Article
Meteorology & Atmospheric Sciences
Malarvizhi Arulraj, Veljko Petkovic, Ralph r. Ferraro, Huan Meng
Summary: This study uses a high-quality radar network and deep learning methods to characterize different precipitation regimes and their vertical profiles. By using clustering algorithms, 18 distinct precipitation patterns are identified based on their structural features and precipitation rate/type. These identified precipitation regimes can be used for physics-guided retrievals and further studying precipitation patterns.
JOURNAL OF HYDROMETEOROLOGY
(2023)
Article
Chemistry, Analytical
Jingyu Lu, Kai Wang, Chen Chen, Weixi Ji
Summary: This study proposes a fault diagnosis method for rolling bearings based on Gramian angular field (GAF) coding technology and an improved ResNet50 model to address the low accuracy and timeliness of traditional fault diagnosis methods. By recoding the one-dimensional vibration signal into a two-dimensional feature image using GAF technology and using it as input for the model, combined with the advantages of ResNet algorithm in image feature extraction and classification recognition, automatic feature extraction and fault diagnosis are achieved, and classification of different fault types is accomplished.
Article
Computer Science, Information Systems
Francisco J. Pulgar, Francisco Charte, Antonio J. Rivera, Maria J. del Jesus
Summary: The paper introduces a new classifier, ClEnDAE, which uses ensemble methods and DAE to reduce dimensionality of input space and improve predictive performance. Experimental results show that the algorithm outperforms other traditional methods in classification.
INFORMATION SCIENCES
(2021)
Article
Environmental Sciences
Md Touhid Islam, Md Rashedul Islam, Md Palash Uddin, Anwaar Ulhaq
Summary: Object classification in hyperspectral images is challenging due to high dimensionality and class imbalance. We propose a framework that addresses these challenges through dimensionality reduction and re-sampling. Our framework employs a subgroup-based dimensionality reduction technique and achieves class balance. The reduced and balanced data are processed by a hybrid CNN model to extract spectral-spatial features and improve classification accuracy.
Article
Engineering, Aerospace
Michele Lazzara, Max Chevalier, Michele Colombo, Jasone Garay Garcia, Corentin Lapeyre, Olivier Teste
Summary: This paper presents an innovative surrogate model based on dual-phase Long-Short Term Memory (LSTM) Autoencoder. It is applied in an industrial context to predict the dynamic landing response of aircraft over time. The proposed model outperforms other surrogate models in predicting high-dimensional temporal system responses.
AEROSPACE SCIENCE AND TECHNOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Rocio del Amor, Adrian Colomer, Carlos Monteagudo, Valery Naranjo
Summary: Epigenetic alterations play a crucial role in cancer development, and the proposed deep embedded refined clustering method for breast cancer differentiation based on DNA methylation achieves high clustering accuracy and low error rate in breast tissue samples. The method involves dimensionality reduction using an autoencoder and clustering in the latent space, showing superior performance compared to other state-of-the-art methods for breast cancer classification using DNA methylation data.
NEURAL COMPUTING & APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Uwe Reuter, Aditha Jayaram, Mina Rezkalla, Wolfgang Weber
Summary: This paper discusses the application of autoencoders in deep learning, particularly their ability to detect and determine dependencies among parameters in input data sets. By using stacked autoencoders and sensitivity measures, these dependencies can be automatically detected.
Article
Computer Science, Artificial Intelligence
Zhengming Ma, Zengrong Zhan, Zijian Feng, Jiajing Guo
Summary: The SGLC-ML algorithm divides data into straight-like geodesics, maps them to straight lines in a low-dimensional Euclidean space, and utilizes local coordinates for dimensionality reduction, achieving good performance compared to other algorithms.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2021)
Article
Automation & Control Systems
Shaowu Pan, Steven L. Brunton, J. Nathan Kutz
Summary: High-dimensional spatio-temporal dynamics can be encoded in a low-dimensional subspace. To address practical engineering challenges, we propose a general framework called Neural Implicit Flow (NIF), which enables mesh-agnostic, low-rank representation of large-scale parametric spatio-temporal data.
JOURNAL OF MACHINE LEARNING RESEARCH
(2023)
Article
Environmental Sciences
Shalini Gakhar, Kailash Chandra Tiwari
Summary: This study extends the target detection method for urban applications using deep learning with AVIRIS-NG data, showing significantly higher results compared to existing literature. However, as the number of PCA components and window size increase, the time complexity rises, leading to a compromise with accuracy.
GEOCARTO INTERNATIONAL
(2022)
Article
Thermodynamics
Lei Zhou, Jiahao Wen, Zhaokun Wang, Pengru Deng, Hongfu Zhang
Summary: In order to optimize wind farm layout and improve wind energy conversion, accurately and efficiently modeling wind turbine wake is of great significance. This study proposes a novel dimensionality reduction method, called delayed proper orthogonal decomposition (d-POD), combined with long short-term memory (LSTM) network to predict the unsteady wind turbine wake velocity. The d-POD-LSTM model shows superior performance compared to the conventional POD-LSTM model, with a reduction of prediction error by 80% when the delayed number increases from 1 to 16.
