Chinese Scientists Innovate Brain Like AI network model

Innovative Brain-Inspired Network Model Developed by Chinese Scientists

Chinese scientist have developed an innovative brain-inspired network model grounded in internal complexity, aimed at overcoming the challenges of traditional models, including high computing resource consumption, as reported by the Institute of Automation at the Chinese Academy of Sciences.

The research has been published in Nature Computational Science.

Objective of Current AI Evolution

An essential aim in the current AI evolution is the development of general-purpose AI, expanding model's cognitive range.

Challenges of Traditional Models

Li Guogi from the Institute of Automation stated that the prevailing method in large models to construct increasingly large, deep, and broad neural networks, leveraging the law of scale to achieve general intelligence through 'external complexity.'

This approach, however, is challenged by issues such as excessive computing resource and energy consumption, and a lack of interpretability.

Efficiency of the Human Brain

On the flip side, the human brain is comprised of roughly 100 billion neurons and 1,000 trillion synaptic links, each neuron with a detailed internal architecture, but it functions with an energy consumption of only approximately 20 watts.

Innovative Approach: Internal Complexity

Influenced by the complex behavior of neuronal networks, scientists from the Institute of Automation, alongside researchers from Tsinghua University and Peking University, have used an 'internal complexity' approach to enhance general intelligence.

Results and Implications

The results of their experiments confirmed the robustness and reliability of the internal complexity model in addressing complex tasks, offering new approaches and theoretical insights for merging the dynamics aspects of neuroscience with AI, and providing practical solutions for refining and elevating AI model functionality.

Source

Top Offers of Amazon