New data-driven approaches to improve probabilistic model structure learning
To learn the network structures used in probabilistic models (e.g., Bayesian network), many researchers proposed structure learning algorithms to extract the network structure from data. However, structure learning is a challenging problem due to the extremely large number of possible structure cand...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
2019
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/84123 http://hdl.handle.net/10220/50443 |
| _version_ | 1811695367300841472 |
|---|---|
| author | Zhao, Jianjun |
| author2 | Pan Jialin, Sinno |
| author_facet | Pan Jialin, Sinno Zhao, Jianjun |
| author_sort | Zhao, Jianjun |
| collection | NTU |
| description | To learn the network structures used in probabilistic models (e.g., Bayesian network), many researchers proposed structure learning algorithms to extract the network structure from data. However, structure learning is a challenging problem due to the extremely large number of possible structure candidates. One challenge relates to structure learning in Bayesian network is the conflicts among local structures obtained from the local structure learning algorithms. This is the so-called symmetry correction problem. Another challenge is the V-structure selection problem, which is related to the determination of edge orientation in Bayesian network.
In this thesis, we investigate the above two challenges in structure learning and propose novel data-driven approaches to overcome these challenges when building a Bayesian network. First, two new data-driven symmetry correction methods are developed to learn an undirected graph of Bayesian network. The proposed methods outperform the existing heuristic rule. Second, a weighted maximum satisfiability (MAX-SAT) problem is formulated to solve the V-structures selection problem. The weights are learned from data to quantify the strength of the V-structures. Our proposed solution outperforms existing methods.
Besides, we investigate how transfer learning can be used for structure learning with limited training examples and a source structure. In particular, we propose a transfer learning approach to learn the structure of a Sum-Product Network (SPN) which can be converted to a Bayesian network under certain conditions. Our novel approach allows one to construct the target SPN with limited training examples, given an existing source SPN from a similar domain. |
| first_indexed | 2024-10-01T07:22:20Z |
| format | Thesis |
| id | ntu-10356/84123 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T07:22:20Z |
| publishDate | 2019 |
| record_format | dspace |
| spelling | ntu-10356/841232020-10-28T08:40:48Z New data-driven approaches to improve probabilistic model structure learning Zhao, Jianjun Pan Jialin, Sinno School of Computer Science and Engineering A*STAR (SINGA) Centre for Computational Intelligence Science::Mathematics::Statistics Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence To learn the network structures used in probabilistic models (e.g., Bayesian network), many researchers proposed structure learning algorithms to extract the network structure from data. However, structure learning is a challenging problem due to the extremely large number of possible structure candidates. One challenge relates to structure learning in Bayesian network is the conflicts among local structures obtained from the local structure learning algorithms. This is the so-called symmetry correction problem. Another challenge is the V-structure selection problem, which is related to the determination of edge orientation in Bayesian network. In this thesis, we investigate the above two challenges in structure learning and propose novel data-driven approaches to overcome these challenges when building a Bayesian network. First, two new data-driven symmetry correction methods are developed to learn an undirected graph of Bayesian network. The proposed methods outperform the existing heuristic rule. Second, a weighted maximum satisfiability (MAX-SAT) problem is formulated to solve the V-structures selection problem. The weights are learned from data to quantify the strength of the V-structures. Our proposed solution outperforms existing methods. Besides, we investigate how transfer learning can be used for structure learning with limited training examples and a source structure. In particular, we propose a transfer learning approach to learn the structure of a Sum-Product Network (SPN) which can be converted to a Bayesian network under certain conditions. Our novel approach allows one to construct the target SPN with limited training examples, given an existing source SPN from a similar domain. Doctor of Philosophy 2019-11-19T12:07:54Z 2019-12-06T15:38:48Z 2019-11-19T12:07:54Z 2019-12-06T15:38:48Z 2019 Thesis Zhao, J. (2019). New data-driven approaches to improve probabilistic model structure learning. Doctoral thesis, Nanyang Technological University, Singapore. https://hdl.handle.net/10356/84123 http://hdl.handle.net/10220/50443 10.32657/10356/84123 en 125 p. application/pdf |
| spellingShingle | Science::Mathematics::Statistics Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence Zhao, Jianjun New data-driven approaches to improve probabilistic model structure learning |
| title | New data-driven approaches to improve probabilistic model structure learning |
| title_full | New data-driven approaches to improve probabilistic model structure learning |
| title_fullStr | New data-driven approaches to improve probabilistic model structure learning |
| title_full_unstemmed | New data-driven approaches to improve probabilistic model structure learning |
| title_short | New data-driven approaches to improve probabilistic model structure learning |
| title_sort | new data driven approaches to improve probabilistic model structure learning |
| topic | Science::Mathematics::Statistics Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence |
| url | https://hdl.handle.net/10356/84123 http://hdl.handle.net/10220/50443 |
| work_keys_str_mv | AT zhaojianjun newdatadrivenapproachestoimproveprobabilisticmodelstructurelearning |