Incorporating uncertainty in predictive species distribution modelling

Motivated by the need to solve ecological problems (climate change, habitat fragmentation and biological invasions), there has been increasing interest in species distribution models (SDMs). Predictions from these models inform conservation policy, invasive species management and disease-control measures. However, predictions are subject to uncertainty, the degree and source of which is often unrecognized. Here, we review the SDM literature in the context of uncertainty, focusing on three main classes of SDM: niche-based models, demographic models and process-based models. We identify sources of uncertainty for each class and discuss how uncertainty can be minimized or included in the modelling process to give realistic measures of confidence around predictions. Because this has typically not been performed, we conclude that uncertainty in SDMs has often been underestimated and a false precision assigned to predictions of geographical distribution. We identify areas where development of new statistical tools will improve predictions from distribution models, notably the development of hierarchical models that link different types of distribution model and their attendant uncertainties across spatial scales. Finally, we discuss the need to develop more defensible methods for assessing predictive performance, quantifying model goodness-of-fit and for assessing the significance of model covariates.

Efficient Marginal Likelihood Computation for Gaussian Process Regression

In a Bayesian learning setting, the posterior distribution of a predictive model arises from a trade-off between its prior distribution and the conditional likelihood of observed data. Such distribution functions usually rely on additional hyperparameters which need to be tuned in order to achieve optimum predictive performance; this operation can be efficiently performed in an Empirical Bayes fashion by maximizing the posterior marginal likelihood of the observed data. Since the score function of this optimization problem is in general characterized by the presence of local optima, it is necessary to resort to global optimization strategies, which require a large number of function evaluations. Given that the evaluation is usually computationally intensive and badly scaled with respect to the dataset size, the maximum number of observations that can be treated simultaneously is quite limited. In this paper, we consider the case of hyperparameter tuning in Gaussian process regression. A straightforward implementation of the posterior log-likelihood for this model requires O(N^3) operations for every iteration of the optimization procedure, where N is the number of examples in the input dataset. We derive a novel set of identities that allow, after an initial overhead of O(N^3), the evaluation of the score function, as well as the Jacobian and Hessian matrices, in O(N) operations. We prove how the proposed identities, that follow from the eigendecomposition of the kernel matrix, yield a reduction of several orders of magnitude in the computation time for the hyperparameter optimization problem. Notably, the proposed solution provides computational advantages even with respect to state of the art approximations that rely on sparse kernel matrices.

Time-Varying Liquidity and Momentum Profits

A basic intuition is that arbitrage is easier when markets are most liquid. Surprisingly, we find that momentum profits are markedly larger in liquid market states. This finding is not explained by variation in liquidity risk, time-varying exposure to risk factors, or changes in macroeconomic condition, cross-sectional return dispersion, and investor sentiment. The predictive performance of aggregate market illiquidity for momentum profits uniformly exceed that of market return and market volatility states. While momentum strategies are unconditionally unprofitable in US, Japan, and Eurozone countries in the last decade, they are substantial following liquid market states.

The performance of concrete exposed to marine environments: predictive modelling and use of laboratory/on site test methods

This paper reports an approach by which laboratory based testing and numerical modelling can be combined to predict the long term performance of a range of concretes exposed to marine environments. Firstly, a critical review of the test methods for assessing the chloride penetration resistance of concrete is given. The repeatability of the different test results is also included. In addition to the test methods, a numerical simulation model is used to explore the test data further to obtain long-term chloride ingress trends. The combined use of testing and modelling is validated with the help of long-term chloride ingress data from a North Sea exposure site. In summary, the paper outlines a methodology for determining the long term performance of concrete in marine environments.

On Infinite Time Performance of Nonlinear Model Predictive Controllers

This paper addresses the problem of infinite time performance of model predictive controllers applied to constrained nonlinear systems. The total performance is compared with a finite horizon optimal cost to reveal performance limits of closed-loop model predictive control systems. Based on the Principle of Optimality, an upper and a lower bound of the ratio between the total performance and the finite horizon optimal cost are obtained explicitly expressed by the optimization horizon. The results also illustrate, from viewpoint of performance, how model predictive controllers approaches to infinite optimal controllers as the optimization horizon increases.

Predictive Current Control of Doubly Fed Induction Generators

This paper presents a predictive current control strategy for doubly-fed induction generators (DFIG). The method predicts the DFIG’s rotor current variations in the synchronous reference frame fixed to the stator flux within a fixed sampling period. This is then used to directly calculate the required rotor voltage to eliminate the current errors at the end of the following sampling period. Space vector modulation is used to generate the required switching pulses within the fixed sampling period. The impact of sampling delay on the accuracy of the sampled rotor current is analyzed and detailed compensation methods are proposed to improve the current control accuracy and system stability. Experimental results for a 1.5 kW DFIG system illustrate the effectiveness and robustness of the proposed control strategy during rotor current steps and rotating speed variation. Tests during negative sequence current injection further demonstrate the excellent dynamic performance of the proposed PCC method.

Drivers of sales performance: a contemporary meta-analysis. Have salespeople become knowledge brokers?

It has been 25 years since the publication of a comprehensive review of the full spectrum of salesperformance drivers. This study takes stock of the contemporary field and synthesizes empirical evidence from the period 1982–2008. The authors revise the classification scheme for sales performance determinants devised by Walker et al. (1977) and estimate both the predictive validity of its sub-categories and the impact of a range of moderators on determinant-sales performance relationships. Based on multivariate causal model analysis, the results make two major observations: (1) Five sub-categories demonstrate significant relationships with sales performance: selling-related knowledge (ß=.28), degree of adaptiveness (ß=.27), role ambiguity (ß=-.25), cognitive aptitude (ß=.23) and work engagement (ß=.23). (2) These sub-categories are moderated by measurement method, research context, and salestype variables. The authors identify managerial implications of the results and offer suggestions for further research, including the conjecture that as the world is moving toward a knowledge-intensive economy, salespeople could be functioning as knowledge-brokers. The results seem to back this supposition and indicate how it might inspire future research in the field of personal selling.

