Wed, Jul 30, 2025 page1

• Research simulates human brain learning, memory

  • By Rachel Lin and Jake Chung / Staff reporter, with staff writer
  • • 
    • 
    • 

  Taiwanese researchers have successfully simulated the human brain’s synaptic learning and memory, demonstrating the possibility of simulated synaptic plasticity, which could be used in smart diagnoses, visual modules for self-driving systems, wearable devices or biorobotics.

  The research, which appeared in the May issue of the scientific journal Small, was led by National Yang Ming Chiao Tung University Department of Photonics professor Liu Po-tsun (劉柏村).

  The research cited recent attention from the scientific community regarding the implementation of neuromorphic devices and their ability to revolutionize computational paradigms by emulating the data transfer mechanisms in biological neural systems.

  

  National Yang Ming Chiao Tung University Department of Photonics professor Liu Po-tsun, front left, poses for photographs with team members who worked on optoelectronic synaptic transistor research in an undated photograph.

  Photo courtesy of National Yang Ming Chiao Tung University via CNA

  The emulation of the “human brain’s computing architecture has ... facilitated the development of neurorobotics and artificial sensory systems, such as self-driving systems and medical diagnostics applications,” Liu said.

  The proposed devices successfully emulated key synaptic behaviors, including short-term memory, long-term memory, paired-pulse facilitation and post-tetanic potentiation, he said.

  The transistor demonstrates strong potential for emulating the human visual system, he said.

  A multilayer perceptron model, employing synaptic weights modulated by 650, 525, and 460nm light in the long-term potentiation region and by electrical stimuli in the long-term depression region, achieves a high recognition accuracy of 87.4 pecent for severely distorted handwritten digits, the report said.

  In terms of the device’s capability to recognize handwriting, the research incorporated datasets from the Modified National Institute of Standards and Technology, Liu said.

  The tests were run through RGB schemes and incorporated disturbances, such as Gaussian, pepper and salt, and stripe noise, to “emulate real-world conditions and evaluate devices’ performance under image distortions,” the report said.

  “By leveraging the high linearity achieved through RGB schemes, the results indicate that under Gaussian noise conditions, the recognition accuracy reached optimal levels, achieving 94.2%, 91.2%, and 91.4% for the R, G, and B schemes, respectively,” it said.

  “These findings highlight the robust performance and high accuracy of the proposed WIST synaptic devices in real-world color perception scenarios, demonstrating its feasibility for emulating the human visual system,” it said.

  The tests defined devices incorporating a crystalline WO3 (c-WO3) layer as WIST, and those without a c-WO3 layer as IST.

  The report concluded that the leveraged simulation method demonstrates the high potential of proposed WIST devices in advanced neuromorphic applications and that the WIST device-based model effectively mimics real-world human visual perception systems.
• Most Popular

  • 1

    Singaporean PM quotes Taiwanese song in national speech

  • 2

    Eight Taiwanese restaurants win first Michelin stars

  • 3

    Mbappe lifts Madrid past Osasuna in La Liga opener

  • 4

    Taiwan troops joined military exercise in US

  • 5

    US eyeing TSMC stake, minister says