skip to main content
Explore Scholarly Publications and Datasets in the NSF-PAR

  • Home
  • Search Results
  • Page 1 of 34

Search for: All records

Creators/Authors contains: "Choi, Y."

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. ; ; ( , bioRxiv)
    Deep neural networks implementing generative models for dimensionality reduction have been extensively used for the visualization and analysis of genomic data. One of their key limitations is lack of interpretability: it is challenging to quantitatively identify which input features are used to construct the embedding dimensions, thus preventing insight into why cells are organized in a particular data visualization, for example. Here we present a scalable, interpretable variational autoencoder (siVAE) that is interpretable by design: it learns feature embeddings that guide the interpretation of the cell embeddings in a manner analogous to factor loadings of factor analysis. siVAE is asmore »powerful and nearly as fast to train as the standard VAE but achieves full interpretability of the embedding dimensions. Using siVAE, we exploit a number of connections between dimensionality reduction and gene network inference to identify gene neighborhoods and gene hubs, without the explicit need for gene network inference. We observe a systematic difference in the gene neighborhoods identified by dimensionality reduction methods and gene network inference algorithms in general, suggesting they provide complementary information about the underlying structure of the gene co-expression network. Finally, we apply siVAE to implicitly learn gene networks for individual iPSC lines and uncover a correlation between neuronal differentiation efficiency and loss of co-expression of several mitochondrial complexes, including NADH dehydrogenase, cytochrome C oxidase, and cytochrome b.« less
    Free, publicly-accessible full text available April 1, 2023
  2. ; ; ; ( , Sustainability)
    University-based makerspaces are receiving increasing attention as promising innovations that may contribute to the development of future engineers. Using a theory of social boundary spaces, we investigated whether the diverse experiences offered at university-based makerspaces may contribute to students’ learning and development of various “soft” or “21st century” skills that go beyond engineering-specific content knowledge. Through interviews with undergraduate student users at two university-based makerspaces in the United States we identified seven different types of boundary spaces (where multiple communities, and the individuals and activities affiliated with those communities, come together). We identified students engaging in the processes of identification,more »reflection, and coordination, which allowed them to make sense of, and navigate, the various boundary spaces they encountered in the makerspaces. These processes provided students with opportunities to engage with, and learn from, individuals and practices affiliated with various communities and disciplines. These opportunities can lead to students’ development of necessary skills to creatively and collaboratively address interdisciplinary socio-scientific problems. We suggest that universitybased makerspaces can offer important developmental experiences for a diverse body of students that may be challenging for a single university department, program, or course to offer. Based on these findings, we recommend university programs and faculty intentionally integrate makerspace activities into undergraduate curricula to support students’ development of skills, knowledge, and practices relevant for engineering as well as 21st century skills more broadly.« less
    Accepted Manuscript Full Text Available
  3. ; ; ; ; ; ; ; ; ; ; et al ( , The Astrophysical Journal Letters)
    Free, publicly-accessible full text available October 1, 2022
  4. ; ; ; ; ; ; ; ; ; ; et al ( , Nature physics)
    Accepted Manuscript Full Text Available
  5. ; ; ; ( , Proceedings of the 34th AAAI Conference on Artificial Intelligence, 2020)
    Full Text Available 
  6. ; ; ; ; ; ; ( , Conference of the American Society of Engineering Education)
    Accepted Manuscript Full Text Available
  7. ; ; ; ; ; ; ( , Zone 1 Conference of the American Society for Engineering Education)
    Building upon our two years of research on the use of makerspaces in undergraduate engineering programs, we engaged in a large-scale data collection from students enrolled in undergraduate engineering preparation programs with affiliated makerspaces established for a minimum of three years. Using web searches, and other sources of information (e.g. references from other researchers or faculty members), we have identified 28 institutions that met our criteria. Working with a third party, we gathered over 574 responses from undergraduate engineering students with makerspace experiences spread across the 28 institutions. To gather our data, we created and validated an online survey withmore »a combination of quantitative and qualitative items. We constructed a survey with subscales aligned with motivation to learn, growth mindset, learning goal orientation, knowledge of engineering as a profession, and belongingness and inclusion, as associated with work within makerspaces. We found significant positive correlations among the variables, positive levels of motivation, growth mindset, knowledge of engineering as a profession, and belongingness. We found differences in levels for gender, engineering majors, and student class standing. We discuss the implications for our findings in the context of undergraduate engineering student learning in makerspaces.« less
    Accepted Manuscript Full Text Available
  8. ; ; ; ; ; ; ; ; ; ; et al ( , Nature Communications)
    Abstract The interconversion of charge and spin currents via spin-Hall effect is essential for spintronics. Energy-efficient and deterministic switching of magnetization can be achieved when spin polarizations of these spin currents are collinear with the magnetization. However, symmetry conditions generally restrict spin polarizations to be orthogonal to both the charge and spin flows. Spin polarizations can deviate from such direction in nonmagnetic materials only when the crystalline symmetry is reduced. Here, we show control of the spin polarization direction by using a non-collinear antiferromagnet Mn 3 GaN, in which the triangular spin structure creates a low magnetic symmetry while maintainingmore »a high crystalline symmetry. We demonstrate that epitaxial Mn 3 GaN/permalloy heterostructures can generate unconventional spin-orbit torques at room temperature corresponding to out-of-plane and Dresselhaus-like spin polarizations which are forbidden in any sample with two-fold rotational symmetry. Our results demonstrate an approach based on spin-structure design for controlling spin-orbit torque, enabling high-efficient antiferromagnetic spintronics.« less
    Full Text Available 
  9. ; ; ; ; ; ; ; ; ; ; et al ( , The Astrophysical Journal)
    Full Text Available 
  10. ; ; ; ( , Future generation computer systems)
    Accepted Manuscript Full Text Available