University of Waterloo

About: University of Waterloo is a education organization based out in Waterloo, Ontario, Canada. It is known for research contribution in the topics: Population & Context (language use). The organization has 36093 authors who have published 93906 publications receiving 2948139 citations. The organization is also known as: UW & uwaterloo.

NewsQA: A Machine Comprehension Dataset

Abstract: We present NewsQA, a challenging machine comprehension dataset of over 100,000 human-generated question-answer pairs. Crowdworkers supply questions and answers based on a set of over 10,000 news articles from CNN, with answers consisting of spans of text in the articles. We collect this dataset through a four-stage process designed to solicit exploratory questions that require reasoning. Analysis confirms that NewsQA demands abilities beyond simple word matching and recognizing textual entailment. We measure human performance on the dataset and compare it to several strong neural models. The performance gap between humans and machines (13.3% F1) indicates that significant progress can be made on NewsQA through future research. The dataset is freely available online.

Matroids and the greedy algorithm

Abstract: Linear-algebra rank is the solution to an especially tractable optimization problem This tractability is viewed abstractly, and extended to certain more general optimization problems which are linear programs relative to certain derived polyhedra

Self-selection and Analyst Coverage

Abstract: We examine implications of the conjecture that analysts announce recommendations and forecasts selectively, based upon whether their information about the firm is favorable. We propose this as an alternative to the common assumption that analysts introduce bias into their forecasts, and provide empirical evidence on this alternative. An effect of selective reporting is that ex post observed distributions of earnings forecast errors appear over-optimistic, even if each forecast is unbiased ex ante. This occurs because high forecasts are both more likely to be observed and more likely to be over-optimistic than low forecasts, for any given realization. We find strong evidence supporting the self-selection conjecture in analyst recommendations, and generally consistent evidence in analyst forecast errors. We also document that analysts avoid issuing negative information by sparse use of sell recommendations and by delaying downgrades, but not by avoiding downgrades altogether.

The Herschel-Heterodyne Instrument for the Far-Infrared (HIFI)

Abstract: Aims. This paper describes the Heterodyne Instrument for the Far-Infrared (HIFI) that was launched onboard ESA's Herschel Space Observatory in May 2009. Methods. The instrument is a set of 7 heterodyne receivers that are electronically tuneable, covering 480-1250 GHz with SIS mixers and the 1410-1910 GHz range with hot electron bolometer (HEB) mixers. The local oscillator (LO) subsystem comprises a Ka-band synthesizer followed by 14 chains of frequency multipliers and 2 chains for each frequency band. A pair of auto-correlators and a pair of acousto-optical spectrometers process the two IF signals from the dual-polarization, single-pixel front-ends to provide instantaneous frequency coverage of 2 × 4 GHz, with a set of resolutions (125 kHz to 1 MHz) that are better than 0.1 km s-1. Results. After a successful qualification and a pre-launch TB/TV test program, the flight instrument is now in-orbit and completed successfully the commissioning and performance verification phase. The in-orbit performance of the receivers matches the pre-launch sensitivities. We also report on the in-orbit performance of the receivers and some first results of HIFI's operations.