A new reverse reduce-error ensemble pruning algorithm

Although greedy algorithms possess high efficiency, they often receive suboptimal solutions of the ensemble pruning problem, since their exploration areas are limited in large extent. And another marked defect of almost all the currently existing ensemble pruning algorithms, including greedy ones, consists in: they simply abandon all of the classifiers which fail in the competition of ensemble selection, causing a considerable waste of useful resources and information. Inspired by these observations, an interesting greedy Reverse Reduce-Error (RRE) pruning algorithm incorporated with the operation of subtraction is proposed in this work. The RRE algorithm makes the best of the defeated candidate networks in a way that, the Worst Single Model (WSM) is chosen, and then, its votes are subtracted from the votes made by those selected components within the pruned ensemble. The reason is because, for most cases, the WSM might make mistakes in its estimation for the test samples. And, different from the classical RE, the near-optimal solution is produced based on the pruned error of all the available sequential subensembles. Besides, the backfitting step of RE algorithm is replaced with the selection step of a WSM in RRE. Moreover, the problem of ties might be solved more naturally with RRE. Finally, soft voting approach is employed in the testing to RRE algorithm. The performances of RE and RRE algorithms, and two baseline methods, i.e., the method which selects the Best Single Model (BSM) in the initial ensemble, and the method which retains all member networks of the initial ensemble (ALL), are evaluated on seven benchmark classification tasks under different initial ensemble setups. The results of the empirical investigation show the superiority of RRE over the other three ensemble pruning algorithms.     

Evolutionary collective behavior decomposition model for time series data mining

In this research, we propose a novel framework referred to as collective game behavior decomposition where complex collective behavior is assumed to be generated by aggregation of several groups of agents following different strategies and complexity emerges from collaboration and competition of individuals. The strategy of an agent is modeled by certain simple game theory models with limited information. Genetic algorithms are used to obtain the optimal collective behavior decomposition based on history data. The trained model can be used for collective behavior prediction. For modeling individual behavior, two simple games, the minority game and mixed game are investigated in experiments on the real-world stock prices and foreign-exchange rate. Experimental results are presented to show the effectiveness of the new proposed model.     

Effects of SO2 on selective catalytic reduction of NO with NH3 over a TiO2 photocatalyst

The effect of SO₂ gas was investigated on the activity of the photo-assisted selective catalytic reduction of nitrogen monoxide (NO) with ammonia (NH₃) over a TiO₂ photocatalyst in the presence of excess oxygen (photo-SCR). The introduction of SO₂ (300 ppm) greatly decreased the activity of the photo-SCR at 373 K. The increment of the reaction temperature enhanced the resistance to SO₂ gas, and at 553 K the conversion of NO was stable for at least 300 min of the reaction. X-ray diffraction, FTIR spectroscopy, thermogravimetry and differential thermal analysis, x-ray photoelectron spectroscopy (XPS), elemental analysis and N₂ adsorption measurement revealed that the ammonium sulfate species were generated after the reaction. There was a strong negative correlation between the deposition amount of the ammonium sulfate species and the specific surface area. Based on the above relationship, we concluded that the deposition of the ammonium sulfate species decreased the specific surface area by plugging the pore structure of the catalyst, and the decrease of the specific surface area resulted in the deactivation of the catalyst.     

TCBR-HMM: An HMM-based text classifier with a CBR system

This paper presents an innovative solution to model distributed adaptive systems in biomedical environments. We present an original TCBR-HMM (Text Case Based Reasoning-Hidden Markov Model) for biomedical text classification based on document content. The main goal is to propose a more effective classifier than current methods in this environment where the model needs to be adapted to new documents in an iterative learning frame. To demonstrate its achievement, we include a set of experiments, which have been performed on OSHUMED corpus. Our classifier is compared with Naive Bayes and SVM techniques, commonly used in text classification tasks. The results suggest that the TCBR-HMM Model is indeed more suitable for document classification. The model is empirically and statistically comparable to the SVM classifier and outperforms it in terms of time efficiency.     

Model checking temporal knowledge and commitments in multi-agent systems using reduction

Though modeling and verifying Multi-Agent Systems (MASs) have long been under study, there are still challenges when many different aspects need to be considered simultaneously. In fact, various frameworks have been carried out for modeling and verifying MASs with respect to knowledge and social commitments independently. However, considering them under the same framework still needs further investigation, particularly from the verification perspective. In this article, we present a new technique for model checking the logic of knowledge and commitments (CTLKC⁺). The proposed technique is fully-automatic and reduction-based in which we transform the problem of model checking CTLKC⁺ into the problem of model checking an existing logic of action called ARCTL. Concretely, we construct a set of transformation rules to formally reduce the CTLKC⁺ model into an ARCTL model and CTLKC⁺ formulae into ARCTL formulae to get benefit from the extended version of NuSMV symbolic model checker of ARCTL. Compared to a recent approach that reduces the problem of model checking CTLKC⁺ to another logic of action called GCTL¹, our technique has better scalability and efficiency. We also analyze the complexity of the proposed model checking technique. The results of this analysis reveal that the complexity of our reduction-based procedure is PSPACE-complete for local concurrent programs with respect to the size of these programs and the length of the formula being checked. From the time perspective, we prove that the complexity of the proposed approach is P-complete with regard to the size of the model and length of the formula, which makes it efficient. Finally, we implement our model checking approach on top of extended NuSMV and report verification results for the verification of the NetBill protocol, taken from business domain, against some desirable properties. The obtained results show the effectiveness of our model checking approach when the system scales up.     

