skip to main content


Search for: All records

Creators/Authors contains: "Zhu, W."

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Free, publicly-accessible full text available November 20, 2024
  2. Optical metasurfaces of sub-wavelength pillars have provided new capabilities for the versatile definition of the amplitude, phase, and polarization of light. In this work we demonstrate that an efficient dielectric metasurface lens can be used to trap and image single neutral atoms. We characterize the high numerical aperture optical tweezers using the trapped atoms and compare to numerical computations of the metasurface lens performance. We predict future metasurfaces for atom trapping can leverage multiple ongoing developments in metasurface design and enable multifunctional control in complex experiments with neutral-atoms arrays. 
    more » « less
  3. In 2019, University of Houston (UH) at Houston, Texas was awarded an NSF Research Experience for Teachers (RET) site grant titled “RET Site: High School Teacher Experience in Engineering Design and Manufacturing.” The goal of the project is to host 12 high school teachers each summer to participate in engineering design and manufacturing research and then convert their experience into high school curriculum. In summer of 2021, the first cohort of 12 teachers from Region 4 of Southeast Texas participated in the RET program at UH College of Technology (COT). This six-week program, open to local high school STEM teachers in Texas, sought to advance educators’ knowledge of concepts in design and manufacturing as a means of enriching high school curriculums and meeting foundational standards set by 2013’s Texas House Bill 5. These standards require enhanced STEM contents in high school curricula as a prerequisite for graduation, detailed in the Texas Essential Knowledge and Skills standard. Due to the pandemic situation, about 50% of the activities are online and the rest are face to face. About 40% of the time, teachers attended online workshops to enhance their knowledge of topics in engineering design and manufacturing before embarking on applicable research projects in the labs. Six UH COT engineering technology professors each led workshops in a week. The four tenure-track engineering mentors, assisted by student research assistants, each mentored three teachers on projects ranging from additive manufacturing to thermal/fluids, materials, and energy. The group also participated in field trips to local companies including ARC Specialties, Master Flo, Re:3D, and Forged Components. They worked with two instructional track engineering technology professors and one professor of education on applying their learnings to lesson plan design. Participants also met weekly for online Brown Bag teacher seminars to share their experiences and discuss curricula, which was organized by the RET master teacher. On the final day of the program, the teachers presented their curriculum prototype for the fall semester to the group and received completion certificates. The program assessment was led by the assessment specialist, Director of Assessment and Accreditation at UH COT. Teacher participants found the research experience with their mentors beneficial not only to them, but also to their students according to our findings from interviews. The mentors will visit their mentees’ classrooms to see the lesson plans being implemented. In the spring of 2022, the teachers will present their refined curricula at a RET symposium to be organized at UH and submit their standards-aligned plans to teachengineering.org for other K-12 educators to access. 
    more » « less
  4. Incorporating group symmetry directly into the learning process has proved to be an effective guideline for model design. By producing features that are guaranteed to transform covariantly to the group actions on the inputs, group-equivariant convolutional neural net- works (G-CNNs) achieve significantly improved generalization performance in learning tasks with intrinsic symmetry. General theory and practical implementation of G-CNNs have been studied for planar images under either rotation or scaling transformation, but only individu- ally. We present, in this paper, a roto-scale-translation equivariant CNN (RST-CNN), that is guaranteed to achieve equivariance jointly over these three groups via coupled group con- volutions. Moreover, as symmetry transformations in reality are rarely perfect and typically subject to input deformation, we provide a stability analysis of the equivariance of representation to input distortion, which motivates the truncated expansion of the convolutional filters under (pre-fixed) low-frequency spatial modes. The resulting model provably achieves deformation-robust RS T equivariance, i.e., the RST symmetry is still “approximately” preserved when the transformation is “contaminated” by a nuisance data deformation, a property that is especially important for out-of-distribution generalization. Numerical experiments on MNIST, Fashion-MNIST, and STL-10 demonstrate that the proposed model yields remarkable gains over prior arts, especially in the small data regime where both rotation and scaling variations are present within the data. 
    more » « less
  5. Incorporating group symmetry directly into the learning process has proved to be an effective guideline for model design. By producing features that are guaranteed to transform covariantly to the group actions on the inputs, group-equivariant convolutional neural networks (G-CNNs) achieve significantly improved generalization performance in learning tasks with intrinsic symmetry. General theory and practical implementation of G-CNNs have been studied for planar images under either rotation or scaling transformation, but only individually. We present, in this paper, a roto-scale-translation equivariant CNN (RST-CNN), that is guaranteed to achieve equivariance jointly over these three groups via coupled group convolutions. Moreover, as symmetry transformations in reality are rarely perfect and typically subject to input deformation, we provide a stability analysis of the equivariance of representation to input distortion, which motivates the truncated expansion of the convolutional filters under (pre-fixed) low-frequency spatial modes. The resulting model provably achieves deformation-robust RST-equivariance, i.e., the RST-symmetry is still "approximately” preserved when the transformation is "contaminated” by a nuisance data deformation, a property that is especially important for out-of-distribution generalization. Numerical experiments on MNIST, Fashion-MNIST, and STL-10 demonstrate that the proposed model yields remarkable gains over prior arts, especially in the small data regime where both rotation and scaling variations are present within the data. 
    more » « less
  6. Generative adversarial networks (GANs), a class of distribution-learning methods based on a two-player game between a generator and a discriminator, can generally be formulated as a minmax problem based on the variational representation of a divergence between the unknown and the generated distributions. We introduce structure-preserving GANs as a data-efficient framework for learning distributions with additional structure such as group symmetry, by developing new variational representations for divergences. Our theory shows that we can reduce the discriminator space to its projection on the invariant discriminator space, using the conditional expectation with respect to the sigma-algebra associated to the underlying structure. In addition, we prove that the discriminator space reduction must be accompanied by a careful design of structured generators, as flawed designs may easily lead to a catastrophic “mode collapse” of the learned distribution. We contextualize our framework by building symmetry-preserving GANs for distributions with intrinsic group symmetry, and demonstrate that both players, namely the equivariant generator and invariant discriminator, play important but distinct roles in the learning process. Empirical experiments and ablation studies across a broad range of data sets, including real-world medical imaging, validate our theory, and show our proposed methods achieve significantly improved sample fidelity and diversity—almost an order of magnitude measured in Frechet Inception Distance—especially in the small data regime 
    more » « less
  7. Generative adversarial networks (GANs), a class of distribution-learning methods based on a two-player game between a generator and a discriminator, can generally be formulated as a minmax problem based on the variational representation of a divergence between the unknown and the generated distributions. We introduce structure-preserving GANs as a data-efficient framework for learning distributions with additional structure such as group symmetry, by developing new variational representations for divergences. Our theory shows that we can reduce the discriminator space to its projection on the invariant discriminator space, using the conditional expectation with respect to the sigma-algebra associated to the underlying structure. In addition, we prove that the discriminator space reduction must be accompanied by a careful design of structured generators, as flawed designs may easily lead to a catastrophic “mode collapse” of the learned distribution. We contextualize our framework by building symmetry-preserving GANs for distributions with intrinsic group symmetry, and demonstrate that both players, namely the equivariant generator and invariant discriminator, play important but distinct roles in the learning process. Empirical experiments and ablation studies across a broad range of data sets, including real-world medical imaging, validate our theory, and show our proposed methods achieve significantly improved sample fidelity and diversity—almost an order of magnitude measured in Frechet Inception Distance—especially in the small data regime. 
    more » « less