A5101411710- Advanced Fiber Optic Sensors
- Domain Adaptation and Few-Shot Learning
- Advanced Photonic Communication Systems
- Anomaly Detection Techniques and Applications
- Advanced Optical Sensing Technologies
- Statistical Methods and Inference
- Bayesian Methods and Mixture Models
- Advanced X-ray and CT Imaging
- Photonic and Optical Devices
- Generative Adversarial Networks and Image Synthesis
- AI in cancer detection
- Structural Health Monitoring Techniques
- Medical Imaging Techniques and Applications
- Speech and Audio Processing
- Statistical Methods and Bayesian Inference
- Gaussian Processes and Bayesian Inference
- Tensor decomposition and applications
- Advanced Optical Network Technologies
- Ocular and Laser Science Research
- Optical Network Technologies
- Sparse and Compressive Sensing Techniques
- Magneto-Optical Properties and Applications
- Semiconductor Lasers and Optical Devices
- Advanced Clustering Algorithms Research
- Machine Learning and ELM
NEC (United States)
2021-2025
Princeton University
2022-2024
Virginia Tech
2021
Duke University
2012-2017
Chinese Academy of Sciences
2009
We review recent advances in distributed fiber optic sensing (DFOS) and their applications. The scattering mechanisms glass, which are exploited for reflectometry-based DFOS, Rayleigh, Brillouin, Raman scatterings. These sensitive to either strain and/or temperature, allowing optical cables monitor ambient environment addition conventional role as a medium telecommunications. Recently, DFOS leveraged technologies developed telecommunications, such coherent detection, digital signal...
We review various applications of distributed fiber optic sensing (DFOS) and machine learning (ML) technologies that particularly benefit telecom operators' networks businesses. By leveraging relative phase shift the reflectance coherent Rayleigh, Brillouin Raman scattering light wave, ambient environmental vibration, acoustic effects, temperature fiber/cable strain can be detected. Fiber technology allows optical to support features in addition its conventional role transmit data...
Contrastive Language-Audio Pretraining (CLAP) models have demonstrated unprecedented performance in various acoustic signal recognition tasks. Fiber-optic-based is one of the most important downstream tasks and plays a significant role environmental sensing. Adapting CLAP for fiber-optic has become an active research area. As non-conventional sensor, presents challenging, domain-specific, low-shot deployment environment with domain shifts due to unique frequency response noise...
Our study introduces two pioneering applications leveraging Distributed Fiber Optic Sensing (DFOS) and Machine Learning (ML) technologies. These innovations offer substantial benefits for fortifying telecom infrastructures public safety. By harnessing existing cables, our solutions excel in perimeter intrusion detection via buried cables impulsive event classification through aerial cables. To achieve comprehensive detection, we introduce a label encoding strategy multitask learning...
A new method for learning variational autoencoders (VAEs) is developed, based on Stein gradient descent. key advantage of this approach that one need not make parametric assumptions about the form encoder distribution. Performance further enhanced by integrating proposed with importance sampling. Excellent performance demonstrated across multiple unsupervised and semi-supervised problems, including analysis ImageNet data, demonstrating scalability model to large datasets.
Polarization-based, multi-span sensing over a link with reflection-back circuits is demonstrated experimentally. By measuring rotation matrices instead of just monitoring polarization, 35 dB extinction in localization achieved regardless the disturbance magnitude.
We present a manhole localization method based on distributed fiber optic sensing and weakly supervised machine learning techniques. For the first time to our knowledge, ambient environment data is used for underground cable mapping with promise of enhancing operational efficiency reducing field work. To effectively accommodate weak informativeness data, selective sampling scheme an attention-based deep multiple instance classification model are adopted, which only requires annotated data....
We report the field trial results of monitoring abnormal activities near deployed cable with fiber-optic-sensing technology for protection. Detection and position determination events evaluating threat to is realized.
We review the distributed-fiber-sensing field trial results over deployed telecom networks. With local AI processing, real-time detection, and localization of abnormal events with cable damage threat assessment are realized for self-protection.
We report the distributed-fiber-sensing field trial results over a 5G-transport-network. A standard communication fiber is used with real-time AI processing for cable self-protection, cable- cut threat assessment and road traffic monitoring in long-term continuous test.
We propose a globally convergent alternating minimization (AM) algorithm for image reconstruction in transmission tomography, which extends automatic relevance determination (ARD) to Poisson noise models with Beer's law. The promotes solutions that are sparse the pixel/voxel--difference domain by introducing additional latent variables, one each pixel/voxel, and then learning these variables from data using hierarchical Bayesian model. Importantly, proposed AM is free of any tuning...
This article is motivated by soccer positional passing networks collected across multiple games. We refer to these data as replicated spatial networks---to accurately model such it necessary take into account the positions of passer and receiver for each event. registration replicates that occur games represent key differences with usual social network data. As a step before investigating how dynamics influence team performance, we focus on developing methods summarizing different team's...
Learning multiple tasks across heterogeneous domains is a challenging problem since the feature space may not be same for different tasks. We assume data in are generated from latent common domain via sparse transforms and propose probit model (LPM) to jointly learn transforms, shared classifier domain. To meaningful task relatedness avoid over-fitting classification, we introduce sparsity matrices, as well classifier. derive theoretical bounds estimation error of terms transforms. An...
In distributed acoustic sensing (DAS) on aerial fiber-optic cables, utility pole localization is a prerequisite for any subsequent event detection. Currently, localizing the poles DAS traces relies human experts who manually label poles' locations by examining signal patterns generated in response to hammer knocks poles. This process inefficient, error-prone and expensive, thus impractical non-scalable industrial applications. this paper, we propose two machine learning approaches automate...
We utilize copulas to constitute a unified framework for constructing and optimizing variational proposals in hierarchical Bayesian models. For models with continuous non-Gaussian hidden variables, we propose semiparametric automated Gaussian copula approach, which the parametric family is able preserve multivariate posterior dependence, nonparametric transformations based on Bernstein polynomials provide ample flexibility characterizing univariate marginal posteriors.
In this paper, we investigate post-processing for the frequency-domain blind source separation (FD-BSS) in hearing aids applications. It is known that segregate quality of FD-BSS degrades severely challenging scenario reverberant enclosures or moving situations. A robust two-stage dynamic programming approach based on inter-frequency correlation presented to solving permutation ambiguity correction problem. Moreover, binary masking method and non-stationary spectral subtraction techniques...
We implement a cascaded learning framework using component-level EDFA models for optical power spectrum prediction in multi-span networks, achieving mean absolute error of 0.17 dB across 6 spans and 12 EDFAs with only one-shot measurement.