Brain-inspired computing (BIC) is an emerging research field that aims to build fundamental theories, models, hardware architectures, and application systems toward more general artificial intelligence (AI) by learning from the information processing mechanisms of biological nervous systems. It is regarded as one of the most promising research directions for future intelligent computing in the post-Moore era. 

In an article published in Proceedings of the IEEE, researchers present a general framework for the real-world applications of BIC systems, which they promise will benefit both AI and brain science. 

BIC Infrastructures

Researchers discuss the concept of BIC, summarizing four components of BIC infrastructure development: 1) modeling/algorithm; 2) hardware platform; 3) software tool; and 4) benchmark data. Taking BIC as the turning point and leveraging the advantages of multidisciplinary cross-integration to promote the common development of various domains and realize more general AI forms a valuable and important topic. 

Four components of BIC infrastructures. The prosperity of BIC depends on the construction of these components.

 

Overview of BIC systems: co-design of four components of BIC becoming increasingly ubiquitous in building an end-to-end BIC system, combining computer science and neuroscience.

 

The researchers are optimistic about the potential of BIC systems, believing that they will spark great interest in the research communities of both AI and brain science, paving the way for exciting future developments. 

Main Concepts of BIC 

To clarify the concept of BIC, the researchers introduce several basic concepts in the related research fields. Specifically, they teach the concepts of ANNs and SNNs, mainly from deep learning and computational neuroscience. The researchers also differentiate between DNN accelerators designed for supporting ANNs and neuromorphic chips for supporting SNNs. Finally, they illustrate the interdisciplinary feature of BIC, pointing out that the main challenge is how BIC models and systems can exploit the advanced achievements of computational neuroscience to achieve more general AI.

Scope of BIC and emerging-related research fields.

 

Neuromorphic Computing uses very large-scale integrated circuit (VLSI) systems to mimic biological functions of the nervous system. In this survey, “neuromorphic” is a subset of Classic BIC because it is more related to hardware, which relies on hardware neurons for computing.

Classic BIC aims to build theories, models, architectures, and hardware systems by learning from biological neural systems’ mechanisms, structures, and functions. In this survey, Classic BIC is a superset of neuromorphic computing in the sense that Classic BIC is not limited to hardware, whose theories, models, architectures, and hardware systems can all be inspired by biological neural systems both behaviorally and physically. 

Brain for AI aims to enhance AI technologies by getting inspiration from 1) the signal transmission and learning rules in the nervous system; 2) the structures and functions of the brain; or 3) the mental or cognitive processes of human beings. Classic BIC is a subset of Brain for AI because the former mainly focuses on the computing/learning capability of the system, while the latter can additionally refer to the general concept related to AI technologies.

AI for Brain can power brain science with the help of AI technologies so that we can: 1) explain complex phenomena in the brain using an AI framework; 2) have a better understanding of the structures and functions of the brain; 3) predict cognition, development, and mental health; or 4) control behaviors and mental processes.

The main challenge is how BIC systems can exploit the advanced achievements of computational neuroscience to bridge the gap between AI and neuroscience. 

Learning Algorithms and Frameworks

Neuroscience has long been an important driver of the progress in AI. The researchers propose that leveraging the neural dynamics inspired by the structure/function of a single neuron or neural system is of great potential. To this end, four key aspects must be considered, including great development potential based on bio-plausibility, development goal oriented to the performance needs of practical application scenarios, the inherent low-energy development advantage brought by sparse spike activities, and the development bottleneck caused by the trainability of large-scale SNNs.

Key considerations in the development of BIC algorithms.

 

Next, the researchers discuss Hardware Platforms, Software Tools, and Benchmark Datasets before reviewing the BIC Systems framework.

Framework for developing a BIC system.

 

Two directions of building BIC systems

 

Discussion and Conclusion

To truly realize the potential of BIC, the entire field needs to make systematic efforts. Although the existing BIC framework stems from learning the brain and seems to share similar terminologies with neuroscience, the learning itself is still limited to phenomenological simulation with computational simplification.

The complex topology and special dynamic phenomena from neural connections may be critical for the brain to coordinate the body’s myriad functions and behaviors and achieve human intelligence. However, BIC researchers still fail to apply them to practical modeling. Accelerating progress in current AI by investing in fundamental research in neural computation has great potential. 

The researchers suggest that BIC may learn from the invalidation of reductionism in neuroscience and develop real multiscale architectures for cross-scale neural dynamics to emerge. They highlight the potential for long-term future research in exploring the ways in which BIC can learn from the brain at a microscopic scale, mesoscopic scale, and macroscopic scale simultaneously, and develop corresponding theories, models, architectures, and hardware systems to deal with real-world applications.

Interested in learning more about Brain Inspired Computing? The IEEE Xplore Digital Library offers over 1,000 publications on Brain Inspired Computing.

Interested in acquiring full-text access to this collection for your entire organization? Request a free demo and trial subscription for your organization.