Neuromorphic Systems for Next-Gen AI: A Comprehensive Review
Keywords:
Brain-inspired architecture, Event-driven processing, Neuromorphic computing, Neuromorphic hardware, Next-generation AI, Spiking Neural Networks (SNNs)Abstract
The idea of neuromorphic systems is transforming the research of artificial intelligence because of its similarity to the brain in terms of architecture and behaviour. These systems are constructed with Spiking Neural Networks (SNNs), that recreate the mechanism of communication between biological neurons implemented by electrical impulses, or spikes. Neuromorphic systems have a union of memory and processing systems, unlike conventional von Neumann systems in which memory and processing are discrete; this allows the neuromorphic systems to realize real-time learning, flexibility, and massively parallel operations. This implies that they are switched on only when necessary, which significantly drops the power levels and latency. Due to these strengths, neuromorphic computing can perfectly fulfill the needs of the next generation AI systems, especially edge devices, where the speed and energy efficiency are paramount. These are robotics, smart sensors, and autonomous vehicles that have to process information and make decisions in real-time without cloud servers. Intel Loihi, IBM TrueNorth, and SpiNNaker systems have shown outstanding potential in real-time pattern recognition, sensory processing and adaptive learning, and with only a fraction of the power that GPUs demand. The paper also explores the implementations available in the current research advancements, challenges and the statistical comparisons with conventional systems and the potential of the neuromorphic platforms in edge computing, robotics, smart sensors and autonomous systems. In general, neuromorphic computing is an encouraging direction in developing future intelligent, efficient, and scalable AI.
References
I. G, L. C S, and D. T., Neuromorphic systems. Elsevier/Academic Press, 2008.
L. Fernando, N. D. Truong, J. Eshraghian, “Neuromorphic Neuromodulation: Towards the next generation of on-device AI-revolution in electroceuticals,” Arxiv.org, 2023. https://arxiv.org/abs/2307.12471
M. Davies, N. Srinivasa, T. Lin, “Loihi: A Neuromorphic Manycore Processor with On-Chip Learning,” IEEE Micro, vol. 38, no. 1, pp. 82–99, Jan. 2018, doi: https://doi.org/10.1109/mm.2018.112130359.
P. A. Merolla, J. Arthur, R. Alvarez-Icaza, A. S. Cassidy, “Artificial brains. A million spiking-neuron integrated circuit with a scalable communication network and interface,” Science (New York, N.Y.), vol. 345, no. 6197, pp. 668–673, Aug. 2014, doi: https://doi.org/10.1126/science.1254642.
S. Furber, “Large-scale neuromorphic computing systems,” Journal of Neural Engineering, vol. 13, no. 5, p. 051001, Aug. 2016, doi: https://doi.org/10.1088/1741-2560/13/5/051001.
G. Indiveri and S.-C. Liu, “Memory and Information Processing in Neuromorphic Systems,” Proceedings of the IEEE, vol. 103, no. 8, pp. 1379–1397, Aug. 2015, doi: https://doi.org/10.1109/jproc.2015.2444094.
Y. Chen, H. Li, C. Wu, “Neuromorphic computing’s yesterday, today, and tomorrow – an evolutionary view,” Integration, vol. 61, pp. 49–61, Mar. 2018, doi: https://doi.org/10.1016/j.vlsi.2017.11.001.
M. Kang, Y. Lee, and M. Park, “Energy Efficiency of Machine Learning in Embedded Systems Using Neuromorphic Hardware,” Electronics, vol. 9, no. 7, p. 1069, Jun. 2020, doi: https://doi.org/10.3390/electronics9071069.
G. Li, D. Deng, H. Tang “Brain-Inspired Computing: A Systematic Survey and Future Trends,” Proceedings of the IEEE, pp. 1–41, Jan. 2024, doi: https://doi.org/10.1109/jproc.2024.3429360.
F. Liao, F. Zhou, and Y. Chai, “Neuromorphic vision sensors: Principle, progress and perspectives,” Journal of Semiconductors, vol. 42, no. 1, p. 013105, Jan. 2021, doi: https://doi.org/10.1088/1674-4926/42/1/013105.
A. Pantazi, B. Rajendran, O. Simeone, and E. Neftci, “Editorial: Neuro-inspired computing for next-gen AI: Computing model, architectures and learning algorithms,” Frontiers in Neuroscience, vol. 16, Jul. 2022, doi: https://doi.org/10.3389/fnins.2022.974627.
D. Ivanov, A. Chezhegov, M. Kiselev, A. Grunin, and D. Larionov, “Neuromorphic artificial intelligence systems,” Frontiers in Neuroscience, vol. 16, Sep. 2022, doi: https://doi.org/10.3389/fnins.2022.959626.
M. D. Pham, A. D’Angiulli, M. M. Dehnavi, and R. Chhabra, “From Brain Models to Robotic Embodied Cognition: How Does Biological Plausibility Inform Neuromorphic Systems?” Brain Sciences, vol. 13, no. 9, p. 1316, Sep. 2023, doi: https://doi.org/10.3390/brainsci13091316.
A. Disney, J. Reynolds, C. D. Schuman, A. Klibisz, A. Young, and J. S. Plank, “DANNA: A neuromorphic software ecosystem,” Biologically Inspired Cognitive Architectures, vol. 17, pp. 49–56, Jul. 2016, doi: https://doi.org/10.1016/j.bica.2016.07.007.
A. A. Firoozi and A. A. Firoozi, “Case Studies and Real-World Applications,” Springer Briefs in Applied Sciences and Technology, pp. 77–82, Jan. 2024, doi: https://doi.org/10.1007/978-3-031-65549-4_8.
J. del Valle, J. G. Ramírez, M. J. Rozenberg, and I. K. Schuller, “Challenges in materials and devices for resistive-switching-based neuromorphic computing,” Journal of Applied Physics, vol. 124, no. 21, p. 211101, Dec. 2018, doi: https://doi.org/10.1063/1.5047800.
