Source link : https://tech365.info/mind-inspired-ai-may-lower-power-use-and-enhance-efficiency/
A convergence unit in retinal topography extracted from Ref [65]. Numbers on this determine are chosen for the aim of illustrating the converging nature of the topography. On common, 50 photoreceptors are linked to a single bipolar cell. Credit score: Neurocomputing (2025). DOI: 10.1016/j.neucom.2025.131740
Synthetic intelligence (AI) may quickly develop into extra energy-efficient and sooner, due to a brand new method developed on the College of Surrey that takes direct inspiration from organic neural networks of the human mind.
In a research revealed in Neurocomputing, researchers from Surrey’s Nature-Impressed Computation and Engineering (NICE) group have proven that mimicking the mind’s sparse and structured neural wiring can considerably enhance the efficiency of synthetic neural networks (ANNs)—utilized in generative AI and different trendy AI fashions akin to ChatGPT—with out sacrificing accuracy.
The strategy, referred to as Topographical Sparse Mapping (TSM), rethinks how AI methods are wired at their most elementary degree. In contrast to standard deep-learning fashions—akin to these used for picture recognition and language processing—which join each neuron in a single layer to all neurons within the subsequent, losing power, TSM connects every neuron solely to close by or associated ones, very similar to how the mind’s visible system organizes data effectively. Via this pure design, the mannequin eliminates the…
—-
Author : tech365
Publish date : 2025-10-30 20:06:00
Copyright for syndicated content belongs to the linked Source.
—-
1 – 2 – 3 – 4 – 5 – 6 – 7 – 8