Empirical Comparison of Multi-Label Classification Algorithms

  title={Empirical Comparison of Multi-Label Classification Algorithms},
  author={Clifford A. Tawiah and Victor S. Sheng},
  journal={Proceedings of the AAAI Conference on Artificial Intelligence},
Multi-label classifications exist in many real world applications. This paper empirically studies the performance of a variety of multi-label classification algorithms. Some of them are developed based on problem transformation. Some of them are developed based on adaption. Our experimental results show that the adaptive Multi-Label K-Nearest Neighbor performs the best, followed by Random k-Label Set, followed by Classifier Chain and Binary Relevance. Adaboost.MH performs the worst… 

Tables from this paper

Comparative Experimental Study of Multi Label Classification using Single Label Ground Truth with Application to Field Majoring Problem
It can be concluded that in certain problems, multi label classification methods can train single label and still produce multi label predictions with fairly good accuracy.
Comparative evaluation of four multi‐label classification algorithms in classifying learning objects
With the increasing number of learning objects (LOs), the possibility of their fast and effective retrieving and storing has become a more critical issue. The classification of LOs enables users to
Binary relevance (BR) method classifier of multi-label classification for arabic text
This study aims to design and develop a new method for multi-label text classification for Arabic texts based on a binary relevance method made up from a different set of machine learning classifiers.
Multi-class wine grades predictions with hierarchical support vector machines
This work achieves high accuracy on predicting a wine's grade through linear kernel support vector machines (SVMs) using a multi-class problem with four classes: 100∼95, 94∼90, 89∼85 and 84 below.
Classification on grade, price, and region with multi-label and multi-target methods in wineinformatics
This work compares several different multi-label/multitarget methods to the single-label method where each label is treated independently, and finds that per-label accuracy has the best performance, although some multi- label methods approach the performance of single- label.
MapReduce-based adaptive random forest algorithm for multi-label classification
The traditional decision tree algorithm was improved, and the traditional RF method was converted into an adaptive RF (ARF) method for multi-label classification, which was implemented on the cloud platform and the effectiveness of the proposed method was verified.
Multi-Label Prediction for Political Text-as-Data
This work applies the multi-label prediction framework to code multiple features of access to information requests made to the Mexican government and country-year human rights reports, and finds that multi- label prediction outperforms standard supervised learning approaches, even in instances where the correlations among one’s multiple labels are low.
A Multi-Label Computer-aided Diagnoses System for Detecting and Diagnosing Diabetic Retinopathy
A multi-label computer-aided diagnosis (ML-CAD) system to apply MLC for different DR grades using color fundus images and outperforms the other diagnosing DR systems.
Improving occupancy presence prediction via multi-label classification
This paper examines the novel use of multi-label classification (MLC) for predicting occupancy of rooms based on data from motion sensors and concludes that SVM provides a more robust performance than other algorithms with a significantly higher count of highest prediction accuracy for observed scenarios.
Student Performance Prediction in Mathematics Course Based on the Random Forest and Simulated Annealing
This study provides a methodological reference for the prediction of student achievement and also makes a marginal contribution to student management work.


Classifier chains for multi-label classification
This paper presents a novel classifier chains method that can model label correlations while maintaining acceptable computational complexity, and illustrates the competitiveness of the chaining method against related and state-of-the-art methods, both in terms of predictive performance and time complexity.
Random k-Labelsets for Multilabel Classification
Empirical evidence indicates that RAkEL manages to improve substantially over LP, especially in domains with large number of labels and exhibits competitive performance against other high-performing multilabel learning methods.
A Pruned Problem Transformation Method for Multi-label Classification
A Pruned Problem Transformation method (PPT) is presented, along with several exten- sions, designed to overcome such disadvantages as well as their respective advantages and disadvantages.
BoosTexter: A Boosting-based System for Text Categorization
This work describes in detail an implementation, called BoosTexter, of the new boosting algorithms for text categorization tasks, and presents results comparing the performance of Boos Texter and a number of other text-categorization algorithms on a variety of tasks.
Data Mining And Knowledge Discovery Handbook
This book first surveys, then provides comprehensive yet concise algorithmic descriptions of methods, including classic methods plus the extensions and novel methods developed recently.
Random klabelsets for multilabel classification
  • IEEE Transactions on Knowledge and Data Engineering,
  • 2011