Biologically inspired heterogeneous learning for accurate, efficient and low-latency neural network
2024

Biologically Inspired Learning for Neural Networks

publication Evidence: high

Author Information

Author(s): Wang Bo, Zhang Yuxuan, Li Hongjue, Dou Hongkun, Guo Yuchen, Deng Yue

Primary Institution: School of Astronautics, Beihang University, Beijing, China

Hypothesis

Can a biologically inspired spiking neural network (SNN) improve accuracy, efficiency, and latency in AI tasks?

Conclusion

The proposed Heterogeneous spIking Framework with self-Inhibiting neurons (HIFI) significantly enhances learning capabilities and outperforms traditional models in various AI tasks.

Supporting Evidence

  • The HIFI model showed 1%–10% improvement in accuracy over traditional SNNs.
  • HIFI achieved a maximum 17.83-fold reduction in energy consumption.
  • HIFI demonstrated a maximum 5-fold improvement in latency for AI tasks.
  • The model successfully identified rare cell types from scRNA-seq data.
  • HIFI outperformed other SNNs in various computer vision benchmarks.

Takeaway

This study shows that by mimicking how real brain cells work, we can make computer programs that learn better and faster.

Methodology

The study developed a spiking neural network model that incorporates self-inhibiting neurons and uses a bi-level programming approach for learning.

Limitations

The model's computational complexity may limit its use in resource-constrained environments.

Digital Object Identifier (DOI)

10.1093/nsr/nwae301

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication