We introduce a generative model of part‐segmented 3D objects: the shape variational auto‐encoder (ShapeVAE). The ShapeVAE describes a joint distribution over the existence of object parts, the ...locations of a dense set of surface points, and over surface normals associated with these points. Our model makes use of a deep encoder‐decoder architecture that leverages the part‐decomposability of 3D objects to embed high‐dimensional shape representations and sample novel instances. Given an input collection of part‐segmented objects with dense point correspondences the ShapeVAE is capable of synthesizing novel, realistic shapes, and by performing conditional inference enables imputation of missing parts or surface normals. In addition, by generating both points and surface normals, our model allows for the use of powerful surface‐reconstruction methods for mesh synthesis. We provide a quantitative evaluation of the ShapeVAE on shape‐completion and test‐set log‐likelihood tasks and demonstrate that the model performs favourably against strong baselines. We demonstrate qualitatively that the ShapeVAE produces plausible shape samples, and that it captures a semantically meaningful shape‐embedding. In addition we show that the ShapeVAE facilitates mesh reconstruction by sampling consistent surface normals.
Gaussian processes (GPs) provide a principled, practical, probabilistic approach to learning in kernel machines. GPs have received increased attention in the machine-learning community over the past ...decade, and this book provides a long-needed systematic and unified treatment of theoretical and practical aspects of GPs in machine learning. The treatment is comprehensive and self-contained, targeted at researchers and students in machine learning and applied statistics.The book deals with the supervised-learning problem for both regression and classification, and includes detailed algorithms. A wide variety of covariance (kernel) functions are presented and their properties discussed. Model selection is discussed both from a Bayesian and a classical perspective. Many connections to other well-known techniques from machine learning and statistics are discussed, including support-vector machines, neural networks, splines, regularization networks, relevance vector machines and others. Theoretical issues including learning curves and the PAC-Bayesian framework are treated, and several approximation methods for learning with large datasets are discussed. The book contains illustrative examples and exercises, and code and datasets are available on the Web. Appendixes provide mathematical background and a discussion of Gaussian Markov processes.
Context. Deuterium fractionation has been used to study the thermal history of prestellar environments. Their formation pathways trace different regions of the disk and may shed light into the ...physical structure of the disk, including locations of important features for planetary formation. Aims. We aim to constrain the radial extent of the main deuterated species; we are particularly interested in spatially characterizing the high and low temperature pathways for enhancing deuteration of these species. Methods. We observed the disk surrounding the Herbig Ae star HD 163296 using ALMA in Band 6 and obtained resolved spectral imaging data of DCO+ (J = 3 − 2), DCN (J = 3 − 2) and N2D+ (J = 3 − 2) with synthesized beam sizes of 0.̋53 × 0.̋42, 0.̋53 × 0.̋42, and 0.̋50 × 0.̋39, respectively. We adopted a physical model of the disk from the literature and use the 3D radiative transfer code LIME to estimate an excitation temperature profile for our detected lines. We modeled the radial emission profiles of DCO+, DCN, and N2D+, assuming their emission is optically thin, using a parametric model of their abundances and our excitation temperature estimates. Results. DCO+ can be described by a three-region model with constant-abundance rings centered at 70 AU, 150 AU, and 260 AU. The DCN radial profile peaks at about 60 AU and N2D+ is seen in a ring at 160 AU. Simple models of both molecules using constant abundances reproduce the data. Assuming reasonable average excitation temperatures for the whole disk, their disk-averaged column densities (and deuterium fractionation ratios) are 1.6–2.6×1012 cm-2 (0.04–0.07), 2.9–5.2×1012 cm-2 (~0.02), and 1.6–2.5×1011 cm-2 (0.34–0.45) for DCO+, DCN, and N2D+, respectively. Conclusions. Our simple best-fit models show a correlation between the radial location of the first two rings in DCO+ and the DCN and N2D+ abundance distributions that can be interpreted as the high and low temperature deuteration pathways regimes. The origin of the third DCO+ ring at 260 AU is unknown but may be due to a local decrease of ultraviolet opacity allowing the photodesorption of CO or due to thermal desorption of CO as a consequence of radial drift and settlement of dust grains. The derived Df values agree with previous estimates of 0.05 for DCO+/HCO+ and 0.02 for DCN/HCN in HD 163296, and 0.3−0.5 for N2D+/N2H+ in AS 209, a T Tauri disk. The high N2D+/N2H+ confirms N2D+ as a good candidate for tracing ionization in the cold outer disk.
