Lightweight Neural Architecture Search for Temporal Convolutional Networks at the Edge

Article Engineering, Electrical & Electronic

Manufacturing as a Data-Driven Practice: Methodologies, Technologies, and Tools

Tania Cerquitelli et al.

Summary: The article discusses the need for powerful software architectures and data-driven methodologies to extract valuable knowledge from the increasing amount of data generated in digital shop floor environments. It covers key functional and methodological aspects, as well as technologies and tools, to add intelligence to data-driven services in manufacturing environments. The deployment of these solutions in research project demonstrators shows their ability to reduce manufacturing line interruptions and associated costs.

PROCEEDINGS OF THE IEEE (2021)

Add to Collection

Article Computer Science, Hardware & Architecture

DORY: Automatic End-to-End Deployment of Real-World DNNs on Low-Cost IoT MCUs

Alessio Burrello et al.

Summary: This work introduces an automatic tool DORY for deploying DNNs on low-cost MCUs, aiming to maximize memory utilization and improve performance through additional heuristics. A case study demonstrates that DORY outperforms other devices in the market on GreenWaves Technologies GAP8.

IEEE TRANSACTIONS ON COMPUTERS (2021)

Add to Collection

Proceedings Paper Automation & Control Systems

Pruning In Time (PIT): A Lightweight Network Architecture Optimizer for Temporal Convolutional Networks

Matteo Risso et al.

Summary: The study introduced an automatic dilation optimizer that learns dilation factors and weights on the time axis through weight pruning, reducing model size and latency on hardware targets without compromising accuracy. This method generates a set of Pareto-optimal TCNs, outperforming hand-designed solutions in both size and accuracy.

2021 58TH ACM/IEEE DESIGN AUTOMATION CONFERENCE (DAC) (2021)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

ECG-TCN: Wearable Cardiac Arrhythmia Detection with a Temporal Convolutional Network

Thorir Mar Ingolfsson et al.

Summary: The study introduces a novel TCN model for accurate classification of ECG signals, achieving similar accuracy to existing technology while improving energy efficiency. Results show that the GAP8 implementation on two different development boards is more energy-efficient and faster compared to the ARM Cortex M4F implementation.

2021 IEEE 3RD INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE CIRCUITS AND SYSTEMS (AICAS) (2021)

Add to Collection

Proceedings Paper Engineering, Electrical & Electronic

Robust and Energy-efficient PPG-based Heart-Rate Monitoring

Matteo Risso et al.

Summary: This study proposes the use of hardware-friendly Temporal Convolutional Networks (TCN) for PPG-based heart estimation to achieve robust heart-rate monitoring in the presence of motion artifacts and deploy smaller yet accurate networks on commercial MCUs, showing high accuracy on PPG datasets.

2021 IEEE INTERNATIONAL SYMPOSIUM ON CIRCUITS AND SYSTEMS (ISCAS) (2021)

Add to Collection

Article Engineering, Biomedical