Neural and wavelet network models for financial distress classification

This work analyzes the use of linear discriminant models, multi-layer perceptron neural networks and wavelet networks for corporate financial distress prediction. Although simple and easy to interpret, linear models require statistical assumptions that may be unrealistic. Neural networks are able to discriminate patterns that are not linearly separable, but the large number of parameters involved in a neural model often causes generalization problems. Wavelet networks are classification models that implement nonlinear discriminant surfaces as the superposition of dilated and translated versions of a single "mother wavelet" function. In this paper, an algorithm is proposed to select dilation and translation parameters that yield a wavelet network classifier with good parsimony characteristics. The models are compared in a case study involving failed and continuing British firms in the period 1997-2000. Problems associated with over-parameterized neural networks are illustrated and the Optimal Brain Damage pruning technique is employed to obtain a parsimonious neural model. The results, supported by a re-sampling study, show that both neural and wavelet networks may be a valid alternative to classical linear discriminant models.

A neural network enhanced generalized minimum variance self-tuning proportional, integral and derivative control algorithm for complex dynamic systems

A neural network enhanced proportional, integral and derivative (PID) controller is presented that combines the attributes of neural network learning with a generalized minimum-variance self-tuning control (STC) strategy. The neuro PID controller is structured with plant model identification and PID parameter tuning. The plants to be controlled are approximated by an equivalent model composed of a simple linear submodel to approximate plant dynamics around operating points, plus an error agent to accommodate the errors induced by linear submodel inaccuracy due to non-linearities and other complexities. A generalized recursive least-squares algorithm is used to identify the linear submodel, and a layered neural network is used to detect the error agent in which the weights are updated on the basis of the error between the plant output and the output from the linear submodel. The procedure for controller design is based on the equivalent model, and therefore the error agent is naturally functioned within the control law. In this way the controller can deal not only with a wide range of linear dynamic plants but also with those complex plants characterized by severe non-linearity, uncertainties and non-minimum phase behaviours. Two simulation studies are provided to demonstrate the effectiveness of the controller design procedure.

Self-tuning proportional, integral and derivative controller based on genetic algorithm least squares

A self-tuning proportional, integral and derivative control scheme based on genetic algorithms (GAs) is proposed and applied to the control of a real industrial plant. This paper explores the improvement in the parameter estimator, which is an essential part of an adaptive controller, through the hybridization of recursive least-squares algorithms by making use of GAs and the possibility of the application of GAs to the control of industrial processes. Both the simulation results and the experiments on a real plant show that the proposed scheme can be applied effectively.

Cannabidivarin is anticonvulsant in mouse and rat in vitro and in seizure models

Summary Background and purpose: Phytocannabinoids in Cannabis sativa have diverse pharmacological targets extending beyond cannabinoid receptors and several exert notable anticonvulsant effects. For the first time, we investigated the anticonvulsant profile of the phytocannabinoid cannabidivarin (CBDV) in vitro and in in vivo seizure models. Experimental approach: The effect of CBDV (1-100μM) on epileptiform local field potentials (LFPs) induced in rat hippocampal brain slices by 4-AP application or Mg2+-free conditions was assessed by in vitro multi-electrode array recordings. Additionally, the anticonvulsant profile of CBDV (50-200 mg kg-1) in vivo was investigated in four rodent seizure models: maximal electroshock (mES) and audiogenic seizures in mice, and pentylenetetrazole (PTZ) and pilocarpine-induced seizures in rat. CBDV effects in combination with commonly-used antiepileptic drugs were investigated in rat seizures. Finally, the motor side effect profile of CBDV was investigated using static beam and gripstrength assays. Key results: CDBV significantly attenuated status epilepticus-like epileptiform LFPs induced by 4-AP and Mg2+-free conditions. CBDV had significant anticonvulsant effects in mES (≥100 mg kg-1), audiogenic (≥50 mg kg-1) and PTZ-induced seizures (≥100 mg kg-1). CBDV alone had no effect against pilocarpine-induced seizures, but significantly attenuated these seizures when administered with valproate or phenobarbital at 200 mg kg-1 CBDV. CBDV had no effect on motor function. Conclusions and Implications: These results indicate that CBDV is an effective anticonvulsant across a broad range of seizure models, does not significantly affect normal motor function and therefore merits further investigation in chronic epilepsy models to justify human trials.

