Dissertation > Excellent graduate degree dissertation topics show

Multiple Kernel Learning Improved by Bi-objective Functions and Its Application to Semi-supervised Learning and Transfer Learning

Author: Liang
Tutor: RenJiangTao
School: Sun Yat-sen University
Course: Software Engineering
Keywords: Kernel Learning Semi-supervised learning Transfer learning Figure Laplace Maximum mean difference
CLC: TP181
Type: Master's thesis
Year: 2011
Downloads: 67
Quote: 0
Read: Download Dissertation

Abstract


Nuclear technology is a data mining in recent years , the field of machine learning is an important research direction , its main feature is to build the kernel function of the similarity between the sample feature space constructed sample . In a recent study of nuclear technology , nuclear learn the latest nuclear technology as a concern , and different from the traditional nuclear technology , nuclear learn through the maximum interval or label registration measure , dynamic learning an optimal kernel function , which Get a better feature space . This dynamic nuclear learning constructed classifier can obtain a better performance than conventional nuclear technology . However, the majority of nuclear learn there are some obvious shortcomings . First, the efficiency of the nuclear learning algorithm are insufficient, and the size of the input data is the efficiency of the algorithm is limited ; Secondly, many of the learning algorithm assumes scene is limited to a supervised learning scene . Semi-supervised learning and transfer learning is to study how learning classifier training set is restricted case , the performance of the general nuclear learning in the scene of the two learning deficiencies , which does not consider learning because of the traditional nuclear technology to the particularity of the above scenario . In order to solve the above problems , the latest research on the basis of the current nuclear Learning , through the expansion of high - efficiency multi-core learning algorithm to construct multi-objective function of nuclear learning framework . In the new framework , the general multicore learning algorithm is capable of binding to a specific objective function to improve the performance of the algorithm in a particular scene . Semi-supervised learning scenarios and transfer learning scenarios , the proposed algorithm specific the the improved multicore learning algorithm based on graph Laplace and multicore learning algorithm based on the maximum mean difference in improvement . Characteristics for multiple objective function optimization problems , how to select a good balance between the objective function . In this experiment performed on the data collected from the real world , the experimental results show that the method after the multicore learning Figure Laplace improvements and the largest mean difference in improvement than general nuclear technology in better performance in the two scenarios , and has comparable with some semi- supervised learning algorithms and transfer learning algorithms .

Related Dissertations

  1. High-performed Kernel Classification Methods Based on Multi-kernel Learning,TP391.41
  2. Semi-supervised Learning and Active Learning of Sentiment Classification Coupled with Domain Knowledge,TP181
  3. Content-based image classification technology goods,TP391.41
  4. Heterogeneous Transfer Learning between Image and Text Data,TP391.41
  5. Comparison and Improvement of Two Methods Based on Semi-Supervised Learning,TP18
  6. The Design and Implementation of an Image Segmentation System Based on Semi-Supervised Learning,TP391.41
  7. Machine Learning Algorithm-based Metaphor Recognition,TP181
  8. The Research on Marine Literature Categorization and Labeling Minimization,TP391.1
  9. Research on Two Clustering Algorithms Based on Semi-Supervised Learning,TP181
  10. Semi-Supervised Learning with Application in MR Image Segmentation,TP391.41
  11. Oversight and semi-supervised canonical correlation analysis and its application,TP181
  12. The Research on Semi-supervised Collaboration-training Algorithm,TP18
  13. Semi-Supervised Manifold Learning Theory and Applications,TP181
  14. Research of Search Engine Web Spam Detection,TP391.3
  15. Research on Technology of Spam Filtering Based on Markov Logic Network,TP393.098
  16. Studies on the Algorithms of the Semi-supervised Classification and Techniques,TP18
  17. Study of Real-Time Acoustic Detection Based on Semi-Supervised Learning,TN912.3
  18. Research on Semi-Supervised Learning and Its Application,TP181
  19. Embedded Space Based Dimensionality Reduction Modeling Research and Its Applications,TP311.13
  20. Based semi - supervised learning Chinese Question Classification,TP391.1

CLC: > Industrial Technology > Automation technology,computer technology > Automated basic theory > Artificial intelligence theory > Automated reasoning,machine learning
© 2012 www.DissertationTopic.Net  Mobile