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

  • Home
  • Search Results
  • Page 1 of 17

Search for: All records

Creators/Authors contains: "Lee, Y.-J."

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. ; ; ; ( , Proceedings of the British Machine Vision Conference (BMVC), 2020)
    Aliasing refers to the phenomenon that high frequency signals degenerate into com- pletely different ones after sampling. It arises as a problem in the context of deep learning as downsampling layers are widely adopted in deep architectures to reduce parameters and computation. The standard solution is to apply a low-pass filter (e.g., Gaussian blur) before downsampling [37]. However, it can be suboptimal to apply the same filter across the entire content, as the frequency of feature maps can vary across both spatial locations and feature channels. To tackle this, we propose an adaptive content-aware low-pass filtering layer, which predicts separatemore »filter weights for each spatial location and chan- nel group of the input feature maps. We investigate the effectiveness and generalization of the proposed method across multiple tasks including ImageNet classification, COCO instance segmentation, and Cityscapes semantic segmentation. Qualitative and quanti- tative results demonstrate that our approach effectively adapts to the different feature frequencies to avoid aliasing while preserving useful information for recognition. Code is available at https://maureenzou.github.io/ddac/.« less
    Accepted Manuscript Full Text Available
  2. ; ; ; ( , Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR))
    Accepted Manuscript Full Text Available
  3. ; ; ; ; ; ( , Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR))
    Full Text Available 
  4. ; ; ( , IEEE Winter Conference on Applications of Computer Vision (WACV))
    Accepted Manuscript Full Text Available
  5. ; ; ( , Proceedings of the IEEEInternational Conference on Computer Vision (ICCV))
    Full Text Available 
  6. ; ; ; ( , Proceedings of the IEEE International Conference on Computer Vision (ICCV))
    Full Text Available 
  7. ; ; ; ; ; ; ; ; ; ; et al ( , Astronomy & Astrophysics)
    Since January 2012, we have been monitoring the behavior of sulfur dioxide and water on Venus, using the Texas Echelon Cross-Echelle Spectrograph imaging spectrometer at the NASA InfraRed Telescope Facility (IRTF, Mauna Kea Observatory). Here, we present new data recorded in February and April 2019 in the 1345 cm −1 (7.4 μ m) spectral range, where SO 2 , CO 2 , and HDO (used as a proxy for H 2 O) transitions were observed. The cloud top of Venus was probed at an altitude of about 64 km. As in our previous studies, the volume mixing ratio (vmr) ofmore »SO 2 was estimated using the SO 2 /CO 2 line depth ratio of weak transitions; the H 2 O volume mixing ratio was derived from the HDO/CO 2 line depth ratio, assuming a D/H ratio of 200 times the Vienna standard mean ocean water. As reported in our previous analyses, the SO 2 mixing ratio shows strong variations with time and also over the disk, showing evidence for the formation of SO 2 plumes with a lifetime of a few hours; in contrast, the H 2 O abundance is remarkably uniform over the disk and shows moderate variations as a function of time. We have used the 2019 data in addition to our previous dataset to study the long-term variations of SO 2 and H 2 O. The data reveal a long-term anti-correlation with a correlation coefficient of −0.80; this coefficient becomes −0.90 if the analysis is restricted to the 2014–2019 time period. The statistical analysis of the SO 2 plumes as a function of local time confirms our previous result with a minimum around 10:00 and two maxima near the terminators. The dependence of the SO 2 vmr with respect to local time shows a higher abundance at the evening terminator with respect to the morning. The dependence of the SO 2 vmr with respect to longitude exhibits a broad maximum at 120–200° east longitudes, near the region of Aphrodite Terra. However, this trend has not been observed by other measurements and has yet to be confirmed.« less
    Full Text Available 
  8. ; ; ; ; ; ; ; ; ; ; et al ( , Journal of High Energy Physics)
    A bstract A search for a heavy resonance decaying into a top quark and a W boson in proton-proton collisions at $$ \sqrt{s} $$ s = 13 TeV is presented. The data analyzed were recorded with the CMS detector at the LHC and correspond to an integrated luminosity of 138 fb − 1 . The top quark is reconstructed as a single jet and the W boson, from its decay into an electron or muon and the corresponding neutrino. A top quark tagging technique based on jet clustering with a variable distance parameter and simultaneous jet grooming is used tomore »identify jets from the collimated top quark decay. The results are interpreted in the context of two benchmark models, where the heavy resonance is either an excited bottom quark b ∗ or a vector-like quark B. A statistical combination with an earlier search by the CMS Collaboration in the all-hadronic final state is performed to place upper cross section limits on these two models. The new analysis extends the lower range of resonance mass probed from 1.4 down to 0.7 TeV. For left-handed, right-handed, and vector-like couplings, b ∗ masses up to 3.0, 3.0, and 3.2 TeV are excluded at 95% confidence level, respectively. The observed upper limits represent the most stringent constraints on the b ∗ model to date.« less
    Free, publicly-accessible full text available April 1, 2023
  9. ; ; ; ; ; ; ; ; ; ; et al ( , Physical Review Letters)
    Free, publicly-accessible full text available March 1, 2023
  10. ; ; ; ; ; ; ; ; ; ; et al ( , Physical Review Letters)
    Free, publicly-accessible full text available January 1, 2023