Cyber-Physical Analytics: Environmental Sound Classification at the Edge

@article{Elliott2020CyberPhysicalAE,
  title={Cyber-Physical Analytics: Environmental Sound Classification at the Edge},
  author={David Elliott and Evan Martino and Carlos E. Otero and Anthony O. Smith and Adrian M. Peter and Benjamin Luchterhand and Eric Lam and Steven Leung},
  journal={2020 IEEE 6th World Forum on Internet of Things (WF-IoT)},
  year={2020},
  pages={1-6}
}
Recent advances in embedded system technology have created opportunities for alleviating or eliminating common Big Data problems, by providing the resources necessary to perform AI/ML algorithms on-board edge devices. This has led to the emergence of a sub-discipline of Measurements and Signal Intelligence (MASINT) known as Cyber-Physical MASINT, wherein analysts can receive and exploit data directly from cyber-physical devices, and execute algorithms on-board, without the need to transfer to… 

Figures and Tables from this paper

Tiny Transformers for Environmental Sound Classification at the Edge
TLDR
This work presents training techniques for audio models in the environmental sound classification at the edge, and shows that a BERT-based Transformer, trained on Mel spectrograms, can outperform a CNN using 99.85% fewer parameters.
Quantized Constant-Q Gabor Atoms for Sparse Binary Representations of Cyber-Physical Signatures
TLDR
Binary metrics are proposed for the quantification of cyber-physical signal characteristics and features, and a standardized constant-Q variation of the Gabor atom is developed for use with wavelet transforms.
Constant-Q Gabor Atoms for Sparse Binary Representations of Cyber-Physical Signatures
TLDR
Binary metrics are proposed for the quantification of cyber10 physical signal characteristics and features, and a standardized constant-Q variation of the Gabor atom is developed for use with wavelet transforms.

References

SHOWING 1-10 OF 33 REFERENCES
Edge Intelligence: Challenges and Opportunities of Near-Sensor Machine Learning Applications
TLDR
It is argued that for a wide range of emerging applications edge intelligence is a necessary evolutionary need, and thus a summary of the challenges and opportunities that arise from this need is provided.
A Survey on Network Methodologies for Real-Time Analytics of Massive IoT Data and Open Research Issues
TLDR
The state-of-the-art of the analytics network methodologies, which are suitable for real-time IoT analytics are reviewed, and a number of prospective research problems and future research directions are presented focusing on thenetwork methodologies for the real- time IoT analytics.
Edge Computing with Embedded AI: Thermal Image Analysis for Occupancy Estimation in Intelligent Buildings
TLDR
An embedded algorithm for room occupancy estimation based on a thermal sensor with accuracy over the state-of-the-art is presented, achieving a prediction accuracy of 98.9% and very low latency achieving execution times under 14 ms.
Real-time room occupancy estimation with Bayesian machine learning using a single PIR sensor and microcontroller
This paper presents the implementation and deployment of a compute/memory intensive non-parametric Bayesian machine learning algorithm on a microcontroller unit (MCU) to estimate room occupancy in a
AI Benchmark: Running Deep Neural Networks on Android Smartphones
TLDR
A study of the current state of deep learning in the Android ecosystem and describe available frameworks, programming models and the limitations of running AI on smartphones, as well as an overview of the hardware acceleration resources available on four main mobile chipset platforms.
Squeezing Deep Learning into Mobile and Embedded Devices
TLDR
This department describes phone, watch, and embedded prototypes that can locally run large-scale deep networks processing audio, images, and inertial sensor data and vastly reduce conventional inference-time overhead of deep models.
Sound Event Detection in the DCASE 2017 Challenge
TLDR
Analysis of the systems behavior reveals that task-specific optimization has a big role in producing good performance; however, often this optimization closely follows the ranking metric, and its maximization/minimization does not result in universally good performance.
Understanding and parameter setting of STA/LTA trigger algorithm
TLDR
T triggered seismic data acquisition serves for the detection of typical seismic signals (earth quakes, controlled source seismic signals, underground nuclear explosion signals, etc.) in the constantly present seismic noise signal.
Detection of a Surface Detonated Nuclear Weapon Using a Photovoltaic Rich Distribution Grid
Detonation of a nuclear weapon differs significantly from traditional chemical blasts in a number of well documented ways. However, at time of detonation, it is not easy to detect which type of blast
Rethinking the Inception Architecture for Computer Vision
TLDR
This work is exploring ways to scale up networks in ways that aim at utilizing the added computation as efficiently as possible by suitably factorized convolutions and aggressive regularization.
...
1
2
3
4
...