for AI. For machines in general, and for AI's especially, operating over extended periods or in extreme environments will require energy usage orders of magnitudes more efficient than exists today. In many operational environments, energy sources will be constrained. The AI's design and function may be dependent upon the type of energy source, as well as its availability and accessibility. Any plans for AI devices operating in a challenging environment must begin with the question of how they are powered, where fuel is located, how energy is stored and made available to the machine, and how long the machine can operate on specific energy units. While one of the key advantages of AI use is to reduce the dimensionality of a complex problem, the fact remains that some energy is required for functionality. Hence, the materials and technologies that provide the needed energy represent a critical challenge toward future use scenarios of AI and should be integrated into their design. Here we look to the brain and other aspects of biology as inspiration for Biomimetic Research for Energy-efficient AI Designs (BREAD).
The random walk is a fundamental stochastic process that underlies many numerical tasks in scientific computing applications. We consider here two neural algorithms that can be used to efficiently implement random walks on spiking neuromorphic hardware. The first method tracks the positions of individual walkers independently by using a modular code inspired by the grid cell spatial representation in the brain. The second method tracks the densities of random walkers at each spatial location directly. We analyze the scaling complexity of each of these methods and illustrate their ability to model random walkers under different probabilistic conditions.
We present a preliminary investigation of the use of Multi-Layer Perceptrons (MLP) and Recurrent Neural Networks (RNNs) as surrogates of parameter-to-prediction maps of com- putational expensive dynamical models. In particular, we target the approximation of Quan- tities of Interest (QoIs) derived from the solution of a Partial Differential Equations (PDEs) at different time instants. In order to limit the scope of our study while targeting a rele- vant application, we focus on the problem of computing variations in the ice sheets mass (our QoI), which is a proxy for global mean sea-level changes. We present a number of neural network formulations and compare their performance with that of Polynomial Chaos Expansions (PCE) constructed on the same data.
The rise of low-power neuromorphic hardware has the potential to change high-performance computing; however much of the focus on brain-inspired hardware has been on machine learning applications. A low-power solution for solving partial differential equations could radically change how we approach large-scale computing in the future. The random walk is a fundamental stochastic process that underlies many numerical tasks in scientific computing applications. We consider here two neural algorithms that can be used to efficiently implement random walks on spiking neuromorphic hardware. The first method tracks the positions of individual walkers independently by using a modular code inspired by grid cells in the brain. The second method tracks the densities of random walkers at each spatial location directly. We present the scaling complexity of each of these methods and illustrate their ability to model random walkers under different probabilistic conditions. Finally, we present implementations of these algorithms on neuromorphic hardware.
Unlike general purpose computer architectures that are comprised of complex processor cores and sequential computation, the brain is innately parallel and contains highly complex connections between computational units (neurons). Key to the architecture of the brain is a functionality enabled by the combined effect of spiking communication and sparse connectivity with unique variable efficacies and temporal latencies. Utilizing these neuroscience principles, we have developed the Spiking Temporal Processing Unit (STPU) architecture which is well-suited for areas such as pattern recognition and natural language processing. In this paper, we formally describe the STPU, implement the STPU on a field programmable gate array, and show measured performance data.
Neural computing has been identified as a computing alternative in the post-Moore's Law era, however much of its attention has been directed at specialized applications such as machine learning. For scientific computing applications, particularly those that often depend on supercomputing, it is not clear that neural machine learning is the exclusive contribution to be made by neuromorphic platforms. In our presentation, we will discuss ways that looking to the brain as a whole and neurons specifically can provide new sources for inspiration for computing beyond current machine learning applications. Particularly for scientific computing, where approximate methods for computation introduce additional challenges, the development of non-approximate methods for neural computation is potentially quite valuable. In addition, the brain's dramatic ability to utilize context at many different scales and incorporate information from many different modalities is a capability currently poorly realized by neural machine learning approaches yet offers considerable potential impact on scientific applications.
Neural machine learning methods, such as deep neural networks (DNN), have achieved remarkable success in a number of complex data processing tasks. These methods have arguably had their strongest impact on tasks such as image and audio processing - data processing domains in which humans have long held clear advantages over conventional algorithms. In contrast to biological neural systems, which are capable of learning continuously, deep artificial networks have a limited ability for incorporating new information in an already trained network. As a result, methods for continuous learning are potentially highly impactful in enabling the application of deep networks to dynamic data sets. Here, inspired by the process of adult neurogenesis in the hippocampus, we explore the potential for adding new neurons to deep layers of artificial neural networks in order to facilitate their acquisition of novel information while preserving previously trained data representations. Our results on the MNIST handwritten digit dataset and the NIST SD 19 dataset, which includes lower and upper case letters and digits, demonstrate that neurogenesis is well suited for addressing the stability-plasticity dilemma that has long challenged adaptive machine learning algorithms.
The dentate gyrus forms a critical link between the entorhinal cortex and CA3 by providing a sparse version of the signal. Concurrent with this increase in sparsity, a widely accepted theory suggests the dentate gyrus performs pattern separation-similar inputs yield decorrelated outputs. Although an active region of study and theory, few logically rigorous arguments detail the dentate gyrus's (DG) coding.We suggest a theoretically tractable, combinatorial model for this action. The model provides formal methods for a highly redundant, arbitrarily sparse, and decorrelated output signal. To explore the value of this model framework, we assess how suitable it is for two notable aspects of DG coding: how it can handle the highly structured grid cell representation in the input entorhinal cortex region and the presence of adult neurogenesis, which has been proposed to produce a heterogeneous code in the DG.We find tailoring themodel to grid cell input yields expansion parameters consistent with the literature. In addition, the heterogeneous coding reflects activity gradation observed experimentally. Finally,we connect this approach with more conventional binary threshold neural circuit models via a formal embedding.
For decades, neural networks have shown promise for next-generation computing, and recent breakthroughs in machine learning techniques, such as deep neural networks, have provided state-of-the-art solutions for inference problems. However, these networks require thousands of training processes and are poorly suited for the precise computations required in scientific or similar arenas. The emergence of dedicated spiking neuromorphic hardware creates a powerful computational paradigm which can be leveraged towards these exact scientific or otherwise objective computing tasks. We forego any learning process and instead construct the network graph by hand. In turn, the networks produce guaranteed success often with easily computable complexity. We demonstrate a number of algorithms exemplifying concepts central to spiking networks including spike timing and synaptic delay. We also discuss the application of cross-correlation particle image velocimetry and provide two spiking algorithms; one uses time-division multiplexing, and the other runs in constant time.