A Fast and Efficient Algorithm for Filtering the Training Dataset
Pages 504 - 512
Abstract
The goal of this paper is to present a new algorithm that filters out inconsistent instances from the training dataset for further usage with machine learning algorithms or learning of neural networks. The idea of this algorithm is based on the previous state-of-the-art algorithm, which uses the concept of local sets. Sophisticated modification of the definition of local sets changes the merits of the algorithm. It is additionally supported by locality-sensitive hashing used for searching for nearest neighbors, composing a new efficient (), and an accurate algorithm.
Results prepared on many benchmarks show that the algorithm is as accurate as previous but strongly reduces the time complexity.
References
[1]
Arnaiz-González, A., Díez-Pastor, J.-F., Rodríguez, J.J., García-Osorio, C.: Instance selection of linear complexity for big data: Knowl.-Based Syst. 107, 83–95 (2016)
[2]
Bawa, M., Condie, T., Ganesan, P.: LSH forest: self-tuning indexes for similarity search. In: Proceedings of the 14th International Conference on World Wide Web, pp. 651–660. Chiba, Japan (2005)
[3]
Brighton H and Mellish C Advances in instance selection for instance-based learning algorithms Data Min. Knowl. Disc. 2002 6 2 153-172
[4]
Cover TM and Hart PE Nearest neighbor pattern classification Instit. Electr. Electron. Eng. Trans. Inf. Theory 1967 13 1 21-27
[5]
Garcia S, Derrac J, Cano J, and Herrera F Prototype selection for nearest neighbor classification: taxonomy and empirical study IEEE Trans. Pattern Anal. Mach. Intell. 2012 34 3 417-435
[6]
Har-Peled S, Indyk P, and Motwani R Approximate nearest neighbor: towards removing the curse of dimensionality Theory Comput. 2012 8 321-350
[7]
Indyk, P., Motwani, R.: Approximate nearest neighbor—towards removing the curse of dimensionality. In: The Thirtieth Annual ACM Symposium on Theory of Computing, pp. 604–613 (1998)
[8]
Grochowski M and Jankowski N Rutkowski L, Siekmann JH, Tadeusiewicz R, and Zadeh LA Comparison of instance selection algorithms II. results and comments Artificial Intelligence and Soft Computing - ICAISC 2004 2004 Heidelberg Springer 580-585
[9]
Jankowski, N., Orliński, M.: Fast encoding length-based prototype selection algorithms. Australian J. Intell. Inf. Process. Syst. 16(3), 59–66 (2019). Special Issue: Neural Information Processing 26th International Conference on Neural Information Processing. http://ajiips.com.au/iconip2019/docs/ajiips/v16n3.pdf
[10]
Leyva E, González A, and Pérez R Three new instance selection methods based on local sets: a comparative study with several approaches from a bi-objective perspective Pattern Recogn. 2015 48 4 1523-1537
[11]
Merz, C.J., Murphy, P.M.: UCI repository of machine learning databases (1998). http://www.ics.uci.edu/~mlearn/MLRepository.html
[12]
Olvera-López JA, Carrasco-Ochoa JA, and Martínez-Trinidad JF A new fast prototype selection method based on clustering Pattern Anal. Appl. 2009 13 2 131-141
[13]
Orliński M and Jankowski N Fast t-SNE algorithm with forest of balanced LSH trees and hybrid computation of repulsive forces Knowl.-Based Syst. 2020 206 1-16
[14]
Orliński, M., Jankowski, N.: instance selection algorithms–RR-DROPs. In: IEEE World Congress on Computational Intelligence, pp. 1–8. IEEE Press (2020). http://www.is.umk.pl/~norbert/publications/20-FastDROP.pdf
[15]
Wilson D Asymptotic properties of nearest neighbor rules using edited data IEEE Trans. Syst. Man Cybern. 1972 2 3 408-421
Recommendations
DCPE co-training for classification
Co-training is a well-known semi-supervised learning technique that applies two basic learners to train the data source, which uses the most confident unlabeled data to augment labeled data in the learning process. In the paper, we use the diversity of ...
Editing training data for multi-label classification with the k-nearest neighbor rule
Multi-label classification allows instances to belong to several classes at once. It has received significant attention in machine learning and has found many real-world applications in recent years, such as text categorization, automatic video ...
Tri-Training: Exploiting Unlabeled Data Using Three Classifiers
In many practical data mining applications, such as Web page classification, unlabeled training examples are readily available, but labeled ones are fairly expensive to obtain. Therefore, semi-supervised learning algorithms such as co-training have ...
Comments
Please enable JavaScript to view thecomments powered by Disqus.Information & Contributors
Information
Published In
Nov 2022
659 pages
ISBN:978-3-031-30104-9
DOI:10.1007/978-3-031-30105-6
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.
Publisher
Springer-Verlag
Berlin, Heidelberg
Publication History
Published: 13 April 2023
Author Tags
Qualifiers
- Article
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 0Total Downloads
- Downloads (Last 12 months)0
- Downloads (Last 6 weeks)0
Reflects downloads up to 21 Dec 2024