A fourth production region for the globally important Antarctic bottom water has been attributed to dense shelf water formation in the Cape Darnley Polynya, adjoining Prydz Bay in East Antarctica. ...Here we show new observations from CTD-instrumented elephant seals in 2011-2013 that provide the first complete assessment of dense shelf water formation in Prydz Bay. After a complex evolution involving opposing contributions from three polynyas (positive) and two ice shelves (negative), dense shelf water (salinity 34.65-34.7) is exported through Prydz Channel. This provides a distinct, relatively fresh contribution to Cape Darnley bottom water. Elsewhere, dense water formation is hindered by the freshwater input from the Amery and West Ice Shelves into the Prydz Bay Gyre. This study highlights the susceptibility of Antarctic bottom water to increased freshwater input from the enhanced melting of ice shelves, and ultimately the potential collapse of Antarctic bottom water formation in a warming climate.
We present a new prospective analysis of deep multi-band imaging with the
James Webb
Space Telescope (JWST). In this work, we investigate the recovery of high-redshift 5 <
z
< 12 galaxies through ...extensive image simulations of accepted JWST programs, including the Early Release Science in the EGS field and the Guaranteed Time Observations in the HUDF. We introduced complete samples of ∼300 000 galaxies with stellar masses of log(
M
*
/
M
⊙
) > 6 and redshifts of 0 <
z
< 15, as well as galactic stars, into realistic mock NIRCam, MIRI, and HST images to properly describe the impact of source blending. We extracted the photometry of the detected sources, as in real images, and estimated the physical properties of galaxies through spectral energy distribution fitting. We find that the photometric redshifts are primarily limited by the availability of blue-band and near-infrared medium-band imaging. The stellar masses and star formation rates are recovered within 0.25 and 0.3 dex, respectively, for galaxies with accurate photometric redshifts. Brown dwarfs contaminating the
z
> 5 galaxy samples can be reduced to < 0.01 arcmin
−2
with a limited impact on galaxy completeness. We investigate multiple high-redshift galaxy selection techniques and find that the best compromise between completeness and purity at 5 <
z
< 10 using the full redshift posterior probability distributions. In the EGS field, the galaxy completeness remains higher than 50% at magnitudes
m
UV
< 27.5 and at all redshifts, and the purity is maintained above 80 and 60% at
z
≤ 7 and 10, respectively. The faint-end slope of the galaxy UV luminosity function is recovered with a precision of 0.1–0.25, and the cosmic star formation rate density within 0.1 dex. We argue in favor of additional observing programs covering larger areas to better constrain the bright end.
Greedy Learning of Binary Latent Trees Harmeling, S; Williams, C K I
IEEE transactions on pattern analysis and machine intelligence,
06/2011, Letnik:
33, Številka:
6
Journal Article
Recenzirano
Odprti dostop
Inferring latent structures from observations helps to model and possibly also understand underlying data generating processes. A rich class of latent structures is the latent trees, i.e., ...tree-structured distributions involving latent variables where the visible variables are leaves. These are also called hierarchical latent class (HLC) models. Zhang and Kočka CHECK END OF SENTENCE proposed a search algorithm for learning such models in the spirit of Bayesian network structure learning. While such an approach can find good solutions, it can be computationally expensive. As an alternative, we investigate two greedy procedures: The BIN-G algorithm determines both the structure of the tree and the cardinality of the latent variables in a bottom-up fashion. The BIN-A algorithm first determines the tree structure using agglomerative hierarchical clustering, and then determines the cardinality of the latent variables as for BIN-G. We show that even with restricting ourselves to binary trees, we obtain HLC models of comparable quality to Zhang's solutions (in terms of cross-validated log-likelihood), while being generally faster to compute. This claim is validated by a comprehensive comparison on several data sets. Furthermore, we demonstrate that our methods are able to estimate interpretable latent structures on real-world data with a large number of variables. By applying our method to a restricted version of the 20 newsgroups data, these models turn out to be related to topic models, and on data from the PASCAL Visual Object Classes (VOC) 2007 challenge, we show how such tree-structured models help us understand how objects co-occur in images. For reproducibility of all experiments in this paper, all code and data sets (or links to data) are available at http://people.kyb.tuebingen.mpg.de/harmeling/code/ltt-1.4.tar.
Summary
We re‐analysed prospective data collected by anaesthetists in the Anaesthesia Sprint Audit of Practice (ASAP‐1) to describe associations with linked outcome data. Mortality was 165/11,085 ...(1.5%) 5 days and 563/11,085 (5.1%) 30 days after surgery and was not associated with anaesthetic technique (general vs. spinal, with or without peripheral nerve blockade). The risk of death increased as blood pressure fell: the odds ratio (95% CI) for mortality within five days after surgery was 0.983 (0.973–0.994) for each 5 mmHg intra‐operative increment in systolic blood pressure, p = 0.0016, and 0.980 (0.967–0.993) for each mmHg increment in mean pressure, p = 0.0039. The equivalent odds ratios (95% CI) for 30‐day mortality were 0.968 (0.951–0.985), p = 0.0003 and 0.976 (0.964–0.988), p = 0.0001, respectively. The lowest systolic blood pressure after intrathecal local anaesthetic relative to before induction was weakly correlated with a higher volume of subarachnoid bupivacaine: r2 −0.10 and −0.16 for hyperbaric and isobaric bupivacaine, respectively. A mean 20% relative fall in systolic blood pressure correlated with an administered volume of 1.44 ml hyperbaric bupivacaine. Future research should focus on refining standardised anaesthesia towards administering lower doses of spinal (and general) anaesthesia and maintaining normotension.
