Adaptive Spike-Encoded Hybrid CNN–SNN Framework for Energy-Efficient Facial Emotion Recognition

DOI

10.2991/978-94-6239-727-9_15How to use a DOI?

Keywords

Facial Emotion Recognition; Hybrid CNN-SNN; Adaptive Spike Encoding; PCA

Abstract

A Major element of affective computing, intelligent surveillance, and human–machine interaction is facial emotion analysis. However, the great accuracy of current deep learning-based models, like Convolutional Neural Networks (CNNs), comes with a high computational and energy cost, making them unsuitable for embedded and real-time systems. To remove these problems, this study has proposed an Adaptive Spike-Encoded Hybrid Convolutional–Spiking Neural Network (CNN–SNN) framework for energy-efficient real-time facial emotion identification. The concept combines ResNet-50 for spatial feature extraction with an adaptive spike encoding method that converts features into temporally efficient spike patterns for neuromorphic processing. Principal Component Analysis (PCA) reduces the dimensionality from 2048 to 256, saving significant spatial features by reducing redundant features. The model is trained using the CK + and FER-2013 datasets, achieving an accuracy of about 86% and demonstrating a 40–45% reduction in energy consumption when compared to traditional CNN-based systems. Because the hybrid network maintains a constant computational complexity and enables real-time inference at 20 frames per second, it is suitable for edge AI devices and VLSI-based neuromorphic hardware. This work bridges the gap between realistic hardware efficiency and biologically inspired neural computation to enable intelligent and sustainable emotion identification systems for next-generation edge platforms.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Aakanksha Taliwal
AU  - Ramji Gupta
PY  - 2026
DA  - 2026/07/22
TI  - Adaptive Spike-Encoded Hybrid CNN–SNN Framework for Energy-Efficient Facial Emotion Recognition
BT  - Proceedings of the International Conference on Sustainable Micro-Nano Materials & Innovative Technology (ICSUMMIT 2026)
PB  - Atlantis Press
SP  - 206
EP  - 216
SN  - 2589-4943
UR  - https://doi.org/10.2991/978-94-6239-727-9_15
DO  - 10.2991/978-94-6239-727-9_15
ID  - Taliwal2026
ER  -