A framework for increasing the value of predictive data-driven models by enriching problem domain characterization with novel features

The need to leverage knowledge through data mining has driven enterprises in a demand for more data. However, there is a gap between the availability of data and the application of extracted knowledge for improving decision support. In fact, more data do not necessarily imply better predictive data-driven marketing models, since it is often the case that the problem domain requires a deeper characterization. Aiming at such characterization, we propose a framework drawn on three feature selection strategies, where the goal is to unveil novel features that can effectively increase the value of data by providing a richer characterization of the problem domain. Such strategies involve encompassing context (e.g., social and economic variables), evaluating past history, and disaggregate the main problem into smaller but interesting subproblems. The framework is evaluated through an empirical analysis for a real bank telemarketing application, with the results proving the benefits of such approach, as the area under the receiver operating characteristic curve increased with each stage, improving previous model in terms of predictive performance.

     

Generalized Hermitian Radial Basis Functions Implicits from polygonal mesh constraints

In this work we investigate a generalized interpolation approach using radial basis functions to reconstruct implicit surfaces from polygonal meshes. With this method, the user can define with great flexibility three sets of constraint interpolants: points, normals, and tangents; allowing to balance computational complexity, precision, and feature modeling. Furthermore, this flexibility makes possible to avoid untrustworthy information, such as normals estimated on triangles with bad aspect ratio. We present results of the method for applications related to the problem of modeling 2D curves from polygons and 3D surfaces from polygonal meshes. We also apply the method to problems involving subdivision surfaces and front-tracking of moving boundaries. Finally, as our technique generalizes the recently proposed HRBF Implicits technique, comparisons with this approach are also conducted.     

Improving the performance of fuzzy rules-based forecasters through application of FCM algorithm

Prediction models based on artificial intelligence techniques have been widely used in Time Series Forecasting in several areas. They are often fuzzy models or neural networks. This paper describes the development of neural and fuzzy models for forecasting time series of practical examples, and shows the comparisons of results between models, including the results of statistical modeling. The use of data clustering algorithms like Fuzzy C-Means is considered in fuzzy models.     

One‐Step Formulation of Protein Microparticles with Tailored Properties: Hard Templating at Soft Conditions

Formulation of therapeutic proteins into particulate forms is a main strategy for site‐specific and prolonged protein delivery as well as for protection against degradation. Precise control over protein particle size, dispersity, purity, as well as mild preparation conditions and minimal processing steps are highly desirable. It is, however, hard to fit all these criteria with conventional preparation techniques. Here a one‐step hard‐templating synthesis of microparticles composed of functional, non‐denatured protein is reported. The method is based on filling porous CaCO₃ microtemplates with the protein near to its isoelectric point (pI) followed by pH‐ or EDTA‐mediated dissolution of the tempplates. In principle, a wide variety of proteins can be converted into microparticles using this approach. The main requirement is an overlap of the protein insolubility and a template solubility for a certain parameter (here pH or EDTA). Here the formulation of insulin particles is studied in detail and it is shown that particles consisting of high molecular weight protein (catalase) can also be prepared. In this context, the synthesis of CaCO₃ templates with controlled size, the mechanism of the protein microparticle formation and mechanical properties of the microparticles are discussed. For the first time, the fabrication of mesoporous monodispersed CaCO₃ microtemplates with identical porocity but tuned diameter from 3 to 20 μm is demonstrated. The protein particle diameter can be adjusted by choosing the appropriate template size that is critical for successful pulmonary delivery of insulin. As a first step towards insulin delivery, the in vitro release of insulin at physiological conditions is studied.     

Gain scheduling adaptive control applied to a particular mixer-settler equipment

The mixer-settler based on phase inversion (MSBPI) is a system used for treatment of wastewaters in order to reduce the content of oils and greases from wastewaters produced in oil industry. Its operation principle combines mixing, extraction and separation by decantation in a unique process. In particular, for control purpose, the level of the interface between organic solvent and water, in the separation section, is a critical variable to guarantee performance and security during operation of the MSBPI. In this work, an image-based detector to monitor the organic solvent–water interface level is proposed and a feedback control loop for the process is presented. Initially, a conventional controller (PID) was implemented and, subsequently, a strategy of gain scheduling adaptive control was developed in order to improve the process response in closed-loop. The experimental results showed that the adaptive control presented an efficient performance, even when disturbances of significant magnitude were applied to the process.     

Benchmarking flexible job-shop scheduling and control systems

Benchmarking is comparing the output of different systems for a given set of input data in order to improve the system’s performance. Faced with the lack of realistic and operational benchmarks that can be used for testing optimization methods and control systems in flexible systems, this paper proposes a benchmark system based on a real production cell. A three-step method is presented: data preparation, experimentation, and reporting. This benchmark allows the evaluation of static optimization performances using traditional operation research tools and the evaluation of control system's robustness faced with unexpected events.     

A new multi-objective optimization method for master production scheduling problems using simulated annealing

Tactic planning or master production scheduling focuses on time and spatial decomposition of the aggregate planning targets and forecasts, as well as, forecast and provision of needed resources. This process becomes extremely hard and time consuming with the increase of number of products, resources and periods considered. In face of such obstacles, this work shows a study of an Artificial Intelligence technique called Simulated Annealing applied to the optimization of production planning problem, more specifically, Master Production Scheduling. This work reviews some of the fundamental theory of simulated annealing, the methodology for master production scheduling calculation, the applicability of simulating annealing to planning problems, most important results and suggestions for further studies.