On achieving intelligent traffic-aware consolidation of virtual machines in a data center using Learning Automata

Jobava, Akaki; Yazidi, Anis; Oommen, B. John; Begnum, Kyrre

doi:10.1016/j.jocs.2017.08.005

Cited by 15 publications

(8 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The sequence imposes an order on the samples; this order must be preserved when training models and making predictions [1]. The goal being to extract knowledge from a continuous sequence of data records, e.g., financial markets, network traffic, weather conditions, among others [2,3,4].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Learning from data streams using kernel least-mean-square with multiple kernel-sizes and adaptive step-size

2019

View full text Add to dashboard Cite

A learning task is sequential if its data samples become available over time; kernel adaptive filters (KAFs) are sequential learning algorithms. There are three main challenges in KAFs: (1) selection of an appropriate Mercer kernel; (2) the lack of an effective method to determine kernel-sizes in an online learning context; (3) how to tune the step-size parameter. This work introduces a framework for online prediction that addresses the latter two of these open challenges. The kernel-sizes, unlike traditional KAF formulations, are both created and updated in an online sequential way. Further, to improve convergence time, we propose an adaptive step-size strategy that minimizes the mean-square-error (MSE) using a stochastic gradient algorithm. The proposed framework has been tested on three real-world data sets; results show both faster convergence to relatively low values of MSE and better accuracy when compared with KAF-based methods, long short-term memory, and recurrent neural networks.

show abstract

Section: Introductionmentioning

confidence: 99%

“…The dataset is publicly available at http://www3.dsi.uminho.pt/pcortez/series/A5M.txt2 The dataset is publicly available at https://www.dukascopy.com3 The dataset is publicly available at https://www.kaggle.com/rpaguirre/tesla-stock-price…”

mentioning

confidence: 99%

Learning from data streams using kernel least-mean-square with multiple kernel-sizes and adaptive step-size

2019

View full text Add to dashboard Cite

show abstract

“…LA is a subclass of Reinforcement Learning (RL), which focuses on training autonomous agents in an interactive environment in order to optimize cumulative rewards. LA have applications in evolutionary optimization [29], Cloud and Grid computing [30], [31], [32], social networks [33], image processing [34], and data clustering [35], to mention a few.…”

Section: Introductionmentioning

confidence: 99%

Partitioning Graphs for the Cloud using Reinforcement Learning

Mofrad,

Melhem,

Hammoud

2019

Preprint

View full text Add to dashboard Cite

In this paper, we propose Revolver, a parallel graph partitioning algorithm capable of partitioning large-scale graphs on a single shared-memory machine. Revolver employs an asynchronous processing framework, which leverages reinforcement learning and label propagation to adaptively partition a graph. In addition, it adopts a vertex-centric view of the graph where each vertex is assigned an autonomous agent responsible for selecting a suitable partition for it, distributing thereby the computation across all vertices. The intuition behind using a vertex-centric view is that it naturally fits the graph partitioning problem, which entails that a graph can be partitioned using local information provided by each vertex's neighborhood. We fully implemented and comprehensively tested Revolver using nine real-world graphs. Our results show that Revolver is scalable and can outperform three popular and state-of-the-art graph partitioners via producing comparable localized partitions, yet without sacrificing the load balance across partitions. s

show abstract

“…• In the field of cloud computing, there is need to distribute traffic across multiple Virtual Machines efficiently. An OMA-based solution, when applied to this problem, was superior to other solutions in the literature, leading to a 90% reduction in performance cost [28,62].…”

Section: The Case For Partitioningmentioning

confidence: 99%

“…28 shows the rate of convergence of the TPEOMA-augmented hierarchical schemes in the Zipf distribution. The key observation from this was that the TPEOMA enhanced scheme quickly converged with the first few queries, and was superior to the MTF and TR standalone schemes.…”

mentioning

confidence: 99%

On Designing Adaptive Data Structures with Adaptive Data "Sub"-Structures

Bisong¹

View full text Add to dashboard Cite

Data structures are key pillars for optimizing computational efficiency, as they contribute in no small measure to enhancing the "speed" in the accessing and subsequent processing of data. The need for enhancing speed is critical for almost all applications and domains, and this is most relevant when real-time or near real-time efficiency is desired. This thesis proposes the use of "Adaptive" Data-Structures (ADSs) that invoke reinforcement learning schemes from the theory of Learning Automata (LA). These operate in conjunction with select reorganization rules to update themselves as they receive queries from the Environment of interaction. The result of such a process is the subsequent minimization of the cost associated with query accesses. The Environments under consideration are those that exhibit a so-called "locality of reference", and are referred to as Non-stationary Environments (NSEs). A hierarchy of data "sub"-structures is used to design Singly-Linked Lists (SLLs) on Singly-Linked Lists, which thus contain outer and sub-list contexts. The elements that are more likely to be accessed together are grouped within the same sub-context, while the sub-lists themselves are moved "en masse" towards the head of the listcontext by following a reorganization strategy. The Object Migration Automaton (OMA) family of reinforcement schemes are employed to capture the probabilistic dependence of the elements in the data structure as it receives query accesses from the Environment. The Enhanced Object Migration Automaton (EOMA), the Pursuit Enhanced Object Migration Automaton (PEOMA), and the Transitivity Pursuit Enhanced Object Migration Automaton (TPEOMA) have each been individually incorporated into the hierarchical SLLs. The consequent results are currently the state-of-the-art methods for adaptive SLLs operating in NSEs. iii Finally, I will want to especially thank my Uncle and Aunt, Achu and Blessing Bisong, for their love, care, support and encouragement throughout the time of this thesis, and indeed, my stay in Canada. They have been a rock in my life. I am grateful to the faculty, the staff and the friends I made at the School of Computer Science at Carleton. They made my stay here enjoyable and fulfilling. iv Bibliography Appendices A OMA-Augmented Hierarchical SLLs Results B EOMA-Augmented Hierarchical SLLs Results C PEOMA-Augmented Hierarchical SLLs Results D TPEOMA-Augmented Hierarchical SLLs Results viii

show abstract

On achieving intelligent traffic-aware consolidation of virtual machines in a data center using Learning Automata

Cited by 15 publications

References 40 publications

Learning from data streams using kernel least-mean-square with multiple kernel-sizes and adaptive step-size

Learning from data streams using kernel least-mean-square with multiple kernel-sizes and adaptive step-size

Partitioning Graphs for the Cloud using Reinforcement Learning

On Designing Adaptive Data Structures with Adaptive Data "Sub"-Structures

Contact Info

Product

Resources

About