A comparison of Shiga-toxin negative Escherichia coli O157 aflagellate and intimin deficient mutants in porcine in vitro and in vivo models of infection

Isolation of Shiga-toxin (Stx) positive Escherichia coli O157:H7 from commercially grown pigs has been reported. Furthermore, experimental infection studies have demonstrated that Stx-positive E. coli O157:H7 can persist in 12-week-old experimentally orally inoculated conventional pigs for up to 2 months and that persistence was not dependent upon intimin. We have shown that the flagellum of Stx-negative E. coli O157:H7 does not have a role to play in pathogenesis in ruminant models whereas, in poultry, the flagellum of E. coli O157:H7 was important for long-term persistent infection. The contribution of the flagellum of Stx-negative E. coli O157 in the colonisation of pigs was investigated by adherence assays on a porcine (IPI-21) cell line, porcine in vitro organ culture (IVOC) and experimental oral inoculation of conventional 14-week-old pigs. E. coli O157:H7 NCTC12900nal(r) and isogenic aflagellate and intimin deficient mutants adhered equally well to IPI-21 cells. In porcine IVOC association assays, E. coli O157:H7 NCTC12900nal(r) was associated in significantly higher numbers to tissues from the caecum and the terminal rectum than other sites. The aflagellate and intimin deficient mutants significantly adhered in greater numbers to more IVOC gastrointestinal tissues than the parent. Groups of 14-week-old pigs were dosed orally with 10(10) CFU/10 ml of either E. coli O157:H7 NCTC12900nal(r) or isogenic aflagellate and intimin deficient mutants and recovery of each test strain was similar. Histological analysis of pig tissues at post mortem examination revealed that E. coli O157 specifically stained bacteria were associated with the mucosa of the ascending and spiral colon. These data suggest that colonisation and persistence of Stx-negative E. coli O157:H7 in pigs, involves mechanisms that do not require the flagellum or intimin.

Systematic winter sea-surface temperature biases in the northern Arabian Sea in HiGEM and the CMIP3 models

Analysis of 20th century simulations of the High resolution Global Environment Model (HiGEM) and the Third Coupled Model Intercomparison Project (CMIP3) models shows that most have a cold sea-surface temperature (SST) bias in the northern Arabian Sea during boreal winter. The association between Arabian Sea SST and the South Asian monsoon has been widely studied in observations and models, with winter cold biases known to be detrimental to rainfall simulation during the subsequent monsoon in coupled general circulation models (GCMs). However, the causes of these SST biases are not well understood. Indeed this is one of the first papers to address causes of the cold biases. The models show anomalously strong north-easterly winter monsoon winds and cold air temperatures in north-west India, Pakistan and beyond. This leads to the anomalous advection of cold, dry air over the Arabian Sea. The cold land region is also associated with an anomalously strong meridional surface temperature gradient during winter, contributing to the enhanced low-level convergence and excessive precipitation over the western equatorial Indian Ocean seen in many models.

Comparison of polar ozone loss rates simulated by one-dimensional and three-dimensional models with Match observations in recent Antarctic and Arctic winters

Simulations of ozone loss rates using a three-dimensional chemical transport model and a box model during recent Antarctic and Arctic winters are compared with experimental loss rates. The study focused on the Antarctic winter 2003, during which the first Antarctic Match campaign was organized, and on Arctic winters 1999/2000, 2002/2003. The maximum ozone loss rates retrieved by the Match technique for the winters and levels studied reached 6 ppbv/sunlit hour and both types of simulations could generally reproduce the observations at 2-sigma error bar level. In some cases, for example, for the Arctic winter 2002/2003 at 475 K level, an excellent agreement within 1-sigma standard deviation level was obtained. An overestimation was also found with the box model simulation at some isentropic levels for the Antarctic winter and the Arctic winter 1999/2000, indicating an overestimation of chlorine activation in the model. Loss rates in the Antarctic show signs of saturation in September, which have to be considered in the comparison. Sensitivity tests were performed with the box model in order to assess the impact of kinetic parameters of the ClO-Cl2O2 catalytic cycle and total bromine content on the ozone loss rate. These tests resulted in a maximum change in ozone loss rates of 1.2 ppbv/sunlit hour, generally in high solar zenith angle conditions. In some cases, a better agreement was achieved with fastest photolysis of Cl2O2 and additional source of total inorganic bromine but at the expense of overestimation of smaller ozone loss rates derived later in the winter.

