Institution
University of Texas at Austin
Education•Austin, Texas, United States•
About: University of Texas at Austin is a education organization based out in Austin, Texas, United States. It is known for research contribution in the topics: Population & Poison control. The organization has 94352 authors who have published 206297 publications receiving 9070052 citations. The organization is also known as: UT-Austin & UT Austin.
Topics: Population, Poison control, Galaxy, Stars, Finite element method
Papers published on a yearly basis
Papers
More filters
••
Memorial Sloan Kettering Cancer Center1, University of Minnesota2, Medical College of Wisconsin3, University of Utah4, University of Texas at Austin5, University of California, Los Angeles6, Fred Hutchinson Cancer Research Center7, University of Kentucky8, Gulf Coast Regional Blood Center9, University of British Columbia10, Stanford University11
TL;DR: The National Marrow Donor Program has benefited a substantial number of patients in need of marrow transplants from closely HLA-matched unrelated donors and has facilitated the recruitment of unrelated donors into the donor pool and the access to suitable marrow.
Abstract: Background and Methods Allogeneic bone marrow transplantation is curative in a substantial number of patients with hematologic cancers, marrow-failure disorders, immunodeficiency syndromes, and certain metabolic diseases. Unfortunately, only 25 to 30 percent of potential recipients have HLA-identical siblings who can act as donors. In 1986 the National Marrow Donor Program was created in the United States to facilitate the finding and procurement of suitable marrow from unrelated donors for patients lacking related donors. Results During the first four years of the program, 462 patients with acquired and congenital lymphohematopoietic disorders or metabolic diseases received marrow transplants from unrelated donors. The probability of engraftment by 100 days after transplantation was 94 percent, although 8 percent of patients later had secondary graft failure. The probability of grade II, III, or IV acute graft-versus-host disease was 64 percent, and the probability of chronic graft-versus-host disease at...
833 citations
••
TL;DR: In this paper, the authors study the moderating role of uncertainty on top management team (TMT) demographic effects and find that TMT international experience, educational heterogeneity, and tenure heterogeneity are positively related to firms' global strategic postures, and functional heterogeneity exhibited a negative association.
Abstract: The complexity surrounding globalization offers a unique context in which to study the moderating role of uncertainty on top management team (TMT) demographic effects. In a sample of United States-based industrial firms, TMT international experience, educational heterogeneity, and tenure heterogeneity were positively related to firms' global strategic postures, and functional heterogeneity exhibited a negative association. However, when the level of uncertainty facing TMTs was accounted for, these associations were found to be nonlinear.
833 citations
••
TL;DR: In this article, photospheric abundances for 27 elements from carbon to europium in 181 F and G dwarfs from a differential local thermodynamic equilibrium (LTE) analysis of high-resolution and high signal-to-noise ratio spectra were presented.
Abstract: Photospheric abundances are presented for 27 elements from carbon to europium in 181 F and G dwarfs from a differential local thermodynamic equilibrium (LTE) analysis of high-resolution and high signal-to-noise ratio spectra. Stellar effective temperatures (T eff) were adopted from an infrared flux method calibration of Stromgren photometry. Stellar surface gravities (g) were calculated from Hipparcos parallaxes and stellar evolutionary tracks. Adopted T eff and g values are in good agreement with spectroscopic estimates. Stellar ages were determined from evolutionary tracks. Stellar space motions (U , V , W ) and a Galactic potential were used to estimate Galactic orbital parameters. These show that the vast majority of the stars belong to the Galactic thin disc. Relative abundances expressed as (X/Fe) generally confirm previously published results. We give results for C, N, O, Na, Mg, Al, Si, S, K, Ca, Sc, Ti, V, Cr, Mn, Co, Ni, Cu, Zn, Sr, Y, Zr, Ba, Ce, Nd and Eu. The α elements - O, Mg, Si, Ca and Ti - show (α/Fe) to increase slightly with decreasing (Fe/H). Heavy elements with dominant contributions at solar metallicity from the s-process show (s/Fe) to decrease slightly with decreasing (Fe/H). Scatter in (X/Fe) at a fixed (Fe/H) is entirely attributable to the small measurement errors, after excluding the few thick disc stars and the s-process-enriched CH subgiants. Tight limits are set on 'cosmic' scatter. If a weak trend with (Fe/H) is taken into account, the composition of a thin disc star expressed as (X/Fe) is independent of the star's age and birthplace for elements contributed in different proportions by massive stars (Type II supernovae), exploding white dwarfs (Type Ia supernovae) and asymptotic red giant branch stars. By combining our sample with various published studies, comparisons between thin and thick disc stars are made. In this composite sample, thick disc stars are primarily identified by their V LSR in the range −40 to −100 km s −1 . These are very old stars with origins in the inner Galaxy and metallicities (Fe/H) −0.4. At the same (Fe/H), the sampled thin disc stars have V LSR ∼ 0k m s −1 , and are generally younger with a birthplace at about the Sun's Galactocentric distance. In the range −0.35 (Fe/H) −0.70, well represented by present thin and thick disc samples, (X/Fe) of the thick disc stars is greater than that of thin disc stars for Mg, Al, Si, Ca, Ti and Eu. (X/Fe) is very similar for the thin and thick disc for - notably - Na and iron-group elements. Barium ((Ba/Fe)) may be underabundant in thick relative to thin disc stars. These results extend previous ideas about composition differences between the thin and thick disc.
