Incremental Learning on Non-stationary Data Stream Using Ensemble Approach

Thalor, Meenakshi Anurag; Patil, Shrishailapa
August 2016
International Journal of Electrical & Computer Engineering (2088;Aug2016, Vol. 6 Issue 4, p1811
Academic Journal
Incremental Learning on non stationary distribution has been shown to be a very challenging problem in machine learning and data mining, because the joint probability distribution between the data and classes changes over time. Many real time problems suffer concept drift as they changes with time. For example, an advertisement recommendation system, in which customer's behavior may change depending on the season of the year, on the inflation and on new products made available. An extra challenge arises when the classes to be learned are not represented equally in the training data i.e. classes are imbalanced, as most machine learning algorithms work well only when the training data is balanced. The objective of this paper is to develop an ensemble based classification algorithm for non-stationary data stream (ENSDS) with focus on two-class problems. In addition, we are presenting here an exhaustive comparison of purposed algorithms with state-of-the-art classification approaches using different evaluation measures like recall, fmeasure and g-mean.


Related Articles

  • A Binary-Classification-Based Metric between Time-Series Distributions and Its Use in Statistical and Learning Problems. Ryabko, Daniil; Mary, Jérémie // Journal of Machine Learning Research;Sep2013, Vol. 14, p2837 

    A metric between time-series distributions is proposed that can be evaluated using binary classification methods, which were originally developed to work on i.i.d. data. It is shown how this metric can be used for solving statistical problems that are seemingly unrelated to classification and...

  • Single Trajectory Learning: Exploration Versus Exploitation. Fu, Qiming; Liu, Quan; Zhong, Shan; Luo, Heng; Wu, Hongjie; Chen, Jianping // International Journal of Pattern Recognition & Artificial Intell;Jun2018, Vol. 32 Issue 6, p-1 

    In reinforcement learning (RL), the exploration/exploitation (E/E) dilemma is a very crucial issue, which can be described as searching between the exploration of the environment to find more profitable actions, and the exploitation of the best empirical actions for the current state. We focus...

  • A Taxonomy of Label Ranking Algorithms. Yangming Zhou; Yangguang Liu; Jiangang Yang; Xiaoqi He; Liangliang Liu // Journal of Computers;Mar2014, Vol. 9 Issue 3, p557 

    The problem of learning label rankings is receiving increasing attention from machine learning and data mining community. Its goal is to learn a mapping from instances to rankings over a finite number of labels. In this paper, we devote to giving an overview of the state-of-the-art in the area...

  • Considering Autocorrelation in Predictive Models. Stojanova, Daniela // Informatica (03505596);Mar2013, Vol. 37 Issue 1, p107 

    This article presents a summary of the doctoral dissertation of the author, which addresses the task of considering autocorrelation in predictive models.

  • A Feature Selection Approach of Inconsistent Decision Systems in Rough Set. Lin Sun; Jiucheng Xu; Yuhui Li // Journal of Computers;Jun2014, Vol. 9 Issue 6, p1333 

    Feature selection has been widely discussed as an important preprocessing step in data mining applications since it reduces a model's complexity. In this paper, limitations of several representative reduction methods are analyzed firstly, and then by distinguishing consistent objects form...

  • A Feature Selection Method from the View Of Complex Network. Ying He; Yan Tang // International Journal of Database Theory & Application;Apr2014, Vol. 7 Issue 2, p71 

    Feature selection as a common method of dimensionality reduction always is one of the hot topics in machine learning and data mining field. Classic algorithms don't consider features' global redundancies fully, which may cause classification accuracy on selected feature subset to be not high...

  • ASSIGNING WEIGHTS TO TRAINING INSTANCES INCREASES CLASSIFICATION ACCURACY. Farid, Dewan Md.; Rahman, Chowdhury Mofizur // International Journal of Data Mining & Knowledge Management Proc;Jan2013, Vol. 3 Issue 1, p13 

    The decision tree (DT) approach is most useful in classification problem. In conventional decision tree learning the weights of every training instances are set to one or equal value, which contradicts general intuition. In this paper, we proposed a new decision tree learning algorithm by...

  • A VSS Algorithm based on Multiple Features for Object Tracking. Bin Xu; Xiaoju Shen; Feiji Ding // Journal of Software (1796217X);Dec2013, Vol. 8 Issue 12, p3029 

    A variable search space (VSS) approach according to the color feature combined with point feature for object tracking is presented. Mean shift is a well-established and fundamental algorithm that works on the basis of color probability distributions, and is robust to given color targets. As it...

  • Distributed consensus for metamorphic systems using a gossip algorithm for CAT (0) metric spaces. Bellachehab, Anass; Jakubowicz, Jérémie // AIP Conference Proceedings;2015, Vol. 1641 Issue 1, p530 

    We present an application of distributed consensus algorithms to metamorphic systems. A metamorphic system is a set of identical units that can self-assemble to form a rigid structure. For instance, one can think of a robotic arm composed of multiple links connected by joints. The system can...


Read the Article


Sorry, but this item is not currently available from your library.

Try another library?
Sign out of this library

Other Topics