Two problems with variational expectation maximisation for time-series models

Variational methods are a key component of the approximate inference and learning toolbox. These methods fill an important middle ground, retaining distributional information about uncertainty in latent variables, unlike maximum a posteriori methods (MAP), and yet generally requiring less computational time than Monte Carlo Markov Chain methods. In particular the variational Expectation Maximisation (vEM) and variational Bayes algorithms, both involving variational optimisation of a free-energy, are widely used in time-series modelling. Here, we investigate the success of vEM in simple probabilistic time-series models. First we consider the inference step of vEM, and show that a consequence of the well-known compactness property of variational inference is a failure to propagate uncertainty in time, thus limiting the usefulness of the retained distributional information. In particular, the uncertainty may appear to be smallest precisely when the approximation is poorest. Second, we consider parameter learning and analytically reveal systematic biases in the parameters found by vEM. Surprisingly, simpler variational approximations (such a mean-field) can lead to less bias than more complicated structured approximations.

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.

Chronic effects of water-borne PFOS exposure on growth, survival and hepatotoxicity in zebrafish: A partial life-cycle test

Perfluorooctane sulfonate (PFOS) is widely distributed and persistent in the environment and wildlife. The main aim of this study was to investigate the impact of long-term exposure to low concentrations of PFOS in zebrafish. Zebrafish fry (F-0, 14d post-fertilization, dpf) were exposed via the water for 70d to 0 (control), 10, 50 and 250 mu g L-1 PFOS, followed by a further 30d to assess recovery in clean water. The effects on survival and growth parameters and liver histopathology were assessed. Although growth suppression (weight and length) was observed in fish treated with high concentrations PFOS during the exposure period, no mortality was observed throughout the 70d experiment. Embryos and larvae (F-1) derived from maternal exposure suffered malformation and mortality. Exposure to 50 and 250 mu g L-1 PFOS could inhibit the growth of the gonads (GSI) in the female zebrafish. Histopathological alterations, primary with lipid droplets accumulation, were most prominently seen in the liver of males and the changes were not reversible, even after the fish were allowed to recover for 30d in clean water. The triiodothyronine (T-3)) levels were not significantly changed in any of the exposure groups. Hepatic vitellogenin (VTG) gene expression was significantly up-regulated in both male and female zebrafish, but the sex ratio was not altered. The overall results suggested that lower concentrations of PFOS in maternal exposure could result in offspring deformation and mortality. (c) 2008 Elsevier Ltd. All rights reserved.

Material point method for coupled hydromechanical problems

This paper describes a new formulation of the material point method (MPM) for solving coupled hydromechanical problems of fluid-saturated soil subjected to large deformation. A soil-pore fluid coupled MPM algorithm based on Biot's mixture theory is proposed for solving hydromechanical interaction problems that include changes in water table location with time. The accuracy of the proposed method is examined by comparing the results of the simulation of a one-dimensional consolidation test with the corresponding analytical solution. A sensitivity analysis of the MPM parameters used in the proposed method is carried out for examining the effect of the number of particles per mesh and mesh size on solution accuracy. For demonstrating the capability of the proposed method, a physical model experiment of a large-scale levee failure by seepage is simulated. The behavior of the levee model with time-dependent changes in water table matches well to the experimental observations. The mechanisms of seepage-induced failure are discussed by examining the pore-water pressures, as well as the effective stresses computed from the simulations © 2013 American Society of Civil Engineers.

A preliminary study on methods for retaining data quality problems in automatically generated test data

Data in an organisation often contains business secrets that organisations do not want to release. However, there are occasions when it is necessary for an organisation to release its data such as when outsourcing work or using the cloud for Data Quality (DQ) related tasks like data cleansing. Currently, there is no mechanism that allows organisations to release their data for DQ tasks while ensuring that it is suitably protected from releasing business related secrets. The aim of this paper is therefore to present our current progress on determining which methods are able to modify secret data and retain DQ problems. So far we have identified the ways in which data swapping and the SHA-2 hash function alterations methods can be used to preserve missing data, incorrectly formatted values, and domain violations DQ problems while minimising the risk of disclosing secrets. © (2012) by the AIS/ICIS Administrative Office All rights reserved.

Classifying data quality problems in asset management

© Springer International Publishing Switzerland 2015. Making sound asset management decisions, such as whether to replace or maintain an ageing underground water pipe, are critical to ensure that organisations maximise the performance of their assets. These decisions are only as good as the data that supports them, and hence many asset management organisations are in desperate need to improve the quality of their data. This chapter reviews the key academic research on data quality (DQ) and Information Quality (IQ) (used interchangeably in this chapter) in asset management, combines this with the current DQ problems faced by asset management organisations in various business sectors, and presents a classification of the most important DQ problems that need to be tackled by asset management organisations. In this research, eleven semi structured interviews were carried out with asset management professionals in a range of business sectors in the UK. The problems described in the academic literature were cross checked against the problems found in industry. In order to support asset management professionals in solving these problems, we categorised them into seven different DQ dimensions, used in the academic literature, so that it is clear how these problems fit within the standard frameworks for assessing and improving data quality. Asset management professionals can therefore now use these frameworks to underpin their DQ improvement initiatives while focussing on the most critical DQ problems.

Dynamics of microcystins-LR and -RR in the phytoplanktivorous silver carp in a sub-chronic toxicity experiment

A sub-chronic toxicity experiment was conducted to examine tissue distribution and depuration of two microcystins (microcystin-LR and microcystin -RR) in the phytoplanktivorous filter-feeding silver carp during a course of 80 days. Two large tanks (A, B) were used, and in Tank A, the fish were fed naturally with fresh Microcystis viridis cells (collected from a eutrophic pond) throughout the experiment, while in Tank B, the food of the fish were M. viridis cells for the first 40 days and then changed to artificial carp feed. High Performance Liquid Chromatography (HPLC) was used to measure MC-LR and MC-RR in the M. viridis cells, the seston, and the intestine, blood, liver and muscle tissue of silver carp at an interval of 20 days. MC-RR and MC-LR in the collected Microcystis cells varied between 268-580 and 110-292 mug g(-1) DW, respectively. In Tank A, MC-RR and MC-LR varied between 41.5-99.5 and 6.9-15.8 mug g(-1) DW in the seston, respectively. The maximum MC-RR in the blood, liver and muscle of the fish was 49.7, 17.8 and 1.77 mug g(-1) DW, respectively. No MC-LR was detectable in the muscle and blood samples of the silver carp in spite of the abundant presence of this toxin in the intestines (for the liver, there was only one case when a relatively minor quantity was detected). These findings contrast with previous experimental results on rainbow trout. Perhaps silver carp has a mechanism to degrade MC-LR actively and to inhibit MC-LR transportation across the intestines. The depuration of MC-RR concentrations occurred slowly than uptakes in blood, liver and muscle, and the depuration rate was in the order of blood > liver > muscle. The grazing ability of silver carp on toxic cyanobacteria suggests an applicability of using phytoplanktivorous fish to counteract cyanotoxin contamination in eutrophic waters. (C) 2003 Elsevier Ltd. All rights reserved.