Efficient Biosignal Processing Using Hyperdimensional Computing: Network Templates for Combined Learning and Classification of ExG Signals

Abstract

Recognizing the very size of the brain's circuits, hyperdimensional (HD) computing can model neural activity patterns with points in a HD space, that is, with HD vectors. Key examined properties of HD computing include: a versatile set of arithmetic operations on HD vectors, generality, scalability, analyzability, one-shot learning, and energy efficiency. These make it a prime candidate for efficient biosignal processing where signals are noisy and nonstationary, training data sets are not huge, individual variability is significant, and energy efficiency constraints are tight. Purely based on native HD computing operators, we describe a combined method for multiclass learning and classification of various ExG biosignals such as electromyography (EMG), electroencephalography (EEG), and electrocorticography (ECoG). We develop a full set of HD network templates that comprehensively encode body potentials and brain neural activity recorded from different electrodes into a single HD vector without requiring domain expert knowledge or ad-hoc electrode selection process. Such encoded HD vector is processed as a single unit for fast one-shot learning, and robust classification. It can be interpreted to identify the most useful features as well. Compared to state-of-the-art counterparts, HD computing enables online, incremental, and fast learning as it demands less than a third as much training data as well as less preprocessing.

Hardware Optimizations of Dense Binary Hyperdimensional Computing: Rematerialization of Hypervectors, Binarized Bundling, and Combinational Associative Memory

“…We present a generic architecture to project multichannel sensory inputs from original representation to hyperdimensional space, where the arithmetic operations are combined to learn and classify examples. While this paper focuses on EMG signals, other streaming multichannel sensor data such as ECoG [1], EEG [31,33], ExG [30], speech [8,34], smell [7] can be equally applicable.…”

Online Learning and Classification of EMG-Based Gestures on a Parallel Ultra-Low Power Platform Using Hyperdimensional Computing

“…Our approaches for learning and classification combine the low-power and high-performance capabilities of a PULP programmable platform with a novel brain-inspired algorithm [44] that is extremely robust against low signal-tonoise ratio (SNR) and large variability in both data and computing platform. Computational complexity of the proposed HD computing approach scales linearly with the number of electrodes [45] and maintains its accuracy with various types of biosignal acquisitions, while the PULP platform is highly optimized for parallel applications that require extreme energy efficiency. Table I provides a quantitative summary of the state of the art of EMG-based pattern recognition embedded systems, including a comparison of energy per classification and battery duration (assuming a 100mAh battery).…”

“…Based on the use of these MAP operations, an encoder can be designed for various tasks, e.g., EMG [20], [22], [49], EEG [23], [50], ECoG [51], ExG [45], or in general pattern processing [52]. The encoder emits a hypervector representing the event of interest that is then fed into an associative memory (AM) for training and inference.…”

Hyperdimensional Computing-based Multimodality Emotion Recognition with Physiological Signals

“…Other leading properties of HD computing includes robustness, energy efficiency, massively parallel operations, and fast one-shot learning [12]. These make it well-suited for efficient biosignal processing [17], e.g., 2× lower energy at iso-accuracy when compared to a highly-optimized SVM on an ARM Cortex M4 [15]. Larger energy saving is achieved by using emerging 3D nanoscale devices [18], [19].…”

“…Novel brain-inspired computational paradigms that support fast learning could lead the way: hyperdimensional (HD) computing [13]-an emerging computational framework based on computing with random HD vectors-provides energy-efficient, robust, and fast learning [14]- [20]. HD computing demonstrates fast learning in various biosignal processing tasks [14]- [16], each of which operates with a specific type of biosignals (see [17] for an overview). In this paper, we extend HD computing for multimodal sensor fusion from different types of physiological signals.…”