A5100690569- Sparse and Compressive Sensing Techniques
- Image and Signal Denoising Methods
- Microwave Imaging and Scattering Analysis
- Direction-of-Arrival Estimation Techniques
- Mathematical Analysis and Transform Methods
- Remote-Sensing Image Classification
- Photoacoustic and Ultrasonic Imaging
- Numerical methods in inverse problems
- Advanced Numerical Analysis Techniques
- Wind and Air Flow Studies
- Hydrocarbon exploration and reservoir analysis
- Digital Filter Design and Implementation
- Manufacturing Process and Optimization
- Advanced Vision and Imaging
- Image Processing Techniques and Applications
- Structural Health Monitoring Techniques
- Approximation Theory and Sequence Spaces
- Robotic Path Planning Algorithms
- Advanced Data Compression Techniques
- Product Development and Customization
- Geological and Geophysical Studies
- Reinforcement Learning in Robotics
- Geophysical Methods and Applications
- Radiation Detection and Scintillator Technologies
- Medical Imaging Techniques and Applications
University of Electronic Science and Technology of China
2024
City College of New York
2022-2023
Zhengzhou University
2023
Changsha University of Science and Technology
2023
Courant Institute of Mathematical Sciences
2019-2022
New York University
2019-2022
Xi'an Jiaotong University
2022
Shanghai Jiao Tong University
2022
Hunan University
2021
University of Maryland, College Park
2016-2020
The problem of imaging point objects can be formulated as estimation an unknown atomic measure from its M+1 consecutive noisy Fourier coefficients. standard resolution this inverse is 1/M and super-resolution refers to the capability resolving atoms at a higher resolution. When any two are less than apart, recovery highly challenging many existing algorithms either cannot deal with situation or require restrictive assumptions on sign measure. ESPRIT efficient method which does not depend...
Abstract This paper discusses the results of a field experiment conducted at Savannah River National Laboratory to test performance several algorithms for localization radioactive materials. In this multirobot system, both an unmanned aerial vehicle, custom hexacopter, and ground vehicle (UGV), ClearPath Jackal, equipped with γ ‐ray spectrometers, were used collect data from two source configurations. Both Fourier scattering transform Laplacian eigenmap detection tested on collected sets....
This paper studies the spectral estimation problem of estimating locations a fixed number point sources given multiple snapshots Fourier measurements collected by uniform array sensors. We prove novel stability bounds for MUSIC and ESPRIT as function noise standard deviation, snapshots, source amplitudes, support. Our most general result is perturbation bound signal space in terms minimum singular value matrices. When are located several separated clumps, we provide an explicit upper...
We consider the inverse problem of recovering locations and amplitudes a collection point sources represented as discrete measure, given $M+1$ its noisy low-frequency Fourier coefficients. Super-resolution refers to stable recovery when distance $\Delta$ between two closest is less than $1/M$. introduce clumps model where are closely spaced within several clumps. Under this assumption, we derive non-asymptotic lower bound for minimum singular value Vandermonde matrix whose nodes determined...
This paper presents a new path planning method for mobile robot in an unstructured and dynamic environment. The consists of two steps: first, probabilistic roadmap (PRM) is constructed stored as graph whose nodes correspond to collision-free world state the robot; second, Q-learning-a reinforcement learning, integrated with PRM determine proper reach goal. In this manner, able use past experience improve its performance avoiding not only static obstacles but also moving obstacles, without...
Recent developments in machine learning and signal processing have resulted many new techniques that are able to effectively capture the intrinsic yet complex properties of hyperspectral imagery (HSI). Tasks ranging from anomaly detection classification can now be solved by taking advantage very efficient algorithms which their roots representation theory computational approximation. Time–frequency methods one example such techniques. They provide means analyze extract spectral content data....
Radar detection has proven to be an effective, nondestructive test for the determination of quality wood-based materials, especially in wooden structures ancient buildings and trees. However, results are usually inaccurate, it is difficult interpret internal anomalies due moisture content wood, individual differences, other factors. In this paper, a new measurement method proposed based on use ground-penetrating radar (GPR) abnormality localization imaging. Firstly, time delay reflected...
We explore the representation capabilities of scattering transforms for classification hyperspectral images. examine several types, including a recently developed technique called Fourier transform. This method is naturally suited data because it decomposes signals into multi-frequency bands and removes small perturbations such as noise. test on four standard datasets, results indicate that transform effective at representing spectral data. also present spatial-spectral combines Wavelet...
In this study we developed three hands-on activities to teach high school students computational thinking (CT) and, specifically, the decomposition skills. The were designed enable solve problems by using application tools. computer science concepts utilized in included binary search, quick sort and iteration. We evaluated effect of utilizing a post-activity questionnaire, post-test, students' worksheets, semi-structured interviews with participating students. results indicated that improved...
This paper studies the problem of recovering a discrete complex measure on torus from finite number corrupted Fourier samples. We assume support unknown satisfies minimum separation condition and we use convex regularization methods to recover approximations original measure. focus two well-known methods, for both, establish an error estimate that bounds smoothed-out in terms target resolution noise level. Our $L^\infty$ approximation rate is entirely new one improves upon previously...
This paper studies stable recovery of a collection point sources from its noisy M+1 low-frequency Fourier coefficients. We focus on the super-resolution regime where minimum separation is below 1/M. propose separated clumps model are clustered in far apart sets, and prove an accurate lower bound matrix with nodes restricted to source locations. estimate gives rise theoretical analysis limit MUSIC algorithm.
This paper presents arobust Q-learning method for path planningin a dynamic environment. The consists of three steps: first, regime-switching Markov decision process (RSMDP) is formed to present the environment; second probabilistic roadmap (PRM) constructed, integrated with RSMDP and stored as graph whose nodes correspond collision-free world state robot; third, an onlineQ-learning stepsize, which facilitates robust convergence Q-value iteration, PRM determine optimal reaching goal. In this...
This paper presents a simulation model for assessing the mutual impacts of Engineering Change Management (ECM) process and New Product Development (NPD) on each other. The discrete-event incorporates ECM into an NPD environment by allowing Changes (EC) to compete limited resources with regular activities. goal is examine how relative size frequency as well ECM, structure (in terms overlapping departmental interaction), operational policy resource using priority that one organization employs...
We study the generalization error of functions that interpolate prescribed data points and are selected by minimizing a weighted norm. Under natural general conditions, we prove both interpolants their errors converge as number parameters grows, limiting interpolant belongs to reproducing kernel Hilbert space. This rigorously establishes an implicit bias minimum norm interpolation explains why minimization may either benefit or suffer from over-parameterization. As special cases this theory,...