About this Research Topic
From building smart home systems to searching exoplanets, the rapid development of artificial intelligence (AI) and machine learning (ML) technologies has given rise to great reforms in a broad spectrum of modern society. However, the contemporary AI/ML systems still have significant architectural limitations. Specifically, the hardware infrastructure is limited to large-scale processing systems and computationally expensive training methods that are built upon the von Neumann computing architecture. Beyond that, the demand for precise cognitive functionality operated in real-time manner cannot keep up with growing data density and malicious input data. Modern studies on human brain led to models describing its working mechanisms and computational principles, which in turn can be used as a guideline to develop emerging materials, devices, circuits, and systems emulating brain functionality.
Neuromorphic computing is a neural-like computing architecture towards the next-generation AI/ML systems, replicating the same physics and properties of brains. Benefited by unified memory and in-memory computing technologies, neuromorphic chips have emerged as an innovative solution to accelerate AI/ML applications, which reduces the costs in latency and energy associated with data movement in von Neumann computing architecture. To this end, neuromorphic systems can be developed to carry out real-time information analysis and solve problems with anomalous or malicious sensory inputs. In hardware, emerging materials/devices – such as memristive and spintronic devices – are being investigated to implement electronic neurons and synapses due to their significant properties, including non-volatility, low latency, scalability, and low-power operation. In software, contemporary learning algorithms – such as associative memory and lifelong learning – have become essential for accurate and reliable neural operations. This Research Topic aims to provide a comprehensive overview of recent trends and prospects of neuromorphic systems, including devices/circuits, architectures, and algorithms.
Topics of interest include, but not limited to:
• Emerging materials, devices, circuits, and systems enabling brain-inspired computing paradigms
• Applications of emerging memory technologies and architectures for near-/in-memory processing
• The use of hardware-aware AI/ML model development in neuromorphic systems
• Hardware/software co-design and co-optimization for edge computing in neuromorphic systems
• Brain inspired algorithms, attack/defense techniques, and simulation methodologies for AI/ML models
• Neuromorphic application-driven AI/ML models/inferences (e.g., computer vision, NLP, robotic, etc.)
Neuromorphic computing is a neural-like computing architecture towards the next-generation AI/ML systems, replicating the same physics and properties of brains. Benefited by unified memory and in-memory computing technologies, neuromorphic chips have emerged as an innovative solution to accelerate AI/ML applications, which reduces the costs in latency and energy associated with data movement in von Neumann computing architecture. To this end, neuromorphic systems can be developed to carry out real-time information analysis and solve problems with anomalous or malicious sensory inputs. In hardware, emerging materials/devices – such as memristive and spintronic devices – are being investigated to implement electronic neurons and synapses due to their significant properties, including non-volatility, low latency, scalability, and low-power operation. In software, contemporary learning algorithms – such as associative memory and lifelong learning – have become essential for accurate and reliable neural operations. This Research Topic aims to provide a comprehensive overview of recent trends and prospects of neuromorphic systems, including devices/circuits, architectures, and algorithms.
Topics of interest include, but not limited to:
• Emerging materials, devices, circuits, and systems enabling brain-inspired computing paradigms
• Applications of emerging memory technologies and architectures for near-/in-memory processing
• The use of hardware-aware AI/ML model development in neuromorphic systems
• Hardware/software co-design and co-optimization for edge computing in neuromorphic systems
• Brain inspired algorithms, attack/defense techniques, and simulation methodologies for AI/ML models
• Neuromorphic application-driven AI/ML models/inferences (e.g., computer vision, NLP, robotic, etc.)
Keywords: Neuromorphic Computing, Edge Computing, Neural Information Processing, Near-/In-memory Processing, Large-scale Integrated Circuits and Systems, Smart Sensing
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.