☛ CPD available at http://www.learnataagbi.org
This paper describes the observations and the first data release (DR1) of the ESO public spectroscopic survey “VANDELS, a deep VIMOS survey of the CANDELS CDFS and UDS fields”. The main targets of ...VANDELS are star-forming galaxies at redshift 2.4 < z < 5.5, an epoch when the Universe had not yet reached 20% of its current age, and massive passive galaxies in the range 1 < z < 2.5. By adopting a strategy of ultra-long exposure times, ranging from a minimum of 20 h to a maximum of 80 h per source, VANDELS is specifically designed to be the deepest-ever spectroscopic survey of the high-redshift Universe. Exploiting the red sensitivity of the refurbished VIMOS spectrograph, the survey is obtaining ultra-deep optical spectroscopy covering the wavelength range 4800–10 000 Å with a sufficiently high signal-to-noise ratio to investigate the astrophysics of high-redshift galaxy evolution via detailed absorption line studies of well-defined samples of high-redshift galaxies. VANDELS-DR1 is the release of all medium-resolution spectroscopic data obtained during the first season of observations, on a 0.2 square degree area centered around the CANDELS-CDFS (Chandra deep-field south) and CANDELS-UDS (ultra-deep survey) areas. It includes data for all galaxies for which the total (or half of the total) scheduled integration time was completed. The DR1 contains 879 individual objects, approximately half in each of the two fields, that have a measured redshift, with the highest reliable redshifts reaching zspec ~ 6. In DR1 we include fully wavelength-calibrated and flux-calibrated 1D spectra, the associated error spectrum and sky spectrum, and the associated wavelength-calibrated 2D spectra. We also provide a catalog with the essential galaxy parameters, including spectroscopic redshifts and redshift quality flags measured by the collaboration. We present the survey layout and observations, the data reduction and redshift measurement procedure, and the general properties of the VANDELS-DR1 sample. In particular, we discuss the spectroscopic redshift distribution and the accuracy of the photometricredshifts for each individual target category, and we provide some examples of data products for the various target typesand the different quality flags. All VANDELS-DR1 data are publicly available and can be retrieved from the ESO archive. Two further data releases are foreseen in the next two years, and a final data release is currently scheduled for June 2020, which will include an improved rereduction of the entire spectroscopic data set.
Endoscopic sensing of alveolar pH Choudhury, D; Tanner, M G; McAughtrie, S ...
Biomedical optics express,
01/2017, Letnik:
8, Številka:
1
Journal Article
Recenzirano
Odprti dostop
Previously unobtainable measurements of alveolar pH were obtained using an endoscope-deployable optrode. The pH sensing was achieved using functionalized gold nanoshell sensors and surface enhanced ...Raman spectroscopy (SERS). The optrode consisted of an asymmetric dual-core optical fiber designed for spatially separating the optical pump delivery and signal collection, in order to circumvent the unwanted Raman signal generated within the fiber. Using this approach, we demonstrate a ~100-fold increase in SERS signal-to-fiber background ratio, and demonstrate multiple site pH sensing with a measurement accuracy of ± 0.07 pH units in the respiratory acini of an
ovine lung model. We also demonstrate that alveolar pH changes in response to ventilation.
GTM: The Generative Topographic Mapping Bishop, Christopher M.; Svensén, Markus; Williams, Christopher K. I.
Neural computation,
01/1998, Letnik:
10, Številka:
1
Journal Article
Recenzirano
Odprti dostop
Latent variable models represent the probability density of data in a space of several dimensions in terms of a smaller number of latent, or hidden, variables. A familiar example is factor analysis, ...which is based on a linear transformation between the latent space and the data space. In this article, we introduce a form of nonlinear latent variable model called the generative topographic mapping, for which the parameters of the model can be determined using the expectation-maximization algorithm. GTM provides a principled alternative to the widely used self-organizing map (SOM) of Kohonen (1982) and overcomes most of the significant limitations of the SOM. We demonstrate the performance of the GTM algorithm on a toy problem and on simulated data from flow diagnostics for a multiphase oil pipeline.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
PDF
