Catálogo de publicaciones - revistas

<jats:p>The demand for artificial intelligence has grown significantly over the past decade, and this growth has been fueled by advances in machine learning techniques and the ability to leverage hardware acceleration. However, to increase the quality of predictions and render machine learning solutions feasible for more complex applications, a substantial amount of training data is required. Although small machine learning models can be trained with modest amounts of data, the input for training larger models such as neural networks grows exponentially with the number of parameters. Since the demand for processing training data has outpaced the increase in computation power of computing machinery, there is a need for distributing the machine learning workload across multiple machines, and turning the centralized into a distributed system. These distributed systems present new challenges: first and foremost, the efficient parallelization of the training process and the creation of a coherent model. This article provides an extensive overview of the current state-of-the-art in the field by outlining the challenges and opportunities of distributed machine learning over conventional (centralized) machine learning, discussing the techniques used for distributed machine learning, and providing an overview of the systems that are available.</jats:p>

<jats:p>Recently, Content Delivery Networks (CDN) have become more and more popular. The technology itself is ahead of academic research in this area. Several dimensions of the technology have not been adequately investigated by academia. These dimensions include outline management, security, and standardization. Discovering and highlighting aspects of this technology that may have or have not been covered by academic research is the first step toward helping academia bridge the gap with industry or even go one step further to lead industry in the right direction. This suggests a comprehensive survey on research works in this regard. The literature in this area has already come up with some surveys and taxonomies, but some of them are outdated or do not cover every aspect of CDN while others fail to detect existing trends or to develop a holistic roadmap for research on the technology. Furthermore, none of the existing surveys aim at enlightening the dark aspects of the technology that have not been subject to academic research. In this survey, we first extract the lifecycle of a CDN as suggested by the existing research. Then, we investigate previous relevant works on each phase of the lifecycle to clarify where the research is currently located and headed. We show how CDN technology tends to converge with emerging paradigms such as cloud computing, edge computing, and machine learning, which are more mature in terms of academic research. This helps us determine the right direction for further research by revealing the deficiencies in existing works.</jats:p>

<jats:p>Multilevel optimization aims at reducing the cost of executing a target network-based algorithm by exploiting coarsened, i.e., reduced or simplified, versions of the network. There is a growing interest in multilevel algorithms in networked systems, mostly motivated by the urge for solutions capable of handling large-scale networks. Notwithstanding the success of multilevel optimization in a multitude of application problems, we were unable to find a representative survey of the state-of-the-art, or consistent descriptions of the method as a general theoretical framework independent of a specific application domain. In this article, we strive to fill this gap, presenting an extensive survey of the literature that contemplates a systematic overview of the state-of-the-art, a panorama of the historical evolution and current challenges, and a formal theoretical framework of the multilevel optimization method in complex networks. We believe our survey provides a useful resource to individuals interested in learning about multilevel strategies, as well as to those engaged in advancing theoretical and practical aspects of the method or in developing solutions in novel application domains.</jats:p>

<jats:p>Holography is usually considered as the ultimate way to visually reproduce a three-dimensional scene. Computer-generated holography constitutes an important branch of holography, which enables visualization of artificially generated scenes as well as real three-dimensional scenes recorded under white-light illumination. In this article, we present a comprehensive survey of methods for synthesis of computer-generated holograms, classifying them into two broad categories: wavefront-based methods and ray-based methods. We examine their modern implementations in terms of the quality of reconstruction and computational efficiency. As it is an integral part of computer-generated holography, we devote a special section to speckle suppression, which is also discussed under two categories following the classification of underlying computer-generated hologram methods.</jats:p>

<jats:p>For over 25 years, most multi-objective evolutionary algorithms (MOEAs) have adopted selection criteria based on Pareto dominance. However, the performance of Pareto-based MOEAs quickly degrades when solving multi-objective optimization problems (MOPs) having four or more objective functions (the so-called many-objective optimization problems), mainly because of the loss of selection pressure. Consequently, in recent years, MOEAs have been coupled with indicator-based selection mechanisms in furtherance of increasing the selection pressure so that they can properly solve many-objective optimization problems. Several research efforts have been conducted since 2003 regarding the design of the so-called indicator-based (IB) MOEAs. In this article, we present a comprehensive survey of IB-MOEAs for continuous search spaces since their origins up to the current state-of-the-art approaches. We propose a taxonomy that classifies IB-mechanisms into two main categories: (1) IB-Selection (which is divided into IB-Environmental Selection, IB-Density Estimation, and IB-Archiving) and (2) IB-Mating Selection. Each of these classes is discussed in detail in this article, emphasizing the advantages and drawbacks of the selection mechanisms. In the final part, we provide some possible paths for future research.</jats:p>

