Spiking Neural Networks for Signal Processing

Brain-inspired AI that fires only when needed, reducing power use for always-on sensing.

Spiking Neural Networks for Signal Processing

Overview

Regular neural networks compute constantly, even when nothing important is happening. Spiking neural networks behave more like real neurons: they stay quiet and activate only when meaningful events appear. For beginners, this is like motion-sensor lights compared with lights that are always on. We apply this idea to signal analysis so systems can watch streams in real time while using far less energy. That makes SNNs attractive for edge devices, smart cameras, and IoT sensors that must stay alert without draining batteries.

Real-World Impact

Enables practical always-on AI for low-power edge products.

Technologies & Techniques

Spiking Neural NetworksNeuromorphic ComputingTDR Signal ProcessingTemporal CodingEdge ComputingEvent-Driven Processing

Key Achievements

Up to 1000x lower energy use compared with conventional dense neural processing

Real-time signal processing capability

Demonstrated on Intel Loihi neuromorphic hardware

Captures time-based signal patterns that standard models often miss

Publications

Citation data synced from scholarly cache.

A lightweight hybrid analog-digital spiking neural network for iot

2024 20th International Conference on Distributed Computing in Smart Systems and the Internet of Things (DCOSS-IoT)

2024

Cited by 4

Read Publication →

A Bio-inspired Low-power Hybrid Analog/Digital Spiking Neural Networks for Pervasive Smart Cameras

2024 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops)

2024

Cited by 2

Read Publication →

Want to Discuss This Research?

I'm always excited to discuss technical details, potential applications, or collaborative opportunities.