James B. Aimone

Dhireesha Kudithipudi, Anurag Daram, Abdullah M. Zyarah et al.

Lifelong learning - an agent's ability to learn throughout its lifetime - is a hallmark of biological learning systems and a central challenge for artificial intelligence (AI). The development of lifelong learning algorithms could lead to a range of novel AI applications, but this will also require the development of appropriate hardware accelerators, particularly if the models are to be deployed on edge platforms, which have strict size, weight, and power constraints. Here, we explore the design of lifelong learning AI accelerators that are intended for deployment in untethered environments. We identify key desirable capabilities for lifelong learning accelerators and highlight metrics to evaluate such accelerators. We then discuss current edge AI accelerators and explore the future design of lifelong learning accelerators, considering the role that different emerging technologies could play.

James B. Aimone, William Severa, J. Darby Smith

Probabilistic artificial neural networks offer intriguing prospects for enabling the uncertainty of artificial intelligence methods to be described explicitly in their function; however, the development of techniques that quantify uncertainty by well-understood methods such as Monte Carlo sampling has been limited by the high costs of stochastic sampling on deterministic computing hardware. Emerging computing systems that are amenable to hardware-level probabilistic computing, such as those that leverage stochastic devices, may make probabilistic neural networks more feasible in the not-too-distant future. This paper describes the scANN technique -- \textit{sampling (by coinflips) artificial neural networks} -- which enables neural networks to be sampled directly by treating the weights as Bernoulli coin flips. This method is natively well suited for probabilistic computing techniques that focus on tunable stochastic devices, nearly matches fully deterministic performance while also describing the uncertainty of correct and incorrect neural network outputs.

Bradley H. Theilman, James B. Aimone

The potential for neuromorphic computing to provide intrinsic fault tolerance has long been speculated, but the brain's robustness in neuromorphic applications has yet to be demonstrated. Here, we show that a previously described, natively spiking neuromorphic algorithm for solving partial differential equations is intrinsically tolerant to structural perturbations in the form of ablated neurons and dropped spikes. The tolerance band for these perturbations is large: we find that as many as 32 percent of the neurons and up to 90 percent of the spikes may be entirely dropped before a significant degradation in the accuracy results. Furthermore, this robustness is tunable through structural hyperparameters. This work demonstrates that the specific brain-like inspiration behind the algorithm contributes to a significant degree of robustness expected from brain-like neuromorphic algorithms.

Karan P. Patel, Andrew Maicke, Jared Arzate et al.

Novel devices and novel computing paradigms are key for energy efficient, performant future computing systems. However, designing devices for new applications is often time consuming and tedious. Here, we investigate the design and optimization of spin orbit torque and spin transfer torque magnetic tunnel junction models as the probabilistic devices for true random number generation. We leverage reinforcement learning and evolutionary optimization to vary key device and material properties of the various device models for stochastic operation. Our AI guided codesign methods generated different candidate devices capable of generating stochastic samples for a desired probability distribution, while also minimizing energy usage for the devices.

Bradley H. Theilman, James B. Aimone

We demonstrate that scalable neuromorphic hardware can implement the finite element method, which is a critical numerical method for engineering and scientific discovery. Our approach maps the sparse interactions between neighboring finite elements to small populations of neurons that dynamically update according to the governing physics of a desired problem description. We show that for the Poisson equation, which describes many physical systems such as gravitational and electrostatic fields, this cortical-inspired neural circuit can achieve comparable levels of numerical accuracy and scaling while enabling the use of inherently parallel and energy-efficient neuromorphic hardware. We demonstrate that this approach can be used on the Intel Loihi 2 platform and illustrate how this approach can be extended to nontrivial mesh geometries and dynamics.

Craig M. Vineyard, William M. Severa, James B. Aimone

Co-design is a prominent topic presently in computing, speaking to the mutual benefit of coordinating design choices of several layers in the technology stack. For example, this may be designing algorithms which can most efficiently take advantage of the acceleration properties of a given architecture, while simultaneously designing the hardware to support the structural needs of a class of computation. The implications of these design decisions are influential enough to be deemed a lottery, enabling an idea to win out over others irrespective of the individual merits. Coordination is a well studied topic in the mathematics of game theory, where in many cases without a coordination mechanism the outcome is sub-optimal. Here we consider what insights game theoretic analysis can offer for computer architecture co-design. In particular, we consider the interplay between algorithm and architecture advances in the field of neuromorphic computing. Analyzing developments of spiking neural network algorithms and neuromorphic hardware as a co-design game we use the Stag Hunt model to illustrate challenges for spiking algorithms or architectures to advance the field independently and advocate for a strategic pursuit to advance neuromorphic computing.

