Search papers, labs, and topics across Lattice.
This paper introduces a neuromorphic-inspired classifier based on the Receptron model, which operates effectively within the computational and memory constraints of mid-range microcontroller units (MCUs). By enabling non-linear decision boundaries without the complexity of multi-layer networks, the Receptron facilitates continuous on-device adaptation, making it suitable for dynamic edge environments. Experimental results demonstrate that the Receptron achieves cross-validated accuracies comparable to traditional machine learning methods, highlighting its potential as an interpretable solution for edge intelligence applications.
The Receptron model achieves competitive accuracy on edge devices while sidestepping the complexities of multi-layer networks, making it a game-changer for IoT applications.
The growing demand for intelligent processing at the edge of IoT networks is constrained by the severe computational and memory limitations of microcontroller units, which render impractical conventional deep learning approaches. We propose a neuromorphicinspired classifier based on the Receptron model, a single-unit architecture capable of implementing non-linearly separable decision boundaries, without resorting to multi-layer networks. The model is designed for direct deployment on mid-range MCUs, while supporting continuous on-device adaptation. Experimental evaluation on basic dataset benchmarks yields cross-validated accuracies compatible with standard machine learning method baselines. These results position the Receptron as a viable and interpretable alternative for resource-constrained neuromorphic edge systems operating in dynamic, non-stationary environments.
