Nonlinear variable selection algorithms for surrogate modeling

Having the ability to analyze, simulate, and optimize complex systems is becoming more important in all engineering disciplines. Decision-making using complex systems usually leads to nonlinear optimization problems, which rely on computationally expensive simulations. Therefore, it is often challenging to detect the actual structure of the optimization problem and formulate these problems with closed-form analytical expressions. Surrogate-based optimization of complex systems is a promising approach that is based on the concept of adaptively fitting and optimizing approximations of the input–output data. Standard surrogate-based optimization assumes the degrees of freedom are known a priori; however, in real applications the sparsity and the actual structure of the black-box formulation may not be known. In this work, we propose to select the correct variables contributing to each objective function and constraints of the black-box problem, by formulating the identification of the true sparsity of the formulation as a nonlinear feature selection problem. We compare three variable selection criteria based on Support Vector Regression and develop efficient algorithms to detect the sparsity of black-box formulations when only a limited amount of deterministic or noisy data is available.     

Prevention of Bone Destruction by Mechanical Loading Is Not Enhanced by the Bruton’s Tyrosine Kinase Inhibitor CC-292 in Myeloma Bone Disease

Limiting bone resorption and regenerating bone tissue are treatment goals in myeloma bone disease (MMBD). Physical stimuli such as mechanical loading prevent bone destruction and enhance bone mass in the MOPC315.BM.Luc model of MMBD. It is unknown whether treatment with the Bruton’s tyrosine kinase inhibitor CC-292 (spebrutinib), which regulates osteoclast differentiation and function, augments the anabolic effect of mechanical loading. CC-292 was administered alone and in combination with axial compressive tibial loading in the MOPC315.BM.Luc model for three weeks. However, neither CC-292 alone nor its use in combination with mechanical loading was more effective in reducing osteolytic bone disease or rescuing bone mass than mechanical stimuli alone, as evidenced by microcomputed tomography (microCT) and histomorphometric analysis. Further studies are needed to investigate novel anti-myeloma and anti-resorptive strategies in combination with physical stimuli to improve treatment of MMBD.     

Study of the Adsorption of Ozone on the Surface of Ferric Oxide by Reversed-Flow Inverse Gas Chromatography

Abstract

In this work, in order to describe the impact of ozone on ferric oxide – a common constituent of building materials – in purely scientific terms, a novel methodology that of Reversed Flow–Inverse Gas Chromatography (RF–IGC or RF–GC) is proposed. Five important physicochemical quantities concerning adsorption of ozone on the solid surface of ferric oxide are determined, in five temperatures, in a time-resolved way. By means of a simple PC-program, the values of local adsorption energy, local adsorption isotherm, local monolayer capacity, probability density function for adsorption energy, and energy from lateral interactions are calculated. Thus, the role of surface heterogeneity, lateral interactions and different distributions of active sites (surface topography) on the adsorption of ozone on the surface of ferric oxide is analyzed in the way the RF–IGC method has it developed.     

Strength-based learning classifier systems revisited: Effective rule evolution in supervised classification tasks

Learning classifier systems (LCS) are machine learning systems designed to work for both multi-step and single-step decision tasks. The latter case presents an interesting challenge for such algorithms, especially when they are applied to real-world data mining (DM) problems. The present investigation departs from the popular approach of applying accuracy-based LCS to single-step classification and aims to uncover the potential of strength-based LCS in such tasks. Although the latter family of algorithms have often been associated with poor generalization and performance, we aim at alleviating these problems by defining appropriate extensions to the traditional strength-based LCS framework. These extensions are detailed and their effect on system performance is studied through the application of the proposed algorithm on a set of artificial problems, designed to challenge its scalability and generalization abilities. The comparison of the proposed algorithm with UCS, its state-of-the-art accuracy-based counterpart, emphasizes the effects of our extended strength-based approach and validates its competitiveness in multi-class problems with various class distributions. Overall, our work presents an investigation of strength-based LCS in the domain of supervised classification. Our extensive analysis of the learning dynamics involved in these systems provides proof of their potential as real-world DM tools, inducing tractable rule-based classification models, even in the presence of severe class imbalances.     

Volatile Bio-ester Production from Orange Pulp-Containing Medium Using Saccharomyces cerevisiae

The dynamics of orange pulp (OP) as a fermentation feedstock for the production of volatile esters of “fruity” aroma by using a commercial wine yeast strain (Saccharomyces cerevisiae var. cerevisiae) was investigated. To achieve this goal, the kinetic behaviour of yeast (cell growth, substrate assimilation and volatile ester formation) was studied in OP complemented with nutritive medium under two different aeration conditions. The results were compared with those obtained from conventional processes in nutritive medium containing glucose as a sole carbon source. The results obtained revealed that the yeast cells were able to grow in the OP-containing medium after a 12-h lag period, probably due to the inhibitory effect of limonene. OP was found to stimulate the de novo synthesis of isoamyl acetate, phenylethyl acetate and ethyl esters (hexanoate, octanoate, decanoate and dodecanoate) by S. cerevisiae. This was strongly evident in the case of limited oxygen supply. Based on the above findings, OP appears to be a promising choice for bioflavour production by yeast.     

