Gaussian process change point models

We combine Bayesian online change point detection with Gaussian processes to create a nonparametric time series model which can handle change points. The model can be used to locate change points in an online manner; and, unlike other Bayesian online change point detection algorithms, is applicable when temporal correlations in a regime are expected. We show three variations on how to apply Gaussian processes in the change point context, each with their own advantages. We present methods to reduce the computational burden of these models and demonstrate it on several real world data sets. Copyright 2010 by the author(s)/owner(s).

A method for identifying scope and assumptions in developing and selecting methods and tools

This technical report describes a practical method consisting of a checklist and a supporting techniques for those planning or just starting to develop or select design tools and methods. The method helps to summarize and illustrate the envisaged tool or method by identifying its scope and the underlying assumptions. The resulting tool or method description clarifies the problem that is addressed, the approach and the possible implications, and can thus be used by a variety of people involved in assessing a tool or method in an early stage. For the developers themselves the method reveals how realistic the envisaged method or tool is, and whether the scope has to be narrowed.

Unsupervised intra-lingual and cross-lingual speaker adaptation for HMM-based speech synthesis using two-pass decision tree construction

Hidden Markov model (HMM)-based speech synthesis systems possess several advantages over concatenative synthesis systems. One such advantage is the relative ease with which HMM-based systems are adapted to speakers not present in the training dataset. Speaker adaptation methods used in the field of HMM-based automatic speech recognition (ASR) are adopted for this task. In the case of unsupervised speaker adaptation, previous work has used a supplementary set of acoustic models to estimate the transcription of the adaptation data. This paper first presents an approach to the unsupervised speaker adaptation task for HMM-based speech synthesis models which avoids the need for such supplementary acoustic models. This is achieved by defining a mapping between HMM-based synthesis models and ASR-style models, via a two-pass decision tree construction process. Second, it is shown that this mapping also enables unsupervised adaptation of HMM-based speech synthesis models without the need to perform linguistic analysis of the estimated transcription of the adaptation data. Third, this paper demonstrates how this technique lends itself to the task of unsupervised cross-lingual adaptation of HMM-based speech synthesis models, and explains the advantages of such an approach. Finally, listener evaluations reveal that the proposed unsupervised adaptation methods deliver performance approaching that of supervised adaptation.

Blade Loading and Its Application in the Mean-Line Design of Low Pressure Turbines

In order to minimize the number of iterations to a turbine design, reasonable choices of the key parameters must be made at the preliminary design stage. The choice of blade loading is of particular concern in the low pressure (LP) turbine of civil aero engines, where the use of high-lift blades is widespread. This paper considers how blade loading should be measured, compares the performance of various loss correlations, and explores the impact of blade lift on performance and lapse rates. To these ends, an analytical design study is presented for a repeating-stage, axial-flow LP turbine. It is demonstrated that the long-established Zweifel lift coefficient (Zweifel, 1945, "The Spacing of Turbomachine Blading, Especially with Large Angular Deflection" Brown Boveri Rev., 32(1), pp. 436-444) is flawed because it does not account for the blade camber. As a result the Zweifel coefficient is only meaningful for a fixed set of flow angles and cannot be used as an absolute measure of blade loading. A lift coefficient based on circulation is instead proposed that accounts for the blade curvature and is independent of the flow angles. Various existing profile and secondary loss correlations are examined for their suitability to preliminary design. A largely qualitative comparison demonstrates that the loss correlations based on Ainley and Mathieson (Ainley and Mathieson, 1957, "A Method of Performance Estimation for Axial-Flow Turbines," ARC Reports and Memoranda No. 2974; Dunham and Came, 1970, "Improvements to the Ainley-Mathieson Method of Turbine Performance Prediction," Trans. ASME: J. Eng. Gas Turbines Power, July, pp. 252-256; Kacker and Okapuu, 1982, "A Mean Line Performance Method for Axial Flow Turbine Efficiency," J. Eng. Power, 104, pp. 111-119). are not realistic, while the profile loss model of Coull and Hodson (Coull and Hodson, 2011, "Predicting the Profile Loss of High-Lift Low Pressure Turbines," J. Turbomach., 134(2), pp. 021002) and the secondary loss model of (Traupel, W, 1977, Thermische Turbomaschinen, Springer-Verlag, Berlin) are arguably the most reasonable. A quantitative comparison with multistage rig data indicates that, together, these methods over-predict lapse rates by around 30%, highlighting the need for improved loss models and a better understanding of the multistage environment. By examining the influence of blade lift across the Smith efficiency chart, the analysis demonstrates that designs with higher flow turning will tend to be less sensitive to increases in blade loading. © 2013 American Society of Mechanical Engineers.

