Conference paper
A Neural Network Engine for Resource Constrained Embedded Systems
This paper introduces a dedicated neural network engine developed for resource constrained embedded devices such as hearing aids. It implements a novel dynamic two-step scaling technique for quantizing the activations in order to minimize word size and thereby memory traffic. This technique requires neither computing a scaling factor during training nor expensive hardware for on-the-fly quantization.
Memory traffic is further reduced by using a 12-element vectorized multiply-accumulate datapath that supports data-reuse. Using a keyword spotting neural network as benchmark, performance of the neural network engine is compared with an implementation on a typical audio digital signal processor used by Demant in some of its hearing instruments.
In general, the neural network engine offers small area as well as low power. It outperforms the digital signal processor and results in significant reduction of, among others, power (5×), memory accesses (5.5×), and memory requirements (3×). Furthermore, the two-step scaling ensures that the engine always executes in a deterministic number of clock cycles for a given neural network.
Language: | English |
---|---|
Publisher: | IEEE |
Year: | 2020 |
Pages: | 125-131 |
Proceedings: | 54<sup>th</sup> Asilomar Conference on Signals, Systems, and Computers |
ISBN: | 0738131261 , 1665447079 , 9780738131269 , 9781665447072 , 0738131245 , 0738131253 , 9780738131245 and 9780738131252 |
ISSN: | 10586393 and 25762303 |
Types: | Conference paper |
DOI: | 10.1109/IEEECONF51394.2020.9443426 |
ORCIDs: | Jelcicova, Zuzana and Sparsø, Jens |
Artificial neural networks Auditory system Benchmark testing Digital signal processors Instruments Memory management Neurons audio digital signal processor digital signal processing chips embedded systems hearing instruments keyword spotting neural network memory traffic neural nets resource constrained embedded systems two-step scaling technique