| Abstract |
As neural networks gain importance with several successful applications of them, this paper raises the question of how they can be applied in the context of coalition operations. A key challenge in military coalition operations is that of energy and severe bandwidth constraints. We address this challenge by exploring the use of Deep Neural Networks (DNNs) and splitting them across multiple edge nodes. Further, we explore the idea of using spiking neural networks that can lower the energy consumption significantly. Spiking Neural Networks offers a bio- plausible alternative towards achieving energy-efficient 'machine intelligence'. Guided by brain-like spiking computational framework, this interdisciplinary field aims at exploring multi-stack optimization across software and compute, with the promise of enabling intelligence, while reducing the energy requirements of underlying computing platforms. Historically, the field began with a motif of implementing biological neural routines with Silicon circuit primitives. Since then, the field has evolved to encompass hardware implementation of artificial intelligence algorithms with spike based encoding and event-driven representations. In this perspective, we provide an overview of major developments in spiking networks, for both algorithms and hardware. We go beyond retrospective discussions highlighting the fundamentals of learning and hardware fabrics, giving insights on major challenges and possible solutions. We discuss future prospects for neuromorphic computing with emphasis on combined efforts comprising algorithm-hardware co-design and how this applies to a coalition distributed tactical setting. |