Complexities of digital technology use and the teaching and learning of function

Many argue that digital technologies have the potential to enhance the teaching and learning of mathematics. However, the availability of technology is not sufficient to realise this potential. The study reported takes a detailed approach to investigate the utility of the particular offerings of the available technologies in the teaching and learning of a specific area of mathematics, functions. Sixteen affordances identified in the data are described. The complexity of the process involved in resolving a situation where particular affordances would be useful so as they are perceived and enacted is detailed. Finally, a grounded theory framework arising from the data analysis from this study that can be used to explain, predict and guide action in other digital environments is presented.     

The Interplay of Modulus,Strength, and Ductility in Adhesive Design Using Biomimetic Polymer Chemistry

High‐performance adhesives require mechanical properties tuned to demands of the surroundings. A mismatch in stiffness between substrate and adhesive leads to stress concentrations and fracture when the bonding is subjected to mechanical load. Balancing material strength versus ductility, as well as considering the relationship between adhesive modulus and substrate modulus, creates stronger joints. However, a detailed understanding of how these properties interplay is lacking. Here, a biomimetic terpolymer is altered systematically to identify regions of optimal bonding. Mechanical properties of these terpolymers are tailored by controlling the amount of a methyl methacrylate stiff monomer versus a similar monomer containing flexible poly(ethylene glycol) chains. Dopamine methacrylamide, the cross‐linking monomer, is a catechol moiety analogous to 3,4‐dihydroxyphenylalanine, a key component in the adhesive proteins of marine mussels. Bulk adhesion of this family of terpolymers is tested on metal and plastic substrates. Incorporating higher amounts of poly(ethylene glycol) into the terpolymer introduces flexibility and ductility. By taking a systematic approach to polymer design, the region in which material strength and ductility are balanced in relation to the substrate modulus is found, thereby yielding the most robust joints.     

The control of sea lice in Atlantic salmon by selective breeding

Sea lice threaten the welfare of farmed Atlantic salmon and the sustainability of fish farming across the world. Chemical treatments are the major method of control but drug resistance means that alternatives are urgently needed. Selective breeding can be a cheap and effective alternative. Here, we combine experimental trials and diagnostics to provide a practical protocol for quantifying resistance to sea lice. We then combined quantitative genetics with epidemiological modelling to make the first prediction of the response to selection, quantified in terms of reduced need for chemical treatments. We infected over 1400 young fish with Lepeophtheirus salmonis, the most important species in the Northern Hemisphere. Mechanisms of resistance were expressed early in infection. Consequently, the number of lice per fish and the ranking of families were very similar at 7 and 17 days post infection, providing a stable window for assessing susceptibility to infection. The heritability of lice numbers within this time window was moderately high at 0.3, confirming that selective breeding is viable. We combined an epidemiological model of sea lice infection and control on a salmon farm with genetic variation in susceptibility among individuals. We simulated 10 generations of selective breeding and examined the frequency of treatments needed to control infection. Our model predicted that substantially fewer chemical treatments are needed to control lice outbreaks in selected populations and chemical treatment could be unnecessary after 10 generations of selection. Selective breeding for sea lice resistance should reduce the impact of sea lice on fish health and thus substantially improve the sustainability of Atlantic salmon production.     

Model repair — aligning process models to reality

Process mining techniques relate observed behavior (i.e., event logs) to modeled behavior (e.g., a BPMN model or a Petri net). Process models can be discovered from event logs and conformance checking techniques can be used to detect and diagnose differences between observed and modeled behavior. Existing process mining techniques can only uncover these differences, but the actual repair of the model is left to the user and is not supported. In this paper we investigate the problem of repairing a process model w.r.t. a log such that the resulting model can replay the log (i.e., conforms to it) and is as similar as possible to the original model. To solve the problem, we use an existing conformance checker that aligns the runs of the given process model to the traces in the log. Based on this information, we decompose the log into several sublogs of non-fitting subtraces. For each sublog, either a loop is discovered that can replay the sublog or a subprocess is derived that is then added to the original model at the appropriate location. The approach is implemented in the process mining toolkit ProM and has been validated on logs and models from several Dutch municipalities.