Share Email Print

Proceedings Paper

Artificial-intelligence-assisted photonics (Conference Presentation)

Paper Abstract

Discovering novel, unconventional optical designs in combination with advanced machine-learning assisted data analysis techniques can uniquely enable new phenomena and breakthrough advances in many areas including on-chip circuitry, imaging, sensing, energy, and quantum information technology. Topology optimization, which has previously revolutionized aerospace and mechanical engineering by providing non-intuitive solutions to highly constrained material distribution problems, has recently emerged as a powerful architect for advanced photonic design. Compared to other inverse-design approaches that require extreme computation power to undertake a comprehensive search within a large parameter space, topology optimization can expand the design space while improving the computational efficiency. This talk will highlight our most recent findings on 1) merging topology optimization with artificial-intelligence-assisted algorithms and 2) integrating machine-learning based analysis with photonic design and quantum optical measurements. Particularly, we will discuss our studies on implementing deep-learning assisted topology optimization for advanced metasurface design development, focusing on highly efficient thermal emitter/absorber development for thermal-photovoltaics applications. We will summarize our research on merging topology optimization technique with quantum device design for achieving ultrafast single-photon source that offers efficient on-chip integration. Finally, we will also describe our recent works on implementing a novel convolutional neural network-based technique for real-time material defect metrology at the quantum level that outperforms all existing approaches in terms of speed and fidelity. This new method rapidly extracts the values of the single-photon autocorrelation function at zero delays from sparse data and ensures one order speed up on solving “bad”/”good” emitter classification problem in comparison with conventional techniques.

Paper Details

Date Published: 9 September 2019
Proc. SPIE 11080, Metamaterials, Metadevices, and Metasystems 2019, 110802K (9 September 2019); doi: 10.1117/12.2528916
Show Author Affiliations
Zhaxylyk A. Kudyshev, Purdue Univ. (United States)
Simeon Bogdanov, Purdue Univ. (United States)
Alexander V. Kildishev, Purdue Univ. (United States)
Alexandra Boltasseva, Purdue Univ. (United States)
Vladimir M. Shalaev, Purdue Univ. (United States)

Published in SPIE Proceedings Vol. 11080:
Metamaterials, Metadevices, and Metasystems 2019
Nader Engheta; Mikhail A. Noginov; Nikolay I. Zheludev, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?