E2CDA: Type I: EXtremely Energy Efficient Collective ELectronics (EXCEL)

随着数十亿智能和联网设备的出现，“数据泛滥”已经成为现实，仅去年一年就产生了超过8泽字节的数据。这项多学科研究工作的主要重点是开发一种新的计算模式，名为极高能效的集体电子（EXCEL），以实现硬件加速的数据分析，可以从未标记和非结构化数据中提取信息。这项研究有望从根本上揭示利用耦合动力系统以节能方式解决计算难题的新方法。随着新材料和设备的创新，芯片级动力系统的实现，架构的变化和关键的基准测试，EXCEL将为新的非冯诺依曼计算范式奠定基础，以实现计算能源效率的数量级改进。外联活动的优先重点是教育后代学生适应信息处理系统即将发生的演变和革命。该研究项目的结构是受益于与工业界的密切合作，这将促进未来的技术转让。参与的PI致力于为本科生和研究生开发新兴纳米技术、非传统计算、机器学习和计算医学领域的课程模块。EXCEL项目通过利用新型设备中的涌现现象和耦合系统的动力学来执行优化，学习和推理任务的集体，合作和可扩展的方式。EXCEL的智能基础在于利用嵌入在耦合网络（振荡和尖峰）的时空动态中的本地信息内容，以大规模并行的方式执行计算。集体计算范式是对传统冯-诺依曼计算模型的及时偏离，该模型依赖于批量、离散时间、迭代更新（缺乏时间局部性）和共享状态（缺乏空间局部性）。EXCEL专注于开发模拟计算的复杂性理论基础。这包括探索连续和离散优化问题以及在线学习的随机机器。EXCEL研究人员将积极追求物理硬件演示，并量化其在解决高性能数据中心，实时网络物理系统和计算医学中不断扩大应用的计算难题方面优于布尔计算机的优势。

With billions of smart and connected devices, "data deluge" is a reality with more than eight zettabytes of data generated in last year alone. The primary focus of this multi-disciplinary research effort is to develop a new paradigm of computing titled Extremely energy efficient Collective Electronics (EXCEL) to enable hardware accelerated data-analytics that can extract information from unlabeled and unstructured data. This research is expected to uncover fundamentally new ways of harnessing coupled dynamical systems for solving computationally hard problems in an energy efficient way. With innovations in novel materials and devices, chip-scale dynamical system implementation, architectural changes and critical benchmarking, EXCEL will lay the foundation for a new non von-Neumann computing paradigm to achieve orders of magnitude improvement in computational energy efficiency. The outreach activities are prioritized around educating future generations of students to adapt to the forthcoming evolution and revolution in information processing systems. The research project is structured to benefit from strong engagement with industry, which will facilitate technology transfer in the future. The participating PIs are committed to developing course modules for both undergraduate and graduate students in the areas of emerging nanotechnologies, unconventional computing, machine learning and computational medicine.The EXCEL project lays the foundation for a radically different approach to energy efficient information processing by leveraging emergent phenomena in novel devices and dynamics of coupled systems to execute optimization, learning and inference tasks in a collective, cooperative and scalable way. The intellectual foundation of EXCEL rests on the utilization of local information content embedded in the spatio-temporal dynamics of coupled networks (oscillatory and spiking) to perform computation in a massively parallel way. The collective computing paradigm is a timely departure from the traditional model of von-Neumann computing which relies on batch, discrete time, iterative updates (lacking temporal locality) and shared states (lacking spatial locality). EXCEL focuses on developing a complexity theoretic foundation in analog computing. This includes exploration of both continuous and discrete optimization problems as well as stochastic machines for on-line learning. The EXCEL researchers will actively pursue physical hardware demonstration and quantify their advantages over Boolean computers in solving computationally hard problems that are finding ever expanding applications in high-performance data centers, real-time cyber-physical systems and computational medicine.