Wisdom of Crowds cluster ensemble

📝 Abstract

The Wisdom of Crowds is a phenomenon described in social science that suggests four criteria applicable to groups of people. It is claimed that, if these criteria are satisfied, then the aggregate decisions made by a group will often be better than those of its individual members. Inspired by this concept, we present a novel feedback framework for the cluster ensemble problem, which we call Wisdom of Crowds Cluster Ensemble (WOCCE). Although many conventional cluster ensemble methods focusing on diversity have recently been proposed, WOCCE analyzes the conditions necessary for a crowd to exhibit this collective wisdom. These include decentralization criteria for generating primary results, independence criteria for the base algorithms, and diversity criteria for the ensemble members. We suggest appropriate procedures for evaluating these measures, and propose a new measure to assess the diversity. We evaluate the performance of WOCCE against some other traditional base algorithms as well as state-of-the-art ensemble methods. The results demonstrate the efficiency of WOCCE’s aggregate decision-making compared to other algorithms.

💡 Analysis

The Wisdom of Crowds is a phenomenon described in social science that suggests four criteria applicable to groups of people. It is claimed that, if these criteria are satisfied, then the aggregate decisions made by a group will often be better than those of its individual members. Inspired by this concept, we present a novel feedback framework for the cluster ensemble problem, which we call Wisdom of Crowds Cluster Ensemble (WOCCE). Although many conventional cluster ensemble methods focusing on diversity have recently been proposed, WOCCE analyzes the conditions necessary for a crowd to exhibit this collective wisdom. These include decentralization criteria for generating primary results, independence criteria for the base algorithms, and diversity criteria for the ensemble members. We suggest appropriate procedures for evaluating these measures, and propose a new measure to assess the diversity. We evaluate the performance of WOCCE against some other traditional base algorithms as well as state-of-the-art ensemble methods. The results demonstrate the efficiency of WOCCE’s aggregate decision-making compared to other algorithms.

📄 Content

1 Wisdom of Crowds Cluster Ensemble Hosein Alizadeh1, Muhammad Yousefnezhad2 and Behrouz Minaei Bidgoli3

Abstract: The Wisdom of Crowds is a phenomenon described in social science that suggests four criteria applicable to groups of people. It is claimed that, if these criteria are satisfied, then the aggregate decisions made by a group will often be better than those of its individual members. Inspired by this concept, we present a novel feedback framework for the cluster ensemble problem, which we call Wisdom of Crowds Cluster Ensemble (WOCCE). Although many conventional cluster ensemble methods focusing on diversity have recently been proposed, WOCCE analyzes the conditions necessary for a crowd to exhibit this collective wisdom. These include decentralization criteria for generating primary results, independence criteria for the base algorithms, and diversity criteria for the ensemble members. We suggest appropriate procedures for evaluating these measures, and propose a new measure to assess the diversity. We evaluate the performance of WOCCE against some other traditional base algorithms as well as state-of-the-art ensemble methods. The results demonstrate the efficiency of WOCCE’s aggregate decision- making compared to other algorithms. Keywords: Ensemble Cluster, Wisdom of Crowds, Diversity, Independence.

1. Introduction Clustering, one of the main tasks of data mining, is used to group non-labeled data to find meaningful patterns. Generally, different models provide predictions with different accuracy rates. Thus, it would be more efficient to develop a number of models using, different data subsets, or utilizing differing conditions within the modeling methodology of choice, to achieve better results. However, selecting the best model is not necessarily the ideal choice because potentially valuable information may be wasted by discarding the results of less-successful models (Perrone and Cooper, 1993; Tumer and Ghosh, 1996; Baker and Ellison, 2008).

1Hosein Alizadeh Department of Computer Eng., Iran University of Science and Technology, Tehran, Iran e-mail: halizadeh@iust.ac.ir

2Muhammad Yousefnezhad Computer Science and Technology Department, Nanjing University of Aeronautics and Astronautics, Nanjing, China e-mail: myousefnezhad@nuaa.edu.cn

3Behrouz Minaei-Bidgoli Department of Computer Eng., Iran University of Science and Technology, Tehran, Iran e-mail: b_minaei@iust.ac.ir 2 This leads to the concept of combining, where the outputs (individual predictions) of several models are pooled to make a better decision (collective prediction) (Tumer and Ghosh, 1996; Baker and Ellison, 2008). Research in the Clustering Combination field has shown that these pooled outputs have more strength, novelty, stability, and flexibility than the results provided by individual algorithms (Strehl and Ghosh, 2002; Topchy et al., 2003; Fred and Lourenco, 2008; Ayad and Kamel, 2008).
In the classic cluster ensemble selection methods, a consensus metric is used to audit the basic results in cluster ensemble selection and use them to produce the final result. There are two problems in the traditional methods; firstly, although the final result is always in accordance with the selected metrics providing the optimized result, there might be other metrics by which a better final result can be generated. Secondly, In order to produce the final result, there is neither any information from other entities of cluster ensemble except auditing basic results and nor that any evaluation of information and errors in other entities can be presented. In order to solve the mentioned problems, this paper proposes wised clustering (WOCCE) as a viable solution. This method audits all entities of cluster ensemble and the errors in result from each entity optimized by information obtained from other entities which dramatically reduces the possibility of any errors to occur in complex data as the result.
In the social science arena, there is a corresponding research field known as the Wisdom of Crowds, after the book by the same name (Surowiecki, 2004), simply claiming that the Wisdom of Crowds (WOC) is the phenomenon whereby decisions made by aggregating the information of groups usually have better results than those made by any single group member. The book presents numerous case studies and anecdotes to illustrate its argument, and touches on several fields, in particular economy and psychology. Surowiecki justifies his own theory, stating that: “If you ask a large enough groups of diverse and independent people to make a prediction or estimate a probability, the average of those answers, will cancel out errors in individual estimation. Each person’s guess, you might say, has two components: information and errors. Subtract the errors, and you’re left with the information” (Surowiecki, 2004). In spite of the lack o

This content is AI-processed based on ArXiv data.