Book Chapter10.1016/b978-0-323-98829-2.00011-6
Photonic matrix computing accelerators
Jianji Dong,Hailong Zhou,Dexiu Huang +2 more
- 01 Jan 2024
- pp 257-293
TL;DR: Photonic matrix computing accelerators accelerate matrix computations and related applications with high speed and low energy consumption.
read more
Abstract: With the growing demand for electronic computing, data centers, and emerging artificial intelligence applications, low latency and high energy efficiency are hard to ensure by traditional electrical methods. Photonic methods can perform high-speed parallel information processing with ultralow energy consumption benefiting from its superior performance. Photonic accelerators are designed to accelerate specific categories of computing in the optical domain, especially matrix computation, to address the growing demand for computing resources and capacity. In this chapter, methods for photonic matrix multiplication are first introduced and compared. Then, the expansions for applications in optical signal processing, optical neural networks, and combinatorial optimization problems with photonic matrix computing accelerators are presented. Finally, general-purpose spatial mode processors, optical neural networks, and Ising machines with photonic accelerators are analyzed in detail.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
References
Learning the parts of objects by non-negative matrix factorization
TL;DR: An algorithm for non-negative matrix factorization is demonstrated that is able to learn parts of faces and semantic features of text and is in contrast to other methods that learn holistic, not parts-based, representations.
14.2K
Mastering the game of Go without human knowledge
David Silver,Julian Schrittwieser,Karen Simonyan,Ioannis Antonoglou,Aja Huang,Arthur Guez,Thomas Hubert,Lucas Baker,Matthew Lai,Adrian Bolton,Yutian Chen,Timothy P. Lillicrap,Fan Hui,Laurent Sifre,George van den Driessche,Thore Graepel,Demis Hassabis +16 more
TL;DR: An algorithm based solely on reinforcement learning is introduced, without human data, guidance or domain knowledge beyond game rules, that achieves superhuman performance, winning 100–0 against the previously published, champion-defeating AlphaGo.
Terabit free-space data transmission employing orbital angular momentum multiplexing
Jian Wang,Jian Wang,Jeng-Yuan Yang,Irfan Fazal,Nisar Ahmed,Yan Yan,Hao Huang,Yongxiong Ren,Yang Yue,S. Dolinar,Moshe Tur,Alan E. Willner +11 more
TL;DR: In this paper, the authors demonstrate the ability to multiplex and transfer data between twisted beams of light with different amounts of orbital angular momentum, which provides new opportunities for increasing the data capacity of free-space optical communications links.
Space-division multiplexing in optical fibres
TL;DR: In this paper, the authors summarized the simultaneous transmission of several independent spatial channels of light along optical fibres to expand the data-carrying capacity of optical communications, and showed that the results achieved in both multicore and multimode optical fibers are documented.
Deep learning with coherent nanophotonic circuits
Yichen Shen,Nicholas C. Harris,Scott Skirlo,Dirk Englund,Marin Soljacic +4 more
- 01 Jul 2017
TL;DR: A new architecture for a fully optical neural network is demonstrated that enables a computational speed enhancement of at least two orders of magnitude and three order of magnitude in power efficiency over state-of-the-art electronics.
2.9K