Low-Power Audio Keyword Spotting using Tsetlin Machines (2021-01-27T00:00:00.000000Z)

TL;DR

A TM-based keyword spotting pipeline is explored to demonstrate low complexity with faster rate of convergence compared to NNs and investigate the scalability with increasing keywords and explore the potential for enabling low-power on-chip KWS.

Abstract

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.

Authors

F. Kawsar

2 papers

Akhil Mathur

2 papers

Jie Lei

1 papers

TL;DR

Abstract

Authors

References56 items

On the Convergence of Tsetlin Machines for the XOR Operator

Low-Latency Asynchronous Logic Design for Inference at the Edge

Adaptive Continuous Feature Binarization for Tsetlin Machines Applied to Forecasting Dengue Incidences in the Philippines

From Arithmetic to Logic based AI: A Comparative Analysis of Neural Networks and Tsetlin Machine

Measuring the Novelty of Natural Language Text Using the Conjunctive Clauses of a Tsetlin Machine Text Classifier

Machine-learning-based prediction models for high-need high-cost patients using nationwide clinical and claims data

Environmental sound classification using a regularized deep convolutional neural network with data augmentation

Learning automata based energy-efficient AI hardware design for IoT applications

Massively Parallel and Asynchronous Tsetlin Machine Architecture Supporting Almost Constant-Time Scaling

A Novel Multi-Step Finite-State Automaton for Arbitrarily Deterministic Tsetlin Machine Learning

Explainability and Dependability Analysis of Learning Automata based AI Hardware

A depthwise separable convolutional neural network for keyword spotting on an embedded system

Heart sound classification based on improved MFCC features and convolutional recurrent neural networks

Edge Machine Learning for AI-Enabled IoT Devices: A Review

Optimizing Convolutional Neural Network Hyperparameters by Enhanced Swarm Intelligence Metaheuristics

14.1 A 510nW 0.41V Low-Memory Low-Computation Keyword-Spotting Chip Using Serial FFT-Based MFCC and Binarized Depthwise Separable Convolutional Neural Network in 28nm CMOS

The regression Tsetlin machine: a novel approach to interpretable nonlinear regression

An Ultra-Low Power Always-On Keyword Spotting Accelerator Using Quantized Convolutional Neural Network and Voltage-Domain Analog Switching Network-Based Approximate Computing

Classification of Environmental Sounds Using Convolutional Neural Network with Bispectral Analysis

Efficient Keyword Spotting through Hardware-Aware Conditional Execution of Deep Neural Networks

Implementation of LSTM Accelerator for Speech Keywords Recognition

A Tsetlin Machine with Multigranular Clauses

Automatic Speech Recognition

18μW SoC for near-microphone Keyword Spotting and Speaker Verification

Edge Intelligence: The Convergence of Humans, Things, and AI

The Convolutional Tsetlin Machine

Multitask Learning of Deep Neural Network-Based Keyword Spotting for IoT Devices

Towards The Internet-of-Smart-Clothing: A Review on IoT Wearables and Garments for Creating Intelligent Connected E-Textiles

DNN based Acoustic Scene Classification using Score Fusion of MFCC and Inverse MFCC

Real-Power Computing

Earables for Personal-Scale Behavior Analytics

Exploring audio and kinetic sensing on earable devices

A 141 UW, 2.46 PJ/Neuron Binarized Convolutional Neural Network Based Self-Learning Speech Recognition Processor in 28NM CMOS

Discriminative Keyword Spotting for limited-data applications

Speech Commands: A Dataset for Limited-Vocabulary Speech Recognition

The Tsetlin Machine - A Game Theoretic Bandit Driven Approach to Optimal Pattern Recognition with Propositional Logic

Hello Edge: Keyword Spotting on Microcontrollers

Music emotion recognition: The combined evidence of MFCC and residual phase

Small-footprint keyword spotting using deep neural networks

Automatic Speech Recognition

Unsupervised speaker segmentation with residual phase and MFCC features

Environmental Sound Recognition With Time–Frequency Audio Features

An Application of Recurrent Neural Networks to Discriminative Keyword Spotting

Quantile based histogram equalization for noise robust large vocabulary speech recognition

Histogram equalization of speech representation for robust speech recognition

Cepstral domain segmental nonlinear feature transformations for robust speech recognition

Signal modeling techniques in speech recognition

Automatic recognition of keywords in unconstrained speech using hidden Markov models

MSP-MFCC: Energy-Efficient MFCC Feature Extraction Method With Mixed-Signal Processing Architecture for Wearable Speech Recognition Applications

Learning Attentive Representations for Environmental Sound Classification

Convolutional neural networks for small-footprint keyword spotting

Feature Extraction and Classification for Automatic Speaker Recognition System – A Review

the TA level

mathematical lemma based analysis of clause convergence using the XOR dataset [43]

Exploiting the natural logic underpinning there are currently ongoing efforts in establishing explainability evaluation and analysis of TMs[17]

implemented with many different programming languages such as, C, C++, C # , Python and Node.js, to name a few

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names