Linking urban water balance and energy balance models to analyse urban design options

Using a water balance modelling framework, this paper analyses the effects of urban design on the water balance, with a focus on evapotranspiration and storm water. First, two quite different urban water balance models are compared: Aquacycle which has been calibrated for a suburban catchment in Canberra, Australia, and the single-source urban evapotranspiration-interception scheme (SUES), an energy-based approach with a biophysically advanced representation of interception and evapotranspiration. A fair agreement between the two modelled estimates of evapotranspiration was significantly improved by allowing the vegetation cover (leaf area index, LAI) to vary seasonally, demonstrating the potential of SUES to quantify the links between water sensitive urban design and microclimates and the advantage of comparing the two modelling approaches. The comparison also revealed where improvements to SUES are needed, chiefly through improved estimates of vegetation cover dynamics as input to SUES, and more rigorous parameterization of the surface resistance equations using local-scale suburban flux measurements. Second, Aquacycle is used to identify the impact of an array of water sensitive urban design features on the water balance terms. This analysis confirms the potential to passively control urban microclimate by suburban design features that maximize evapotranspiration, such as vegetated roofs. The subsequent effects on daily maximum air temperatures are estimated using an atmospheric boundary layer budget. Potential energy savings of about 2% in summer cooling are estimated from this analysis. This is a clear ‘return on investment’ of using water to maintain urban greenspace, whether as parks distributed throughout an urban area or individual gardens or vegetated roofs.

Forecasting with difference and trend stationary models

Although difference-stationary (DS) and trend-stationary (TS) processes have been subject to considerable analysis, there are no direct comparisons for each being the data-generation process (DGP). We examine incorrect choice between these models for forecasting for both known and estimated parameters. Three sets of Monte Carlo simulations illustrate the analysis, to evaluate the biases in conventional standard errors when each model is mis-specified, compute the relative mean-square forecast errors of the two models for both DGPs, and investigate autocorrelated errors, so both models can better approximate the converse DGP. The outcomes are surprisingly different from established results.

Lifestyle-induced metabolic inflexibility and accelerated ageing syndrome: insulin resistance, friend or foe?

The metabolic syndrome may have its origins in thriftiness, insulin resistance and one of the most ancient of all signalling systems, redox. Thriftiness results from an evolutionarily-driven propensity to minimise energy expenditure. This has to be balanced with the need to resist the oxidative stress from cellular signalling and pathogen resistance, giving rise to something we call 'redox-thriftiness'. This is based on the notion that mitochondria may be able to both amplify membrane-derived redox growth signals as well as negatively regulate them, resulting in an increased ATP/ROS ratio. We suggest that 'redox-thriftiness' leads to insulin resistance, which has the effect of both protecting the individual cell from excessive growth/inflammatory stress, while ensuring energy is channelled to the brain, the immune system, and for storage. We also suggest that fine tuning of redox-thriftiness is achieved by hormetic (mild stress) signals that stimulate mitochondrial biogenesis and resistance to oxidative stress, which improves metabolic flexibility. However, in a non-hormetic environment with excessive calories, the protective nature of this system may lead to escalating insulin resistance and rising oxidative stress due to metabolic inflexibility and mitochondrial overload. Thus, the mitochondrially-associated resistance to oxidative stress (and metabolic flexibility) may determine insulin resistance. Genetically and environmentally determined mitochondrial function may define a 'tipping point' where protective insulin resistance tips over to inflammatory insulin resistance. Many hormetic factors may induce mild mitochondrial stress and biogenesis, including exercise, fasting, temperature extremes, unsaturated fats, polyphenols, alcohol, and even metformin and statins. Without hormesis, a proposed redox-thriftiness tipping point might lead to a feed forward insulin resistance cycle in the presence of excess calories. We therefore suggest that as oxidative stress determines functional longevity, a rather more descriptive term for the metabolic syndrome is the 'lifestyle-induced metabolic inflexibility and accelerated ageing syndrome'. Ultimately, thriftiness is good for us as long as we have hormetic stimuli; unfortunately, mankind is attempting to remove all hormetic (stressful) stimuli from his environment.

Threshold autoregressive and Markov switching models: an application to commercial real estate

Although financial theory rests heavily upon the assumption that asset returns are normally distributed, value indices of commercial real estate display significant departures from normality. In this paper, we apply and compare the properties of two recently proposed regime switching models for value indices of commercial real estate in the US and the UK, both of which relax the assumption that observations are drawn from a single distribution with constant mean and variance. Statistical tests of the models' specification indicate that the Markov switching model is better able to capture the non-stationary features of the data than the threshold autoregressive model, although both represent superior descriptions of the data than the models that allow for only one state. Our results have several implications for theoretical models and empirical research in finance.