Article
Computer Science, Artificial Intelligence
Yunlong Gao, Yisong Zhang, Jinyan Pan, Sizhe Luo, Chengyu Yang
Summary: This research integrates discriminant manifold learning with discriminant analysis, introducing an adaptive adjacency factor to propose a novel method called discriminant analysis based on reliability of local neighborhood (DA-RoLN) to address the drawbacks of existing methods and emphasize the importance of valid samples while reducing the influence of outliers. Extensive experimental results demonstrate the effectiveness of DA-RoLN.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Mathematics
Nagwan Abdel Samee, Ghada Atteia, Souham Meshoul, Mugahed A. Al-antari, Yasser M. Kadah
Summary: The article discusses a breast cancer classification framework built using transfer learning and deep learning techniques, successfully achieving an accuracy of 98.50%, sensitivity of 98.06%, specificity of 98.99%, and precision of 98.98% by overcoming the deep feature dimensionality issue.
Article
Nanoscience & Nanotechnology
Soumyashree S. Panda, Ravi S. Hegde
Summary: This study introduces a deep learning-based design methodology for the inverse design of extended unit-cell metagratings. Unlike previous approaches, this method learns the spectral response of the metagrating across its reflected and transmitted orders, accelerating the optimization of multiple functionalities. The proposed methodology is not limited to proof-of-concept demonstrations and can be widely applied to nanophotonic system design.
Article
Computer Science, Artificial Intelligence
Gengshi Huang, Zhengming Ma, Tianshi Luo
Summary: This paper introduces a statistic for measuring the correlation between two random variables using kernelized cross-covariance criterion (kCCC), and applies kCCC to data dimensionality reduction combined with local geometric property preservation method and manifold learning dimensionality reduction method, to maintain both global statistical characteristics and local geometric characteristics simultaneously.
JOURNAL OF MATHEMATICAL IMAGING AND VISION
(2021)
Article
Chemistry, Multidisciplinary
Sajjad Abdollahramezani, Omid Hemmatyar, Mohammad Taghinejad, Hossein Taghinejad, Yashar Kiarashinejad, Mohammadreza Zandehshahvar, Tianren Fan, Sanchit Deshmukh, Ali A. Eftekhar, Wenshan Cai, Eric Pop, Mostafa A. El-Sayed, Ali Adibi
Summary: This study introduces a reconfigurable hybrid metasurface platform by incorporating the phase-change material Ge2Sb2Te5 (GST) into metal-dielectric meta-atoms for active and nonvolatile tuning of properties of light. The reduced-dimension meta-atom can selectively control the hybrid plasmonic-photonic resonances of the metasurface via the dynamic change of optical constants of GST, demonstrating promising applications in high-contrast optical switching and efficient beam deflection. Findings suggest dynamic hybrid metasurfaces as compelling candidates for next-generation reprogrammable meta-optics.
Article
Optics
Hossein Taghinejad, Sajjad Abdollahramezani, Ali A. Eftekhar, Tianren Fan, Amir H. Hosseinnia, Omid Hemmatyar, Ali Eshaghian Dorche, Alexander Gallmon, Ali Adibi
Summary: The study introduces a method utilizing GST phase-change materials and ITO microheater platform for electrically reconfigurable optical phase shifters. The potential applications of this hybrid GST/ITO platform in nanophotonics are demonstrated, providing new possibilities for miniaturized integrated photonic structures.
Article
Optics
Muliang Zhu, Sajjad Abdollahramezani, Tianren Fan, Ali Adibi
Summary: Germanium antimony telluride (GST) is a promising material for reconfiguring subwavelength nanostructures due to its strong non-volatile change of the refractive index between amorphous and crystalline states. By incorporating GST into an electromagnetically-induced-transparency-based silicon metasurface, researchers successfully demonstrated a giant third-harmonic generation (THG) switch with high modulation depth. This study shows the high potential of GST-based fast dynamic nonlinear photonic switches for real-world applications.
Article
Multidisciplinary Sciences
Sajjad Abdollahramezani, Omid Hemmatyar, Mohammad Taghinejad, Hossein Taghinejad, Alex Krasnok, Ali A. Eftekhar, Christian Teichrib, Sanchit Deshmukh, Mostafa A. El-Sayed, Eric Pop, Matthias Wuttig, Andrea Alu, Wenshan Cai, Ali Adibi
Summary: The authors demonstrate an efficient platform for electrically driven reconfigurable metasurfaces using phase-change materials. This platform allows for non-volatile, reversible, multilevel, and fast optical modulation and wavefront engineering in the near-infrared spectral range. The study represents a critical advance towards the development of fully integrable dynamic metasurfaces and their potential for beamforming applications.
NATURE COMMUNICATIONS
(2022)
Article
Nanoscience & Nanotechnology
Sajjad Abdollahramezani, Hossein Taghinejad, Tianren Fan, Mahmood Reza Marzban, Ali A. Eftekhar, Ali Adibi
Summary: We present a hybrid device platform for creating an electrically reconfigurable metasurface by integrating plasmonic nanostructures with phase-change material GST. By changing the phase of GST, a wide range of responses can be achieved, leading to the realization of a broadband electrically tunable multifunctional metadevice.
