Near-channel classifier: symbiotic communication and classification in high-dimensional space

Abstract

Brain-inspired high-dimensional (HD) computing represents and manipulates data using very long, random vectors with dimensionality in the thousands. This representation provides great robustness for various classification tasks where classifiers operate at low signal-to-noise ratio (SNR) conditions. Similarly, hyperdimensional modulation (HDM) leverages the robustness of complex-valued HD representations to reliably transmit information over a wireless channel, achieving a similar SNR gain compared to state-of-the-art codes. Here, we first propose methods to improve HDM in two ways: (1) reducing the complexity of encoding and decoding operations by generating, manipulating, and transmitting bipolar or integer vectors instead of complex vectors; (2) increasing the SNR gain by 0.2 dB using a new soft-feedback decoder; it can also increase the additive superposition capacity of HD vectors up to 1.7$$\times$$ × in noise-free cases. Secondly, we propose to combine encoding/decoding aspects of communication with classification into a single framework by relying on multifaceted HD representations. This leads to a near-channel classification (NCC) approach that avoids transformations between different representations and the overhead of multiple layers of encoding/decoding, hence reducing latency and complexity of a wireless smart distributed system while providing robustness against noise and interference from other nodes. We provide a use-case for wearable hand gesture recognition with 5 classes from 64 EMG sensors, where the encoded vectors are transmitted to a remote node for either performing NCC, or reconstruction of the encoded data. In NCC mode, the original classification accuracy of 94% is maintained, even in the channel at SNR of 0 dB, by transmitting 10,000-bit vectors. We remove the redundancy by reducing the vector dimensionality to 2048-bit that still exhibits a graceful degradation: less than 6% accuracy loss is occurred in the channel at − 5 dB, and with the interference from 6 nodes that simultaneously transmit their encoded vectors. In the reconstruction mode, it improves the mean-squared error by up to 20 dB, compared to standard decoding, when transmitting 2048-dimensional vectors.

Wireless On-Chip Communications for Scalable In-memory Hyperdimensional Computing

“…The M encoders at the left compute the different query hypervectors, which will be bundled later on through the majority operation. Each encoder can encode data from e.g., different sensory modalities [26], [27], or streaming channels [18]. This is highly desirable since by doing a bundling of M queries, we virtually increase the throughput by a factor of M .…”

“…By its very nature, HDC is extremely robust in the presence of failures, defects, variations, and noise, all of which are synonymous to ultra-low energy computation. It has been shown that HDC degrades very gracefully in the presence of various faults compared to baseline classifiers: HDC tolerates intermittent errors [29], permanent hard errors (in memory [30] and logic [31]), and spatio-temporal variations [32] in emerging technologies as well as noise and interference in the communication channels [15], [18]. These demonstrate robust operations of HDC under low signal-tonoise ratio and high variability conditions.…”

“…HDC has been employed in a range of applications such as cognitive computing [6]- [8], robotics [9], distributed com-puting [10]- [12], communications [13]- [18], and in various aspects of machine learning. It has shown significant promise in machine learning applications that especially demand fewshot learning [19]- [23], in-sensor adaptive learning [24], [25], multimodal learning [26], [27], and always-on smart sensing [28].…”

A Survey on Hyperdimensional Computing aka Vector Symbolic Architectures, Part I: Models and Data Transformations

“…Some of these results can be obtained analytically using the capacity theory. Additionally, [Summers-Stay et al, 2018], [Frady et al, 2018b], [Kim, 2018], [Hersche et al, 2021] elaborated on methods for recovering information from compositional HVs beyond the standard nearest neighbor search in the item memory, reaching to the capacity of up to 1.2 bits/component [Hersche et al, 2021]. The works above were focused on the case where a single HV was used to store information but as it was demonstrated in [Danihelka et al, 2016] the decoding from HVs can be improved if the redundant storage is used.…”

WHYPE: A Scale-Out Architecture With Wireless Over-the-Air Majority for Scalable In-Memory Hyperdimensional Computing

“…HDC has been employed in a range of applications including cognitive computing [3], [4], robotics [5], distributed computing [6]- [8], communications [9]- [14], and in various aspects of machine learning. See [15] for a comprehensive review.…”