Abstract
Lot sizing and scheduling by mixed integer programming has been a hot research topic inthe last 20 years. Researchers have been trying to develop stronger formulations, as well as to incorporatereal-world requirements from different applications. This paper illustrates some of these requirements anddemonstrates how small- and big-bucket models have been adapted and extended. Motivation comes fromdifferent industries, especially from process and fast-moving consumer goods industries. © 2015 Brazilian Operations Research Society.

Abstract
Lot sizing and scheduling by mixed integer programming has been a hot research topic in the last 20 years. Researchers have been trying to develop stronger formulations, as well as to incorporate real-world requirements from different applications. In this paper we illustrate some of these requirements and show howmodels have been adapted and extended. Motivation comes from different industries, especially from process and fast moving consumer goods industries.

Abstract
Unicer, a major Portuguese beverage company, improved its tactical distribution planning decisions and study alternative scenarios for its supply strategies and network configuration as result of an operations research (OR)-driven process. In this paper, we present the decision support system responsible for this new methodology. At the core of this system is a mathematical programming-based heuristic that includes decision variables that address transportation and inventory management problems. Unicer runs a set of production and distribution platforms with various characteristics to fulfill customers demand. The main challenge of our work was to develop a tactical distribution plan, which Unicer calls an annual distribution budget, as realistically as possible without jeopardizing the nature of the strategic and tactical tool. The company had a complex tactical distribution planning problem because of the increasing variety of its stock-keeping units and its need for a flexible distribution network to satisfy its customers, who demand a very fragmented set of products. Atypical flows of finished products from Unicer's distribution centers to its production platforms are a major cause of this complexity, which yields an intricate supply chain. The quality of the solutions we provided and the implementation of a user-friendly interface and editable inputs and outputs for our decision support system motivated company practitioners to use it. Unicer saves approximately two million euros annually and provides better information to its decision makers. As a result, these decision makers now view their operations from a more OR-based perspective.

Abstract
This work addresses an integrated framework for deciding about the supplier selection in processed food supply chains that accounts for tactical production and distribution planning. We are especially concerned with the option of producing with local or mainstream raw materials. The contribution of this paper is two-fold. Firstly, it proposes a new multi-objective two-stage stochastic mixed-integer programming model for the supplier selection that maximizes the profit and minimizes the risk of a low customer service. Secondly, the main complexities of processed food supply chains management are considered: perishability of raw materials and final products, uncertainty at downstream and upstream parameters, and customer willingness to pay. Results indicate that dual sourcing is a strategy to be pursued across several scenarios. The multi-objective approach shows that a small decrease in the expected value of profit results in a significant increase in the customer service. Acknowledging the increase in customers willing to pay for local products is also fundamental.

Abstract
The possibilities of combining simulation and optimization are vast and the appropriate design highly depends on the problem characteristics. Therefore, it is very important to have a good overview of the different approaches. The taxonomies and classifications proposed in the literature do not cover the complete range of methods and overlook some important criteria. We provide a taxonomy that aims at giving an overview of the full spectrum of current simulation-optimization approaches. Our study may guide researchers who want to use one of the existing methods, give insights into the cross-fertilization of the ideas applied in those methods and create a standard for a better communication in the scientific community. Future reviews can use the taxonomy here described to classify both general approaches and methods for specific application fields.

Abstract
An unsupervised method for convolutional neural network (CNN) architecture design is proposed. The method relies on a variable neighborhood search-based approach for finding CNN architectures and hyperparameter values that improve classification performance. For this purpose, t-Distributed Stochastic Neighbor Embedding (t-SNE) is applied to effectively represent the solution space in 2D. Then, k-Means clustering divides this representation space having in account the relative distance between neighbors. The algorithm is tested in the CIFAR-10 image dataset. The obtained solution improves the CNN validation loss by over 15% and the respective accuracy by 5%. Moreover, the network shows higher predictive power and robustness, validating our method for the optimization of CNN design.