Article
Materials Science, Multidisciplinary
Muliang Zhu, Sajjad Abdollahramezani, Chentao Li, Tianren Fan, Hayk Harutyunyan, Ali Adibi
Summary: In this study, a dynamically reconfigurable asymmetric Fabry-Perot cavity based on phase-change alloy Ge2Sb2Te5 (GST) is numerically designed and experimentally demonstrated, showing a large shift of the third-harmonic generation (THG) resonant band. Continuous resonant spectral shifting is achieved through the precise control of the semicrystalline phase of GST. The tunable THG source provides efficient broadband harmonic generation in the violet-blue visible wavelength range.
ADVANCED PHOTONICS RESEARCH
(2022)
Article
Nanoscience & Nanotechnology
Muliang Zhu, Sajjad Abdollahramezani, Chentao Li, Tianren Fan, Hayk Harutyunyan, Ali Adibi
Summary: This study demonstrates experimentally controllable second-harmonic generation (SHG) switches in a tunable metasurface by actively controlling the crystalline phase of germanium antimony telluride (GST). The results show that high modulation depths and resonant SHG efficiencies can be achieved by controlling the phase of GST, making these switches potentially useful for practical applications such as microscopy, optical communication, and photonic computing in the nonlinear regime.
Proceedings Paper
Engineering, Electrical & Electronic
Muliang Zhu, Chentao Li, Tianren Fan, Sajjad Abdollahramezani, Xi Wu, Hayk Harutyunyan, Ali Adibi
Summary: The study demonstrates an amorphous silicon metasurface that efficiently generates third harmonic generation by breaking the symmetry of both co-polarized and cross-polarized. This is achieved through quasibound states in the continuum.
2021 CONFERENCE ON LASERS AND ELECTRO-OPTICS (CLEO)
(2021)
Proceedings Paper
Engineering, Electrical & Electronic
Muliang Zhu, Sajjad Abdollahramezani, Chentao Li, Tianren Fan, Hayk Harutyunyan, Ali Adibi
Summary: The study demonstrates broadband continuous tuning of third-harmonic generation (THG) using GST material in a Fabry-Perot cavity, highlighting the potential applications of this technology.
2021 CONFERENCE ON LASERS AND ELECTRO-OPTICS (CLEO)
(2021)
Proceedings Paper
Engineering, Electrical & Electronic
Sajjad Abdollahramezani, Omid Hemmatyar, Hossein Taghinejad, Muliang Zhu, Alexander Gallmon, Ali Adibi
Summary: This study presents a tunable hybrid metasurface for non-volatile optical modulation, utilizing phase-change materials and plasmon hybridization, and investigates the impact of structural parameters on optical performance through machine learning algorithms.
2021 CONFERENCE ON LASERS AND ELECTRO-OPTICS (CLEO)
(2021)
Proceedings Paper
Engineering, Electrical & Electronic
Sajjad Abdollahramezani, Omid Hemmatyar, Hossein Taghinejad, Muliang Zhu, Alexander Gallmon, Ali Adibi
Summary: This experimental study shows the active modulation of amplitude/phase profiles of optical wavefronts by utilizing the interplay of surface plasmon polariton and electric/magnetic Mie resonance modes in hybrid plasmonic-dielectric metasurface platforms incorporating chalcogenide phase-change materials.
2021 CONFERENCE ON LASERS AND ELECTRO-OPTICS (CLEO)
(2021)
Proceedings Paper
Engineering, Electrical & Electronic
Omid Hemmatyar, Sajjad Abdollahramezani, Tyler Brown, Ali Adibi
Summary: We demonstrate nanoscale high-saturation color switching using Mie scattering resonances supported by an all-dielectric metasurface made of phase-change material GeSe3 nanopillars.
2021 CONFERENCE ON LASERS AND ELECTRO-OPTICS (CLEO)
(2021)
Proceedings Paper
Engineering, Electrical & Electronic
Omid Hemmatyar, Sajjad Abdollahramezani, Hossein Taghinejad, Ali Adibi
Summary: This study presents the design and experimental demonstration of an electrically-tunable all-dielectric metasurface using phase-change material GST and transparent conductive indium tin oxide (ITO) to control light absorption for multi-level electrooptic modulation with unprecedented sensitivity.
2021 CONFERENCE ON LASERS AND ELECTRO-OPTICS (CLEO)
(2021)
Proceedings Paper
Engineering, Electrical & Electronic
Omid Hemmatyar, Sajjad Abdollahramezani, Hossein Taghinejad, Ali Adibi
2020 CONFERENCE ON LASERS AND ELECTRO-OPTICS (CLEO)
(2020)
Proceedings Paper
Engineering, Electrical & Electronic
Omid Hemmatyar, Sajjad Abdollahramezani, Yashar Kiarashinejad, Mohammadreza Zandehshahvar, Ali Adibi
2020 CONFERENCE ON LASERS AND ELECTRO-OPTICS (CLEO)
(2020)