Large-scale global optimization (LSGO) has been an active research field. Part of this interest is supported by its application to cutting-edge research, such as Deep Learning, Big Data, and complex real-world problems, such as image encryption, real-time traffic management, and more. However, the high dimensionality makes solving LSGO a significant challenge. Some recent research deal with the high dimensionality by mapping the optimization process to a reduced alternative space. Nonetheless, these works suffer from the changes in the search space topology and the loss of information caused by the dimensionality reduction. This article proposes a hybrid metaheuristic, so-called low-dimensional space modeling-based differential evolution (LSMDE), that uses the singular value decomposition to build a low-dimensional search space from the features of candidate solutions generated by a new SHADE-based algorithm (GM-SHADE). GM-SHADE combines a Gaussian mixture model (GMM) and two specialized local algorithms: 1) MTS-LS1 and 2) L-BFGS-B, to promote a better exploration of the reduced search space. GMM mitigates the loss of information in mapping high-dimensional individuals to low-dimensional individuals. Furthermore, the proposal does not require prior knowledge of the search space topology, which makes it more flexible and adaptable to different LSGO problems. The results indicate that LSMDE is the most efficient method to deal with partially separable functions compared to other state-of-the-art algorithms and has the best overall performance in two of the three proposed experiments. Experimental results also show that the new approach achieves competitive results for nonseparable and overlapping functions on the most recent test suite for LSGO problems.

The hybridization of metaheuristics with data mining techniques has been successfully applied to combinatorial optimization problems. Examples of this type of strategy are DM-GRASP and MDM-GRASP, hybrid versions of the Greedy Randomized Adaptive Search Procedure (GRASP) metaheuristic, which incorporate data mining techniques. This type of hybrid method is called Data-Driven Metaheuristics and aims at extracting useful knowledge from the data generated by metaheuristics in their search process. Despite success in combinatorial problems like the set packing problem and maximum diversity problem, proposals of this type to solve continuous optimization problems are still scarce in the literature. This work presents a data mining hybrid version of C-GRASP, an adaptation of GRASP for problems with continuous variables. We call this new version DMC-GRASP, which identifies patterns in high-quality solutions and generates new solutions guided by these patterns. We performed computational experiments with DMC-GRASP on a set of well-known mathematical benchmark functions, and the results showed that metaheuristics for continuous optimization could also benefit from using patterns to guide the search for better solutions.

Simultaneous localization and mapping (SLAM) is a fundamental problem in mobile robotics. One solution to this problem is RatSLAM, which is a SLAM algorithm inspired by the navigation system in rodent brains. RatSLAM has a set of parameters that have to be adjusted individually for each environment to generate a suitable map. To date, there is no automatic tuning of these parameters and they are hand-tuned by trial-and-error. The present work proposes an automatic parameter tuning method, which employs the Iterative Closest Point (ICP) algorithm to evaluate the generated maps as well as the Irace algorithm to find the best parameters settings. The process is formulated as an optimization problem, where the objective function takes into account the parameters values, the generated RatSLAM's maps, and a ground truth map. Irace automatically generates a unique combination of RatSLAM parameter set, where each combination generates a map. Then, deviations between the generated and a ground truth map are evaluated with the ICP algorithm. These evaluations are taken into account by Irace to generate new combinations of RatSLAM parameters until this deviation reaches a stop criterion. The proposed methodology was successfully tested on three different scenarios. It was able to automatically estimate parameters to generate experience maps close to the ground truth maps. Furthermore, it showed how the tuned parameters can be shared across different scenarios, which might pave the way to develop similar approaches for other SLAM algorithms that also depend on fine-tuning of their parameters.

This paper presents a neuro-inspired mapping approach that uses partial information shared by multiple robots to reduce the time to create a map of an entire environment. Robots using a neurobiologically inspired algorithm, namely RatSLAM, map an environment sharing video information among themselves. RatSLAM, which is based on the navigation system present in the hippocampus of rodents' brain, has been widely used on simultaneous localization and mapping (SLAM) problem. This proposal has been able to generate suitable maps mainly when there is redundant information, e.g. a scene is seen more than once, since this fact activates local view cells that inject activity inside the pose cells via an excitatory link. The work here reported extends this approach by merging partial information acquired by multiple robots. The results from the performed experiments show that the final map built by two robots with shared information is similar to one built by two robots performing the same mapping task individually, i.e. without sharing information. However, the time spent to generate the whole map with the proposed shared approach was smaller than the one without the shared information. Thus, the current approach allows creating a complete map of an environment within a reduced time using multiple robots.

Autonomous exploration aims to develop mechanisms that enable a robot to navigate through an unknown environment, seeking to collect information that is relevant to its objective, with minimal or no human intervention. This article presents an autonomous exploration strategy based on a multimodal continuous utility function and optimization metaheuristic. The problem of autonomous exploration of mobile robots is formulated as an optimization problem, providing data for a Firefly-based algorithm that is able to search for points in the solution space, representing cells on the map being constructed, that best meet the exploration objectives. Simulations in a Robot Operating System environment showed that the proposed approach is able to guide robot through good trajectories, leading to full mapping, without any human intervention.

Clustering Search (*CS) has been proposed as a generic way of combining search metaheuristics with clustering to detect promising search areas before applying local search procedures. The clustering process may keep representative solutions associated to different search subspaces (search areas). In this work, new approaches are proposed, based on Artificial Bee Colony (ABC) and Differential Evolution (DE), observing the inherent characteristics of detecting promising food sources employed by that metaheuristic. The proposed hybrid algorithms, performing a Hooke & Jeeves based local, are compared against another hybrid versions of ABC and DE, exploring an elitist criteria.

The web has become an important medium for access to information and users increasingly depend on it for their routine activities. Hence, the demand for higher quality services has grown strong and overload control has become a key problem to be addressed in web servers. This paper analyzes the implementation of a fuzzy admission controller in a QoS-enabled web server architecture and compares its results to traditional approaches. Our fuzzy controller allows better system performance tuning and contributes to smaller and more stable response times regarding http request processing. The fuzzy approach also provides added flexibility to system administration, as different associative matrices can be used, thus customizing the web server to different workload profiles.

This paper describes an application of a Constructive Genetic Algorithm (CGA) to the gate matrix layout problem (GMLP). The GMLP happens in very large scale integration design and can be described as a problem of assigning a set of circuit nodes (gates) in an optimal sequence, such that the layout area is minimized, as a consequence of optimizing the number of tracks necessary to cover the gates interconnection. The CGA has a number of new features compared to a traditional genetic algorithm. These include a population of dynamic size composed of schemata and structures and the possibility of using heuristics in structure representation and in the fitness function definitions. The application of CGA to GMLP uses a 2-Opt-like heuristic to define the fitness functions and the mutation operator. Computational tests are presented using available instances taken from the literature.