A Design Strategy for the Efficient Implementation of Random Basis Neural Networks on Resource-Constrained Devices

Ragusa, Edoardo; Gianoglio, Christian; Zunino, Rodolfo; Gastaldo, Paolo

doi:10.1007/s11063-019-10165-y

Cited by 16 publications

(22 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [17], [18], they propose an efficient decomposition method to accelerate the computation of the pseudo-inverse for the hidden layer output matrix. In [19], the properties of random networks and hard-limiter activation functions are exploited to implement ELM on FPGA. For SoC design, [20] efficiently implements online sequential ELM for real-time applications.…”

Section: Related Work For Elm Hardware Implementationmentioning

confidence: 99%

An Arbitrarily Reconfigurable Extreme Learning Machine Inference Engine for Robust ECG Anomaly Detection

Chuang

Chen

et al. 2021

IEEE Open J. Circuits Syst.

View full text Add to dashboard Cite

Extreme learning machine (ELM) has shown to be an effective and low-power approach for real-time electrocardiography (ECG) anomaly detection. However, prior ELM inference chips are noise-prone and lacking in reconfigurability. In this article, we present an arbitrarily reconfigurable ELM inference engine fabricated in 40-nm CMOS technology for robust ECG anomaly detection. By combining Adaptive boosting (Adaboost) and Eigenspace denoising with ELM (AE-ELM), robust classification under noisy conditions is achieved and saves the number of required multiplications by 95.9%. For chip implementation, a reconfigurable VLSI architecture is designed to support arbitrary complexity of AE-ELM, accounting for dynamic change in application requirements. On the other hand, we propose to construct the input weight matrix of ELM as a Bernoulli random matrix, which further reduces the number of multiplications by 55.2%. For real-time detection, parallel computing is exploited to reduce the latency by up to 86.8%. Overall, the 0.21-mm 2 AE-ELM inference engine shows its robustness against noisy signals and achieves 1.83x AEE compared with the state-of-the-art ELM design. INDEX TERMS Adaptive boosting (Adaboost), eigenspace denoising, electrocardiography (ECG) anomaly detection, extreme learning machine (ELM), reconfigurable chip design.

show abstract

Section: Related Work For Elm Hardware Implementationmentioning

confidence: 99%

An Arbitrarily Reconfigurable Extreme Learning Machine Inference Engine for Robust ECG Anomaly Detection

Chuang

Chen

et al. 2021

IEEE Open J. Circuits Syst.

View full text Add to dashboard Cite

show abstract

“…These paradigms features fast learning and an efficient forward phase due to the fact that the hidden parameters are not tuned during the learning process. The literature shows that SLFNs [ 41 , 46 ] offer a viable solution for low-power, resource-constrained digital implementations of the inference function [ 47 , 48 ], even making the on-device support of the training process possible on dedicated systems on chip [ 49 , 50 , 51 , 52 ]. In addition, SLFNs can be straightforwardly extended to Online Sequential Learning [ 53 ].…”

Section: Related Workmentioning

confidence: 99%

Emotion Recognition on Edge Devices: Training and Deployment

Pandelea

Ragusa

Apicella

et al. 2021

Sensors

Self Cite

View full text Add to dashboard Cite

Emotion recognition, among other natural language processing tasks, has greatly benefited from the use of large transformer models. Deploying these models on resource-constrained devices, however, is a major challenge due to their computational cost. In this paper, we show that the combination of large transformers, as high-quality feature extractors, and simple hardware-friendly classifiers based on linear separators can achieve competitive performance while allowing real-time inference and fast training. Various solutions including batch and Online Sequential Learning are analyzed. Additionally, our experiments show that latency and performance can be further improved via dimensionality reduction and pre-training, respectively. The resulting system is implemented on two types of edge device, namely an edge accelerator and two smartphones.

show abstract

“…Within the paradigm of edge computing, the local embedded system just implements the inference function, which receives its parameters from the cloud. The literature provides some examples of hardware-friendly implementations of inference functions [40][41][42] that could be integrated in our low resources system. Such works exploited the paradigm of random basis neural networks to implement an inference function supported by a single hidden-layer feedforward neural network.…”

Section: Iot-based Predictive Systemmentioning

confidence: 99%

Unsupervised Monitoring System for Predictive Maintenance of High Voltage Apparatus

et al. 2020

Self Cite

View full text Add to dashboard Cite

The online monitoring of a high voltage apparatus is a crucial aspect for a predictive maintenance program. Partial discharges (PDs) phenomena affect the insulation system of an electrical machine and—in the long term—can lead to a breakdown, with a consequent, significant economic loss; wind turbines provide an excellent example. Embedded solutions are therefore required to monitor the insulation status. The paper presents an online system that adopts unsupervised methodologies for assessing the condition of the monitored machine in real time. The monitoring process does not rely on any prior knowledge about the apparatus; nonetheless, the method can identify the relevant drifts in the machine status. In addition, the system is specifically designed to run on low-cost embedded devices.

show abstract

A Design Strategy for the Efficient Implementation of Random Basis Neural Networks on Resource-Constrained Devices

Cited by 16 publications

References 25 publications

An Arbitrarily Reconfigurable Extreme Learning Machine Inference Engine for Robust ECG Anomaly Detection

An Arbitrarily Reconfigurable Extreme Learning Machine Inference Engine for Robust ECG Anomaly Detection

Emotion Recognition on Edge Devices: Training and Deployment

Unsupervised Monitoring System for Predictive Maintenance of High Voltage Apparatus

Contact Info

Product

Resources

About