<jats:p>As the demand for computing resources grows, cloud computing becomes more and more popular as a pay-as-you-go model, in which the computing resources and services are provided to cloud users efficiently. For cloud providers, the typical goal is to maximize their profits. However, maximizing profits in a highly competitive cloud market is a huge challenge for cloud providers. In this article, a survey of profit optimization techniques is proposed to increase cloud provider profitability through service quality improvement, service pricing, energy consumption reduction, and virtual network function (VNF) deployment. The strategy of improving user service quality is discussed first, followed by the pricing strategy for cloud resources to maximize revenue. Then, this article summarizes the techniques for cloud data centers to reduce server power consumption. Finally, various heuristic algorithms for VNF deployment in the cloud are further described to reduce the cost of cloud providers while maintaining performance. We classify research works based on components of profit and methods used to demonstrate similarities and differences in these studies. We hope this survey will provide researchers with insights into cloud profit optimization techniques.</jats:p>

<jats:p>In this survey, we propose to explore and discuss the common rules behind knowledge transfer works for vision recognition tasks. To achieve this, we firstly discuss the different kinds of reusable knowledge existing in a vision recognition task, and then we categorize different knowledge transfer approaches depending on where the knowledge comes from and where the knowledge goes. Compared to previous surveys on knowledge transfer that are from the problem-oriented perspective or from the technique-oriented perspective, our viewpoint is closer to the nature of knowledge transfer and reveals common rules behind different transfer learning settings and applications. Besides different knowledge transfer categories, we also show some research works that study the transferability between different vision recognition tasks. We further give a discussion about the introduced research works and show some potential research directions in this field.</jats:p>

<jats:p>Image classification is a key task in image understanding, and multi-label image classification has become a popular topic in recent years. However, the success of multi-label image classification is closely related to the way of constructing a training set. As active learning aims to construct an effective training set through iteratively selecting the most informative examples to query labels from annotators, it was introduced into multi-label image classification. Accordingly, multi-label active learning is becoming an important research direction. In this work, we first review existing multi-label active learning algorithms for image classification. These algorithms can be categorized into two top groups from two aspects respectively: sampling and annotation. The most important component of multi-label active learning is to design an effective sampling strategy that actively selects the examples with the highest informativeness from an unlabeled data pool, according to various information measures. Thus, different informativeness measures are emphasized in this survey. Furthermore, this work also makes a deep investigation on existing challenging issues and future promises in multi-label active learning with a focus on four core aspects: example dimension, label dimension, annotation, and application extension.</jats:p>

<jats:p>The use of reduced precision to improve performance metrics such as computation latency and power consumption is a common practice in the embedded systems field. This practice is emerging as a new trend in High Performance Computing (HPC), especially when new error-tolerant applications are considered. However, standard compiler frameworks do not support automated precision customization, and manual tuning and code transformation is the approach usually adopted in most domains. In recent years, research have been studying ways to improve the automation of this process. This article surveys this body of work, identifying the critical steps of this process, the most advanced tools available, and the open challenges in this research area. We conclude that, while several mature tools exist, there is still a gap to close, especially for tools based on static analysis rather than profiling, as well as for integration within mainstream, industry-strength compiler frameworks.</jats:p>

<jats:p>Virtualization-based technologies have become ubiquitous in computing. While they provide an easy-to-implement platform for scalable, high-availability services, they also introduce new security issues. Traditionally, discussions on security vulnerabilities in server platforms have been focused on stand-alone (i.e., non-virtualized) environments. For cloud and virtualized platforms, the discussion focuses on the shared usage of resources and the lack of control over the infrastructure. However, the impact virtualization technologies can have on exploit mitigation mechanisms of host machines is often neglected. Therefore, this survey discusses the following issues: first, the security issues and challenges that are introduced by the migration from stand-alone solutions to virtualized environments—special attention is given to the Virtual Machine Monitor, since it is a core component in a virtualized solution; second, the impact (sometimes negative) that these new technologies have on existing security strategies for hosts; third, how virtualization technologies can be leveraged to provide new security mechanisms not previously available.; and, finally, how virtualization technologies can be used for malicious purposes.</jats:p>