Towards artificial general intelligence with hybrid Tianjic chip architecture
Towards artificial general intelligence with hybrid Tianjic chip architecture
复制标题
DOI：
10.1038/s41586-019-1424-8
复制
发表时间：
2019-08-01
期刊：
影响因子：
64.8
通讯作者：
Shi, Luping
中科院分区：
文献类型：
--
作者：
Pei, Jing;Deng, Lei;Shi, Luping
开发通用人工智能（AGI）一般有两种方法：面向计算机科学和面向神经科学。由于其表述和编码方案的根本差异，这两种方法依赖于不同且不兼容的平台（2-8），阻碍了AGI的发展。一个通用的平台，可以支持流行的基于计算机科学的人工神经网络，以及神经科学启发的模型和算法是非常可取的。在这里，我们提出了集成了这两种方法的“天机”芯片，以提供一个混合的、协同的平台。“天极”芯片采用多核架构、可重构的构建模块和具有混合编码方案的流线型数据流，不仅可以适应基于计算机科学的机器学习算法，还可以轻松实现大脑启发电路和多种编码方案。我们仅使用一个芯片，演示了在无人驾驶自行车系统中同时处理多种算法和模型，实现实时目标检测、跟踪、语音控制、避障和平衡控制。我们的研究有望通过为更通用的硬件平台铺平道路来刺激AGI的发展。
There are two general approaches to developing artificial general intelligence (AGI) 1: computer-science-oriented and neuroscience-oriented. Because of the fundamental differences in their formulations and coding schemes, these two approaches rely on distinct and incompatible platforms(2-8), retarding the development of AGI. A general platform that could support the prevailing computer-science-based artificial neural networks as well as neuroscience-inspired models and algorithms is highly desirable. Here we present the Tianjic chip, which integrates the two approaches to provide a hybrid, synergistic platform. The Tianjic chip adopts a many-core architecture, reconfigurable building blocks and a streamlined dataflow with hybrid coding schemes, and can not only accommodate computer-science-based machine-learning algorithms, but also easily implement brain-inspired circuits and several coding schemes. Using just one chip, we demonstrate the simultaneous processing of versatile algorithms and models in an unmanned bicycle system, realizing real-time object detection, tracking, voice control, obstacle avoidance and balance control. Our study is expected to stimulate AGI development by paving the way to more generalized hardware platforms.