832 citations
••
University of Oxford1, Broad Institute2, Swiss Federal Institute of Aquatic Science and Technology3, University of Bern4, Wellcome Trust Sanger Institute5, Wellcome Trust/Cancer Research UK Gurdon Institute6, University of Konstanz7, Agency for Science, Technology and Research8, Reed College9, Stanford University10, California Institute of Technology11, Benaroya Research Institute12, University of Rennes13, Georgia Institute of Technology14, University of Maryland, College Park15, University of Basel16, University of Texas at Austin17, Tokyo Institute of Technology18, National Museum of Natural History19, University of Stirling20, Carnegie Institution for Science21, National Cheng Kung University22, Science for Life Laboratory23, Norwich University24
TL;DR: This article found an excess of gene duplications in the East African lineage compared to Nile tilapia and other teleosts, an abundance of non-coding element divergence, accelerated coding sequence evolution, expression divergence associated with transposable element insertions, and regulation by novel microRNAs.
Abstract: Cichlid fishes are famous for large, diverse and replicated adaptive radiations in the Great Lakes of East Africa. To understand the molecular mechanisms underlying cichlid phenotypic diversity, we sequenced the genomes and transcriptomes of five lineages of African cichlids: the Nile tilapia (Oreochromis niloticus), an ancestral lineage with low diversity; and four members of the East African lineage: Neolamprologus brichardi/pulcher (older radiation, Lake Tanganyika), Metriaclima zebra (recent radiation, Lake Malawi), Pundamilia nyererei (very recent radiation, Lake Victoria), and Astatotilapia burtoni (riverine species around Lake Tanganyika). We found an excess of gene duplications in the East African lineage compared to tilapia and other teleosts, an abundance of non-coding element divergence, accelerated coding sequence evolution, expression divergence associated with transposable element insertions, and regulation by novel microRNAs. In addition, we analysed sequence data from sixty individuals representing six closely related species from Lake Victoria, and show genome-wide diversifying selection on coding and regulatory variants, some of which were recruited from ancient polymorphisms. We conclude that a number of molecular mechanisms shaped East African cichlid genomes, and that amassing of standing variation during periods of relaxed purifying selection may have been important in facilitating subsequent evolutionary diversification.
832 citations
•
TL;DR: This work proves why stochastic gradient descent can find global minima on the training objective of DNNs in $\textit{polynomial time}$ and implies an equivalence between over-parameterized neural networks and neural tangent kernel (NTK) in the finite (and polynomial) width setting.
Abstract: Deep neural networks (DNNs) have demonstrated dominating performance in many fields; since AlexNet, networks used in practice are going wider and deeper. On the theoretical side, a long line of works has been focusing on training neural networks with one hidden layer. The theory of multi-layer networks remains largely unsettled.
In this work, we prove why stochastic gradient descent (SGD) can find $\textit{global minima}$ on the training objective of DNNs in $\textit{polynomial time}$. We only make two assumptions: the inputs are non-degenerate and the network is over-parameterized. The latter means the network width is sufficiently large: $\textit{polynomial}$ in $L$, the number of layers and in $n$, the number of samples.
Our key technique is to derive that, in a sufficiently large neighborhood of the random initialization, the optimization landscape is almost-convex and semi-smooth even with ReLU activations. This implies an equivalence between over-parameterized neural networks and neural tangent kernel (NTK) in the finite (and polynomial) width setting.
As concrete examples, starting from randomly initialized weights, we prove that SGD can attain 100% training accuracy in classification tasks, or minimize regression loss in linear convergence speed, with running time polynomial in $n,L$. Our theory applies to the widely-used but non-smooth ReLU activation, and to any smooth and possibly non-convex loss functions. In terms of network architectures, our theory at least applies to fully-connected neural networks, convolutional neural networks (CNN), and residual neural networks (ResNet).
832 citations
Authors
Showing all 95138 results
Name | H-index | Papers | Citations |
---|---|---|---|
George M. Whitesides | 240 | 1739 | 269833 |
Eugene Braunwald | 230 | 1711 | 264576 |
Yi Chen | 217 | 4342 | 293080 |
Robert J. Lefkowitz | 214 | 860 | 147995 |
Joseph L. Goldstein | 207 | 556 | 149527 |
Eric N. Olson | 206 | 814 | 144586 |
Hagop M. Kantarjian | 204 | 3708 | 210208 |
Rakesh K. Jain | 200 | 1467 | 177727 |
Francis S. Collins | 196 | 743 | 250787 |
Gordon B. Mills | 187 | 1273 | 186451 |
Scott M. Grundy | 187 | 841 | 231821 |
Michael S. Brown | 185 | 422 | 123723 |
Eric Boerwinkle | 183 | 1321 | 170971 |
Aaron R. Folsom | 181 | 1118 | 134044 |
Jiaguo Yu | 178 | 730 | 113300 |