Verbal pragmatics following unilateral stroke: Emotional content and valence.

Verbal pragmatic aspects of discourse production were examined in 16 right brain-damaged (RBD), 16 left brain-damaged (LBD), and 16 normal control right-handed adults. The facilitation effect of emotional content, valence hypothesis, and relationship between pragmatics and emotion were evaluated. Participants produced monologues while recollecting emotional and nonemotional experiences. Transcribed monologues were rated for appropriateness on 6 pragmatic features: conciseness, lexical selection, quantity, relevancy, specificity, and topic maintenance. Overall, brain-damaged groups were rated as significantly less appropriate than normals. Consistent with the facilitation effect, emotional content enhanced pragmatic performance of LBD aphasic participants yet suppressed performance of RBD participants. Contrary to the valence hypothesis, RBD participants were more impaired for positive emotions and LBD participants for negative emotions. Pragmatic appropriateness was not strongly correlated with a measure of emotional intensity. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Challenges in the Processing Line of Spanish Style cv. Chalkidiki Green Table Olives Spontaneously Fermented in Reduced NaCl Content Brines

In this work, the effect of different treatments using selected combinations of chloride salts (NaCl, KCl, CaCl_2, MgCl₂) in fermentation brine to assess the safety, quality, and nutritional value of the Spanish style cv. Chalkidiki green table olives spontaneously fermented is studied. Examination of safety parameters (bacteria/yeasts, pH, acidity) and changes in olive color and texture characteristics in brines during fermentation, and in mineral content of olives upon storage up to 12 months, is conducted using appropriate methods. Sensory evaluation takes place in certain cases. Among the tested salt combinations, those of 4.08% NaCl, 2.8% KCl, 1% CaCl₂, and 0.12% MgCl₂ (w/v) are found to be the most effective regarding safety, color, texture, and taste. Findings support that Spanish style cv. Chalkidiki green olives can be spontaneously fermented in reduced NaCl content brine. The 50% lower NaCl content of the fermented olives compared with that in traditionally fermented ones suggests that reformulation of cv. Chalkidiki green table olives by safe intervention is feasible. Practical Applications: cv. Chalkidiki green table olives are of high importance for the Greek economy. Traditionally they are spontaneously fermented in NaCl brine (8% w/v). Partial replacement of NaCl content with other salts in such a line results in a product acceptable in terms of safety and quality characteristics. The Greek table olive industry may profit from the findings in an attempt to remodel the traditional spontaneous fermentation in green olive processing according to international dietary guidelines.     

Sparse episode identification in environmental datasets: The case of air quality assessment

Sparse episode identification in environmental datasets is not only a multi-faceted and computationally challenging problem for machine learning algorithms, but also a difficult task for human-decision makers: the strict regulatory framework, in combination with the public demand for better information services, poses the need for robust, efficient and, more importantly, understandable forecasting models. Additionally, these models need to provide decision-makers with “summarized” and valuable knowledge, that has to be subjected to a thorough evaluation procedure, easily translated to services and/or actions in actual decision making situations, and integratable with existing Environmental Management Systems (EMSs).On this basis, our current study investigates the potential of various machine learning algorithms as tools for air quality (AQ) episode forecasting and assesses them – given the corresponding domain-specific requirements – using an evaluation procedure, tailored to the task at hand. Among the algorithms employed in the experimental phase, our main focus is on ZCS-DM, an evolutionary rule-induction algorithm specifically designed to tackle this class of problems – that is classification problems with skewed class distributions, where cost-sensitive model building is required.Overall, we consider this investigation successful, in terms of its aforementioned goals and constraints: obtained experimental results reveal the potential of rule-based algorithms for urban AQ forecasting, and point towards ZCS-DM as the most suitable algorithm for the target domain, providing the best trade-off between model performance and understandability.     

Scheduling multi-product tree-structure pipelines

The problem discussed in this paper is short-term scheduling of distribution of petroleum derivatives from a single oil refinery to a number of depots through a tree-structure pipeline. Scheduling product batches in pipelines is a very complex task with many constraints to be considered. Batches of refined products and grades are pumped back-to-back in the pipeline, often with no separation device between batches. In this work a continuous-time, MILP problem representation for tree-structure pipelines is proposed. The approach is successfully applied to a number of pipeline scheduling problems, including a real-world problem. The data and experimental results are reported.     

Recognizing People From Their Movement.

Human observers demonstrate impressive visual sensitivity to human movement. What defines this sensitivity? If motor experience influences the visual analysis of action, then observers should be most sensitive to their own movements. If view-dependent visual experience determines visual sensitivity to human movement, then observers should be most sensitive to the movements of their friends. To test these predictions, participants viewed sagittal displays of point-light depictions of themselves, their friends, and strangers performing various actions. In actor identification and discrimination tasks, sensitivity to one's own motion was highest. Visual sensitivity to friends', but not strangers', actions was above chance. Performance was action dependent. Control studies yielded chance performance with inverted and static displays, suggesting that form and low-motion cues did not define performance. These results suggest that both motor and visual experience define visual sensitivity to human action. (PsycINFO Database Record (c) 2010 APA, all rights reserved)