High-order detached eddy simulation, zonal LES and URANS of cavity and labyrinth seal flows

Hybrid numerical large eddy simulation (NLES), detached eddy simulation (DES) and URANS methods are assessed on a cavity and a labyrinth seal geometry. A high sixth-order discretization scheme is used and is validated using the test case of a two-dimensional vortex. The hybrid approach adopts a new blending function. For the URANS simulations, the flow within the cavity remains steady, and the results show significant variation between models. Surprisingly, low levels of resolved turbulence are observed in the cavity for the DES simulation, and the cavity shear layer remains two dimensional. The hybrid RANS-NLES approach does not suffer from this trait.For the labyrinth seal, both the URANS and DES approaches give low levels of resolved turbulence. The zonal Hamilton-Jacobi approach on the other had given significantly more resolved content. Both DES and hybrid RANS-NLES give good agreement with the experimentally measured velocity profiles. Again, there is significant variation between the URANS models, and swirl velocities are overpredicted. © 2013 John Wiley & Sons, Ltd.

Coupling model uncertainty for coupled rainfall/runoff and surface water quality models in river problems

Coupled hydrology and water quality models are an important tool today, used in the understanding and management of surface water and watershed areas. Such problems are generally subject to substantial uncertainty in parameters, process understanding, and data. Component models, drawing on different data, concepts, and structures, are affected differently by each of these uncertain elements. This paper proposes a framework wherein the response of component models to their respective uncertain elements can be quantified and assessed, using a hydrological model and water quality model as two exemplars. The resulting assessments can be used to identify model coupling strategies that permit more appropriate use and calibration of individual models, and a better overall coupled model response. One key finding was that an approximate balance of water quality and hydrological model responses can be obtained using both the QUAL2E and Mike11 water quality models. The balance point, however, does not support a particularly narrow surface response (or stringent calibration criteria) with respect to the water quality calibration data, at least in the case examined here. Additionally, it is clear from the results presented that the structural source of uncertainty is at least as significant as parameter-based uncertainties in areal models. © 2012 John Wiley & Sons, Ltd.

Automatic construction and natural-language description of nonparametric regression models

Copyright © 2014, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. This paper presents the beginnings of an automatic statistician, focusing on regression problems. Our system explores an open-ended space of statistical models to discover a good explanation of a data set, and then produces a detailed report with figures and natural- language text. Our approach treats unknown regression functions non- parametrically using Gaussian processes, which has two important consequences. First, Gaussian processes can model functions in terms of high-level properties (e.g. smoothness, trends, periodicity, changepoints). Taken together with the compositional structure of our language of models this allows us to automatically describe functions in simple terms. Second, the use of flexible nonparametric models and a rich language for composing them in an open-ended manner also results in state- of-the-art extrapolation performance evaluated over 13 real time series data sets from various domains.

Separating viscoelasticity and poroelasticity of gels with different length and time scales

Viscoelasticity and poroelasticity commonly coexist as time-dependent behaviors in polymer gels. Engineering applications often require knowledge of both behaviors separated; however, few methods exist to decouple viscoelastic and poroelastic properties of gels. We propose a method capable of separating viscoelasticity and poroelasticity of gels in various mechanical tests. The viscoelastic characteristic time and the poroelastic diffusivity of a gel define an intrinsic material length scale of the gel. The experimental setup gives a sample length scale, over which the solvent migrates in the gel. By setting the sample length to be much larger or smaller than the material length, the viscoelasticity and poroelasticity of the gel will dominate at different time scales in a test. Therefore, the viscoelastic and poroelastic properties of the gel can be probed separately at different time scales of the test. We further validate the method by finite-element models and stress-relaxation experiments. © 2014 The Chinese Society of Theoretical and Applied Mechanics; Institute of Mechanics, Chinese Academy of Sciences and Springer-Verlag Berlin Heidelberg.

"Plateau"-related summary statistics are uninformative for comparing working memory models.

Performance on visual working memory tasks decreases as more items need to be remembered. Over the past decade, a debate has unfolded between proponents of slot models and slotless models of this phenomenon (Ma, Husain, Bays (Nature Neuroscience 17, 347-356, 2014). Zhang and Luck (Nature 453, (7192), 233-235, 2008) and Anderson, Vogel, and Awh (Attention, Perception, Psychophys 74, (5), 891-910, 2011) noticed that as more items need to be remembered, "memory noise" seems to first increase and then reach a "stable plateau." They argued that three summary statistics characterizing this plateau are consistent with slot models, but not with slotless models. Here, we assess the validity of their methods. We generated synthetic data both from a leading slot model and from a recent slotless model and quantified model evidence using log Bayes factors. We found that the summary statistics provided at most 0.15 % of the expected model evidence in the raw data. In a model recovery analysis, a total of more than a million trials were required to achieve 99 % correct recovery when models were compared on the basis of summary statistics, whereas fewer than 1,000 trials were sufficient when raw data were used. Therefore, at realistic numbers of trials, plateau-related summary statistics are highly unreliable for model comparison. Applying the same analyses to subject data from Anderson et al. (Attention, Perception, Psychophys 74, (5), 891-910, 2011), we found that the evidence in the summary statistics was at most 0.12 % of the evidence in the raw data and far too weak to warrant any conclusions. The evidence in the raw data, in fact, strongly favored the slotless model. These findings call into question claims about working memory that are based on summary statistics.