RTML: Small: Design of System Software to Facilitate Real-Time Neuromorphic Computing

神经网络等机器学习方法已经成功地应用于实时计算机视觉和信号处理领域。模拟生物神经元和突触的神经形态系统可用于在能量受限的计算平台中实现这些神经网络。然而，由于缺乏用户友好的编程界面，神经形态系统的使用目前仅限于研究。该项目将开发这样一个接口，使这些系统更容易编程，并被美国更广泛的科学和工程社区使用。该项目将把本科教育整合到多学科研究生研究中，让不同学科的各级学生参与拟议的研究活动。高中女生将在夏季参加研讨会，学习如何为配备神经形态硬件的机器人编程。开源编程工具将使美国神经形态系统的开发和商业化更快，并促进与全球其他此类社区的合作。在计算机上执行程序包括几个步骤：编译、资源分配和运行时映射。虽然在主流计算机中已经有了很好的定义，但是在神经形态系统中还没有系统地研究过这些步骤。该项目将开发编译器工具链，将用户的机器学习程序翻译为可由神经形态系统解释的低级语言。一个关键的举措是开发跨不同平台的通用表示。将制定资源优化策略以提高项目绩效；以及一个类似操作系统的框架，允许程序员轻松地在神经形态系统上部署他们的机器学习程序。技术贡献将通过两个案例研究来展示：实时睡眠呼吸暂停检测和实时视频图像分割。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Machine learning methods such as neural networks have been successfully used in real-time computer vision and signal processing areas. Neuromorphic systems, which mimic biological neurons and synapses, can be used to implement these neural networks in energy-constrained computing platforms. However, due to the absence of a user-friendly programming interface, the use of neuromorphic systems is currently limited to research. This project will develop such an interface, allowing for these systems to be more easily programmed and used by the broader science and engineering community within the U.S. The project will integrate undergraduate education within multidisciplinary graduate research, with students at all levels and across different disciplines participating in the proposed research activities. High school girls will participate in workshops during the summer, learning how to program robots equipped with neuromorphic hardware. The open-sourced programming tools will enable faster development and commercialization of neuromorphic systems in the U.S. and facilitate collaboration with other such communities worldwide.Executing a program on a computer involves several steps: compilation, resource allocation, and run-time mapping. Although very well defined for mainstream computers, no prior work has investigated these steps in a systematic manner for neuromorphic systems. This project will develop compiler tool chains to translate a user's machine learning program to low-level languages that can be interpreted by neuromorphic systems. A key initiative is to develop a common representation across different platforms. Resource optimization strategies will be developed to improve program performance; as well as an Operating System like framework that will allow programmers to easily deploy their machine learning programs on neuromorphic systems. The technical contributions will be demonstrated using two case studies: real-time sleep apnea detection and real-time image segmentation from video.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.