Low-Power Audio Keyword Spotting using Tsetlin Machines
| Autor: | Tousif Rahman, Jie Lei, Alex Yakovlev, Fahim Kawsar, Akhil Mathur, Ole-Christoffer Granmo, Adrian Wheeldon, Rishad Shafik |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
FOS: Computer and information sciences
speech command Sound (cs.SD) Computer science Speech recognition 02 engineering and technology keyword spotting Machine learning computer.software_genre Computer Science - Sound Reduction (complexity) Audio and Speech Processing (eess.AS) 020204 information systems FOS: Electrical engineering electronic engineering information engineering 0202 electrical engineering electronic engineering information engineering Electrical and Electronic Engineering Artificial neural network Learning automata business.industry learning automata lcsh:Applications of electric power 020206 networking & telecommunications lcsh:TK4001-4102 Pipeline (software) Power (physics) machine learning Tsetlin Machine MFCC Keyword spotting electrical_electronic_engineering Scalability Memory footprint pervasive AI 020201 artificial intelligence & image processing Mel-frequency cepstrum Artificial intelligence business computer artificial neural network Efficient energy use Electrical Engineering and Systems Science - Audio and Speech Processing |
| Zdroj: | Journal of Low Power Electronics and Applications, Vol 11, Iss 18, p 18 (2021) Journal of Low Power Electronics and Applications Volume 11 Issue 2 |
| Popis: | The emergence of Artificial Intelligence (AI) driven Keyword Spotting (KWS) technologies has revolutionized human to machine interaction. Yet, the challenge of end-to-end energy efficiency, memory footprint and system complexity of current Neural Network (NN) powered AI-KWS pipelines has remained ever present. This paper evaluates KWS utilizing a learning automata powered machine learning algorithm called the Tsetlin Machine (TM). Through significant reduction in parameter requirements and choosing logic over arithmetic based processing, the TM offers new opportunities for low-power KWS while maintaining high learning efficacy. In this paper we explore a TM based keyword spotting (KWS) pipeline to demonstrate low complexity with faster rate of convergence compared to NNs. Further, we investigate the scalability with increasing keywords and explore the potential for enabling low-power on-chip KWS. 20 pp |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|