Predictive position and force control for shape memory alloy cylinders

In this paper, a linear lightweight electric cylinder constructed using shape memory alloy (SMA) is proposed. Spring SMA is used as the actuator to control the position and force of the cylinder rod. The model predictive control algorithm is investigated to compensate SMA hysteresis phenomenon and control the cylinder. In the predictive algorithm, the future output of the cylinder is computed based on the cylinder model, and the control signal is computed to minimize the error and power criterion. The cylinder model parameters are estimated by an online identification algorithm. Experimental results show that the SMA cylinder is able to precisely control position and force by using the predictive control strategy though the hysteresis effect existing in the actuator. The performance of the proposed controller is compared with that of a conventional PID controller

White matter fractional anisotrophy predicts balance performance in older adults

Aging is characterized by brain structural changes that may compromise motor functions. In the context of postural control, white matter integrity is crucial for the efficient transfer of visual, proprioceptive and vestibular feedback in the brain. To determine the role of age-related white matter decline as a function of the sensory feedback necessary to correct posture, we acquired diffusion weighted images in young and old subjects. A force platform was used to measure changes in body posture under conditions of compromised proprioceptive and/or visual feedback. In the young group, no significant brain structure-balance relations were found. In the elderly however, the integrity of a cluster in the frontal forceps explained 21% of the variance in postural control when proprioceptive information was compromised. Additionally, when only the vestibular system supplied reliable information, the occipital forceps was the best predictor of balance performance (42%). Age-related white matter decline may thus be predictive of balance performance in the elderly when sensory systems start to degrade.

Efficiently exploring architectural design spaces via predictive modeling

Architects use cycle-by-cycle simulation to evaluate design choices and understand tradeoffs and interactions among design parameters. Efficiently exploring exponential-size design spaces with many interacting parameters remains an open problem: the sheer number of experiments renders detailed simulation intractable. We attack this problem via an automated approach that builds accurate, confident predictive design-space models. We simulate sampled points, using the results to teach our models the function describing relationships among design parameters. The models produce highly accurate performance estimates for other points in the space, can be queried to predict performance impacts of architectural changes, and are very fast compared to simulation, enabling efficient discovery of tradeoffs among parameters in different regions. We validate our approach via sensitivity studies on memory hierarchy and CPU design spaces: our models generally predict IPC with only 1-2% error and reduce required simulation by two orders of magnitude. We also show the efficacy of our technique for exploring chip multiprocessor (CMP) design spaces: when trained on a 1% sample drawn from a CMP design space with 250K points and up to 55x performance swings among different system configurations, our models predict performance with only 4-5% error on average. Our approach combines with techniques to reduce time per simulation, achieving net time savings of three-four orders of magnitude. Copyright © 2006 ACM.

Efficient Architectural Design Space Exploration via Predictive Modeling

Efficiently exploring exponential-size architectural design spaces with many interacting parameters remains an open problem: the sheer number of experiments required renders detailed simulation intractable.We attack this via an automated approach that builds accurate predictive models. We simulate sampled points, using results to teach our models the function describing relationships among design parameters. The models can be queried and are very fast, enabling efficient design tradeoff discovery. We validate our approach via two uniprocessor sensitivity studies, predicting IPC with only 1–2% error. In an experimental study using the approach, training on 1% of a 250-K-point CMP design space allows our models to predict performance with only 4–5% error. Our predictive modeling combines well with techniques that reduce the time taken by each simulation experiment, achieving net time savings of three-four orders of magnitude.

Machine Learning for Predictive Maintenance: A Multiple Classifiers Approach

In this paper a multiple classifier machine learning methodology for Predictive Maintenance (PdM) is presented. PdM is a prominent strategy for dealing with maintenance issues given the increasing need to minimize downtime and associated costs. One of the challenges with PdM is generating so called ’health factors’ or quantitative indicators of the status of a system associated with a given maintenance issue, and determining their relationship to operating costs and failure risk. The proposed PdM methodology allows dynamical decision rules to be adopted for maintenance management and can be used with high-dimensional and censored data problems. This is achieved by training multiple classification modules with different prediction horizons to provide different performance trade-offs in terms of frequency of unexpected breaks and unexploited lifetime and then employing this information in an operating cost based maintenance decision system to minimise expected costs. The effectiveness of the methodology is demonstrated using a simulated example and a benchmark semiconductor manufacturing maintenance problem.

Predictors of numeracy performance in undergraduate psychology, nursing and medical students

Psychology, nursing and medicine are undergraduate degrees that require students to attain a level of numerical competence for graduation. Yet, the numeracy aspect of these courses is often actively disliked and poorly performed. This study's aim was to identify what factors most strongly predict performance in such courses. Three hundred and twenty-five undergraduate students from these three disciplines were given measures of numeracy performance, maths anxiety, maths attitudes and various demographic and educational variables. From these data three separate path analysis models were formed, showing the predictive effects of affective, demographic and educational variables on numeracy performance. Maths anxiety was the strongest affective predictor for psychology and nursing students, with motivation being more important for medical students. Across participant groups, pre-university maths qualifications were the strongest demographic/educational predictor of performance. The results can be used to suggest ways to improve performance in students having difficulty with numeracy-based modules.