J. Darby Smith, Aaron J. Hill, Leah E. Reeder et al.

Computing stands to be radically improved by neuromorphic computing (NMC) approaches inspired by the brain's incredible efficiency and capabilities. Most NMC research, which aims to replicate the brain's computational structure and architecture in man-made hardware, has focused on artificial intelligence; however, less explored is whether this brain-inspired hardware can provide value beyond cognitive tasks. We demonstrate that high-degree parallelism and configurability of spiking neuromorphic architectures makes them well-suited to implement random walks via discrete time Markov chains. Such random walks are useful in Monte Carlo methods, which represent a fundamental computational tool for solving a wide range of numerical computing tasks. Additionally, we show how the mathematical basis for a probabilistic solution involving a class of stochastic differential equations can leverage those simulations to provide solutions for a range of broadly applicable computational tasks. Despite being in an early development stage, we find that NMC platforms, at a sufficient scale, can drastically reduce the energy demands of high-performance computing (HPC) platforms.

Ojas Parekh, Cynthia A. Phillips, Conrad D. James et al.

Boolean circuits of McCulloch-Pitts threshold gates are a classic model of neural computation studied heavily in the late 20th century as a model of general computation. Recent advances in large-scale neural computing hardware has made their practical implementation a near-term possibility. We describe a theoretical approach for multiplying two $N$ by $N$ matrices that integrates threshold gate logic with conventional fast matrix multiplication algorithms, that perform $O(N^ω)$ arithmetic operations for a positive constant $ω< 3$. Our approach converts such a fast matrix multiplication algorithm into a constant-depth threshold circuit with approximately $O(N^ω)$ gates. Prior to our work, it was not known whether the $Θ(N^3)$-gate barrier for matrix multiplication was surmountable by constant-depth threshold circuits. Dense matrix multiplication is a core operation in convolutional neural network training. Performing this work on a neural architecture instead of off-loading it to a GPU may be an appealing option.

J. Darby Smith, William Severa, Aaron J. Hill et al.

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.

William Severa, Craig M. Vineyard, Ryan Dellana et al.

This paper presents a new technique for training networks for low-precision communication. Targeting minimal communication between nodes not only enables the use of emerging spiking neuromorphic platforms, but may additionally streamline processing conventionally. Low-power and embedded neuromorphic processors potentially offer dramatic performance-per-Watt improvements over traditional von Neumann processors, however programming these brain-inspired platforms generally requires platform-specific expertise which limits their applicability. To date, the majority of artificial neural networks have not operated using discrete spike-like communication. We present a method for training deep spiking neural networks using an iterative modification of the backpropagation optimization algorithm. This method, which we call Whetstone, effectively and reliably configures a network for a spiking hardware target with little, if any, loss in performance. Whetstone networks use single time step binary communication and do not require a rate code or other spike-based coding scheme, thus producing networks comparable in timing and size to conventional ANNs, albeit with binarized communication. We demonstrate Whetstone on a number of image classification networks, describing how the sharpening process interacts with different training optimizers and changes the distribution of activity within the network. We further note that Whetstone is compatible with several non-classification neural network applications, such as autoencoders and semantic segmentation. Whetstone is widely extendable and currently implemented using custom activation functions within the Keras wrapper to the popular TensorFlow machine learning framework.

Aleksandra Faust, James B. Aimone, Conrad D. James et al.

Robots and autonomous agents often complete goal-based tasks with limited resources, relying on imperfect models and sensor measurements. In particular, reinforcement learning (RL) and feedback control can be used to help a robot achieve a goal. Taking advantage of this body of work, this paper formulates general computation as a feedback-control problem, which allows the agent to autonomously overcome some limitations of standard procedural language programming: resilience to errors and early program termination. Our formulation considers computation to be trajectory generation in the program's variable space. The computing then becomes a sequential decision making problem, solved with reinforcement learning (RL), and analyzed with Lyapunov stability theory to assess the agent's resilience and progression to the goal. We do this through a case study on a quintessential computer science problem, array sorting. Evaluations show that our RL sorting agent makes steady progress to an asymptotically stable goal, is resilient to faulty components, and performs less array manipulations than traditional Quicksort and Bubble sort.

William Severa, Rich Lehoucq, Ojas Parekh et al.

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.

Justin E. Doak, Joe B. Ingram, Sam A. Mulder et al.

A forensics investigation after a breach often uncovers network and host indicators of compromise (IOCs) that can be deployed to sensors to allow early detection of the adversary in the future. Over time, the adversary will change tactics, techniques, and procedures (TTPs), which will also change the data generated. If the IOCs are not kept up-to-date with the adversary's new TTPs, the adversary will no longer be detected once all of the IOCs become invalid. Tracking the Known (TTK) is the problem of keeping IOCs, in this case regular expressions (regexes), up-to-date with a dynamic adversary. Our framework solves the TTK problem in an automated, cyclic fashion to bracket a previously discovered adversary. This tracking is accomplished through a data-driven approach of self-adapting a given model based on its own detection capabilities. In our initial experiments, we found that the true positive rate (TPR) of the adaptive solution degrades much less significantly over time than the naive solution, suggesting that self-updating the model allows the continued detection of positives (i.e., adversaries). The cost for this performance is in the false positive rate (FPR), which increases over time for the adaptive solution, but remains constant for the naive solution. However, the difference in overall detection performance, as measured by the area under the curve (AUC), between the two methods is negligible. This result suggests that self-updating the model over time should be done in practice to continue to detect known, evolving adversaries.

