科学家们建造了一个小型的, 由大脑启发的人工智能模型, 模仿猴子的视觉, 缩小复杂性, 同时增进对大脑如何识别物体的理解。 Scientists built a tiny, brain-inspired AI model that mimics monkey vision, slashing complexity while boosting understanding of how brains recognize objects.

科学家们创建了一个由黑猴子大脑启发的小型AI模型,将复杂的视觉系统从6 000万变数减少到仅1万个,同时保持强劲的性能。 Scientists have created a compact AI model inspired by macaque monkey brains, reducing a complex vision system from 60 million variables to just 10,000 while maintaining strong performance. 模型模拟了V4神经元,能检测颜色、质地和形状,如曲线和模式,帮助解释生物大脑如何有效地识别物体。 The model mimics V4 neurons that detect colors, textures, and shapes like curves and patterns, helping explain how biological brains recognize objects efficiently. 研究人员利用猴子神经数据和压缩技术,建立了一个小型、透明的人工智能系统,该系统小到足以通过电子邮件发送,对视觉处理和潜在应用提供更清晰的洞察力,以了解像阿尔茨海默氏病这样的大脑疾病。 Using monkey neural data and compression techniques, researchers built a small, transparent AI system small enough to send via email, offering clearer insights into visual processing and potential applications in understanding brain disorders like Alzheimer’s. 该模型的效率表明,目前的AI可能受益于更新的神经科学,因为它仍然在与现实世界的识别任务作斗争,而现实世界的识别任务是人类容易处理的。 The model’s efficiency suggests current AI may benefit from updated neuroscience, as it still struggles with real-world recognition tasks that humans handle easily.