BZPEERReading, trust and collaboration for researchDiscover
Workspaces
Network
Opportunities
Account
Researcher profile
Mark Reid contributes to research discovery and scholarly infrastructure.
Trust snapshot
Actions
Identity and collaboration
Claiming links this public author record to a researcher profile and unlocks direct collaboration workflows.
Log in to claimDirect collaboration
Claim this author entity first to unlock direct invitations.
Research graph
Inspect adjacent work, topics, institutions and collaborators without jumping out to a separate graph page.
BZPEER is loading the nearby papers, people, topics and institutions for this page.
Published work
This document presents the draft specification for delivering machine learning services over HTTP, developed as part of the Protocols and Structures for Inference project, which concluded in 2013. It presents the motivation for providing machine learning as a service, followed by a description of the essential and optional components of such a service.
We discuss the unique opportunities for maser astrometry with the inclusion of the Square Kilometre Array (SKA) in Very Long Baseline Interferometry (VLBI) networks. The first phase of the SKA will enable observations of hydroxyl and methanol masers, positioning the latter to an accuracy of 5 microarcseconds, and the second phase may allow water maser observations. These observations will provide trigonometric distances with errors as small as 1%. The unrivalled sensitivity of the SKA will enable large-scale surveys and, through joint operations, will turn any VLBI network into a fast astrometry device. Both evolved stars and high mass star formation regions will be accessible throughout the (Southern) Milky Way, completing our understanding of the content, dynamics and history of our Galaxy. Maser velocities and proper motions will be measurable in the Local Group of galaxies and beyond, providing new insights into their kinematics and evolution.
This white paper discusses how a "next-generation" Very Large Array (ngVLA) operating in the frequency range 1-116 GHz could be a groundbreaking tool to study the detailed astrophysics of the "matter cycle" in the Milky Way and other galaxies. If optimized for high brightness sensitivity, the ngVLA would bring detailed microwave spectroscopy and modeling of the full radio spectral energy distribution into regular use as survey tools at resolutions of 0.1- 1 arcseconds. This wavelength range includes powerful diagnostics of density, excitation, and chemistry in the cold ISM, as well as multiple tracers of the rate of recent star formation, the magnetic field, shocks, and properties of the ionized ISM. We highlight design considerations that would make this facility revolutionary in this area, the foremost of which is a large amount of collecting area on ~km-length baselines. We also emphasize the strong case for harnessing the large proposed collecting area of the ngVLA for very long baseline applications as part of the core design. This would allow measurements of the three dimensional space motions of galaxies to beyond the Local Group and mapping of the Milky Way out to the far side of the disk. Finally, we discuss the gains from the proposed combination of very high resolution and sensitivity to thermal emission, which include observing the feeding of black holes and resolving forming protoclusters.
We propose a novel hybrid loss for multiclass and structured prediction problems that is a convex combination of a log loss for Conditional Random Fields (CRFs) and a multiclass hinge loss for Support Vector Machines (SVMs). We provide a sufficient condition for when the hybrid loss is Fisher consistent for classification. This condition depends on a measure of dominance between labels--specifically, the gap between the probabilities of the best label and the second best label. We also prove Fisher consistency is necessary for parametric consistency when learning models such as CRFs. We demonstrate empirically that the hybrid loss typically performs least as well as--and often better than--both of its constituent losses on a variety of tasks, such as human action recognition. In doing so we also provide an empirical comparison of the efficacy of probabilistic and margin based approaches to multiclass and structured prediction.
Adding VLBI capability to the SKA arrays will greatly broaden the science of the SKA, and is feasible within the current specifications. SKA-VLBI can be initially implemented by providing phased-array outputs for SKA1-MID and SKA1-SUR and using these extremely sensitive stations with other radio telescopes, and in SKA2 by realising a distributed configuration providing baselines up to thousands of km, merging it with existing VLBI networks. The motivation for and the possible realization of SKA-VLBI is described in this paper.
We report a new geometric maser distance estimate to the active galaxy NGC 4258. The data for the new model are maser line-of-sight velocities and sky positions from 18 epochs of Very Long Baseline Interferometry observations, and line-of-sight accelerations measured from a 10-year monitoring program of the 22 GHz maser emission of NGC 4258. The new model includes both disk warping and confocal elliptical maser orbits with differential precession. The distance to NGC 4258 is 7.60 +/- 0.17 +/- 0.15 Mpc, a 3% uncertainty including formal fitting and systematic terms. The resulting Hubble Constant, based on the use of the Cepheid Variables in NGC 4258 to recalibrate the Cepheid distance scale (Riess et al. 2011), is H_0 = 72.0 +/- 3.0 km/s/Mpc.
We consider composite loss functions for multiclass prediction comprising a proper (i.e., Fisher-consistent) loss over probability distributions and an inverse link function. We establish conditions for their (strong) convexity and explore the implications. We also show how the separation of concerns afforded by using this composite representation allows for the design of families of losses with the same Bayes risk.
We show that the variational representations for f-divergences currently used in the literature can be tightened. This has implications to a number of methods recently proposed based on this representation. As an example application we use our tighter representation to derive a general f-divergence estimator based on two i.i.d. samples and derive the dual program for this estimator that performs well empirically. We also point out a connection between our estimator and MMD.
We present sensitive Very Large Array observations with an angular resolution of a few arcseconds of the $J= 1 - 0$ line of SiO in the $v$=1 and 2 vibrationally excited states toward a sample of 60 Galactic regions in which stars of high or intermediate mass are currently forming and/or have recently formed. We report the detection of SiO maser emission in \textit{both} vibrationally excited transitions toward only three very luminous regions: Orion-KL, W51N and Sgr B2(M). Toward all three, SiO maser emission had previously been reported, in Orion-KL in both lines, in W51N only in the $v=2$ line and in Sgr B2(M) only in the $v=1$ line. Our work confirms that SiO maser emission in star-forming regions is a rare phenomenon, indeed, that requires special, probably extreme, physical and chemical conditions not commonly found. In addition to this SiO maser survey, we also present images of the simultaneously observed 7 mm continuum emission from a subset of our sample of star-forming regions where such emission was detected. This is in most cases likely to be free-free emission from compact- and ultracompact-HII regions.
A spectral-line survey of IRC+10216 in the 345 GHz band has been undertaken with the Submillimeter Array. Although not yet completed, it has already yielded a fairly large sample of narrow molecular emission lines with line-widths indicating expansion velocities of ~4 km/s, less than 3 times the well-known value of the terminal expansion velocity (14.5 km/s) of the outer envelope. Five of these narrow lines have now been identified as rotational transitions in vibrationally excited states of previously detected molecules: the v=1, J=17--16 and J=19--18 lines of Si34S and 29SiS and the v=2, J=7--6 line of CS. Maps of these lines show that the emission is confined to a region within ~60 AU of the star, indicating that the narrow-line emission is probing the region of dust-formation where the stellar wind is still being accelerated.