James B. Aimone, William M. Severa

Complex architectures of biological neural circuits, such as parallel processing pathways, has been behaviorally implicated in many cognitive studies. However, the theoretical consequences of circuit complexity on neural computation have only been explored in limited cases. Here, we introduce a mechanism by which direct and indirect pathways from cortex to the CA3 region of the hippocampus can balance both contextual gating of memory formation and driving network activity. We implement this concept in a deep artificial neural network by enabling a context-sensitive bias. The motivation for this is to improve performance of a size-constrained network. Using direct knowledge of the superclass information in the CIFAR-100 and Fashion-MNIST datasets, we show a dramatic increase in performance without an increase in network size.

Michael R. Smith, Joe B. Ingram, Christopher C. Lamb et al.

It is needed to ensure the integrity of systems that process sensitive information and control many aspects of everyday life. We examine the use of machine learning algorithms to detect malware using the system calls generated by executables-alleviating attempts at obfuscation as the behavior is monitored rather than the bytes of an executable. We examine several machine learning techniques for detecting malware including random forests, deep learning techniques, and liquid state machines. The experiments examine the effects of concept drift on each algorithm to understand how well the algorithms generalize to novel malware samples by testing them on data that was collected after the training data. The results suggest that each of the examined machine learning algorithms is a viable solution to detect malware-achieving between 90% and 95% class-averaged accuracy (CAA). In real-world scenarios, the performance evaluation on an operational network may not match the performance achieved in training. Namely, the CAA may be about the same, but the values for precision and recall over the malware can change significantly. We structure experiments to highlight these caveats and offer insights into expected performance in operational environments. In addition, we use the induced models to gain a better understanding about what differentiates the malware samples from the goodware, which can further be used as a forensics tool to understand what the malware (or goodware) was doing to provide directions for investigation and remediation.

William M. Severa, Jerilyn A. Timlin, Suraj Kholwadwala et al.

The high dimensionality of hyperspectral imaging forces unique challenges in scope, size and processing requirements. Motivated by the potential for an in-the-field cell sorting detector, we examine a $\textit{Synechocystis sp.}$ PCC 6803 dataset wherein cells are grown alternatively in nitrogen rich or deplete cultures. We use deep learning techniques to both successfully classify cells and generate a mask segmenting the cells/condition from the background. Further, we use the classification accuracy to guide a data-driven, iterative feature selection method, allowing the design neural networks requiring 90% fewer input features with little accuracy degradation.

James B. Aimone

Although the brain has long been considered a potential inspiration for future computing, Moore's Law - the scaling property that has seen revolutions in technologies ranging from supercomputers to smart phones - has largely been driven by advances in materials science. As the ability to miniaturize transistors is coming to an end, there is increasing attention on new approaches to computation, including renewed enthusiasm around the potential of neural computation. This paper describes how recent advances in neurotechnologies, many of which have been aided by computing's rapid progression over recent decades, are now reigniting this opportunity to bring neural computation insights into broader computing applications. As we understand more about the brain, our ability to motivate new computing paradigms with continue to progress. These new approaches to computing, which we are already seeing in techniques such as deep learning and neuromorphic hardware, will themselves improve our ability to learn about the brain and accordingly can be projected to give rise to even further insights. This paper will describe how this positive feedback has the potential to change the complexion of how computing sciences and neurosciences interact, and suggests that the next form of exponential scaling in computing may emerge from our progressive understanding of the brain.

Michael R. Smith, Aaron J. Hill, Kristofor D. Carlson et al.

Information in neural networks is represented as weighted connections, or synapses, between neurons. This poses a problem as the primary computational bottleneck for neural networks is the vector-matrix multiply when inputs are multiplied by the neural network weights. Conventional processing architectures are not well suited for simulating neural networks, often requiring large amounts of energy and time. Additionally, synapses in biological neural networks are not binary connections, but exhibit a nonlinear response function as neurotransmitters are emitted and diffuse between neurons. Inspired by neuroscience principles, we present a digital neuromorphic architecture, the Spiking Temporal Processing Unit (STPU), capable of modeling arbitrary complex synaptic response functions without requiring additional hardware components. We consider the paradigm of spiking neurons with temporally coded information as opposed to non-spiking rate coded neurons used in most neural networks. In this paradigm we examine liquid state machines applied to speech recognition and show how a liquid state machine with temporal dynamics maps onto the STPU-demonstrating the flexibility and efficiency of the STPU for instantiating neural algorithms.

Timothy J. Draelos, Nadine E. Miner, Christopher C. Lamb et al.

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.