scholarly journals Cross-Dimensional Weighting for Aggregated Deep Convolutional Features

Author(s):  
Yannis Kalantidis ◽  
Clayton Mellina ◽  
Simon Osindero
2019 ◽  
Author(s):  
Simon De Deyne ◽  
Wouter Voorspoels ◽  
Steven Verheyen ◽  
Danielle Navarro ◽  
Gert Storms

In contrast to noun categories, little is known about the graded structure of adjective categories. In this study, we investigated whether adjective categories show a similar graded structure and what determines this structure. The results show that adjective categories like nouns exhibit a reliable graded structure. Similar to nouns, we investigated whether similarity is the main determinant of the graded structure. We derived a low-dimensional similarity representation for adjective categories and found that valence differences in adjectives constitute an important organising principle in this similarity space. Valence was not implicated in the categories' graded structure, however. A formal similarity-based model using exemplars accounted for the graded structure by effectively discarding the valence differences between adjectives in the similarity representation through dimensional weighting. Our results generalise similarity-based accounts of graded structure and highlight a closely knit relationship between adjectives and nouns on a representational level.


2008 ◽  
Vol 65 (12) ◽  
pp. 3695-3718 ◽  
Author(s):  
Dong L. Wu ◽  
Stephen D. Eckermann

Abstract The gravity wave (GW)–resolving capabilities of 118-GHz saturated thermal radiances acquired throughout the stratosphere by the Microwave Limb Sounder (MLS) on the Aura satellite are investigated and initial results presented. Because the saturated (optically thick) radiances resolve GW perturbations from a given altitude at different horizontal locations, variances are evaluated at 12 pressure altitudes between ∼21 and 51 km using the 40 saturated radiances found at the bottom of each limb scan. Forward modeling simulations show that these variances are controlled mostly by GWs with vertical wavelengths λz > 5 km and horizontal along-track wavelengths of λy ∼ 100–200 km. The tilted cigar-shaped three-dimensional weighting functions yield highly selective responses to GWs of high intrinsic frequency that propagate toward the instrument. The latter property is used to infer the net meridional component of GW propagation by differencing the variances acquired from ascending (A) and descending (D) orbits. Because of improved vertical resolution and sensitivity, Aura MLS GW variances are ∼5–8 times larger than those from the Upper Atmosphere Research Satellite (UARS) MLS. Like UARS MLS variances, monthly-mean Aura MLS variances in January and July 2005 are enhanced when local background wind speeds are large, due largely to GW visibility effects. Zonal asymmetries in variance maps reveal enhanced GW activity at high latitudes due to forcing by flow over major mountain ranges and at tropical and subtropical latitudes due to enhanced deep convective generation as inferred from contemporaneous MLS cloud-ice data. At 21–28-km altitude (heights not measured by the UARS MLS), GW variance in the tropics is systematically enhanced and shows clear variations with the phase of the quasi-biennial oscillation, in general agreement with GW temperature variances derived from radiosonde, rocketsonde, and limb-scan vertical profiles. GW-induced temperature variances at ∼44-km altitude derived from operational global analysis fields of the ECMWF Integrated Forecast System in August 2006 reveal latitudinal bands of enhanced GW variance and preferred GW meridional propagation directions that are similar to those inferred from the MLS variances, highlighting the potential of MLS GW data for validating the stratospheric GWs simulated and/or parameterized in global models.


2013 ◽  
Vol 13 (3) ◽  
pp. 25-25 ◽  
Author(s):  
R. Weidner ◽  
H. J. Muller

2019 ◽  
Vol 363 ◽  
pp. 17-26 ◽  
Author(s):  
Qi Wang ◽  
Jinxiang Lai ◽  
Zhenguo Yang ◽  
Kai Xu ◽  
Peipei Kan ◽  
...  

2007 ◽  
Vol 44 (2) ◽  
pp. 277-292 ◽  
Author(s):  
Klaus Gramann ◽  
Thomas Toellner ◽  
Joseph Krummenacher ◽  
Martin Eimer ◽  
Hermann J. Müller

Author(s):  
Joost C. F. de Winter ◽  
Max Mulder ◽  
M. M. van Paassen ◽  
David A. Abbink ◽  
Peter A. Wieringa

Sign in / Sign up

Export Citation Format

Share Document