| Presentation | 2023-03-03 Forest Construction of Gaussian and Discrete Variables based on WBIC Ashraful Islam, Joe Suzuki, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Mutual information is a metric that determines the association between two random variables by measuring the amount of information one variable holds about the other. It quantifies the dependency between them. A higher mutual information value indicates a stronger relationship between the random variables. It is connected to entropy and finds application in various fields, such as information theory, statistics, and machine learning. To determine the mutual information between discrete variables, we need to use estimated joint probabilities based on samples of each categorized variable. Nevertheless, the traditional method could be more efficient in figuring out the mutual information of a mix of discrete and continuous random variables because it needs to determine the conditional probabilities for discrete variables given continuous variables. We propose a new MI method that can handle both discrete and continuous random variables at the same time. Our approach calculates the free energy of mutual information using the Watanabe Bayesian information criterion (WBIC), enabling the estimation of MI between discrete and continuous variables. This method can handle mixed variables more skillfully by overcoming the shortcomings of conventional mutual information estimation techniques. When incorporated into the Chow-Liu algorithm, the new MI estimator can create a forest rather than a spanning tree. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Mutual Information / Chow-Liu Algorithm / Forest / Free Energy / WBIC |
| Paper # | PRMU2022-126,IBISML2022-133 |
| Date of Issue | 2023-02-23 (PRMU, IBISML) |
| Conference Information | |
| Committee | PRMU / IBISML / IPSJ-CVIM |
|---|---|
| Conference Date | 2023/3/2(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Future University Hakodate |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Seiichi Uchida(Kyushu Univ.) / Masashi Sugiyama(Univ. of Tokyo) |
| Vice Chair | Takuya Funatomi(NAIST) / Mitsuru Anpai(Denso IT Lab.) / Toshihiro Kamishima(AIST) / Koji Tsuda(Univ. of Tokyo) |
| Secretary | Takuya Funatomi(CyberAgent) / Mitsuru Anpai(Univ. of Tokyo) / Toshihiro Kamishima(NTT) / Koji Tsuda(Hokkaido Univ.) |
| Assistant | Nakamasa Inoue(Tokyo Inst. of Tech.) / Yasutomo Kawanishi(Riken) / Yoshinobu Kawahara(Osaka Univ.) / Taiji Suzuki(Tokyo Inst. of Tech.) |
| Paper Information | |
| Registration To | Technical Committee on Pattern Recognition and Media Understanding / Technical Committee on Information-Based Induction Sciences and Machine Learning / Special Interest Group on Computer Vision and Image Media |
|---|---|
| Language | ENG |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Forest Construction of Gaussian and Discrete Variables based on WBIC |
| Sub Title (in English) | |
| Keyword(1) | Mutual Information |
| Keyword(2) | Chow-Liu Algorithm |
| Keyword(3) | Forest |
| Keyword(4) | Free Energy |
| Keyword(5) | WBIC |
| 1st Author's Name | Ashraful Islam |
| 1st Author's Affiliation | Osaka University(Osaka Univ.) |
| 2nd Author's Name | Joe Suzuki |
| 2nd Author's Affiliation | Osaka University(Osaka Univ.) |
| Date | 2023-03-03 |
| Paper # | PRMU2022-126,IBISML2022-133 |
| Volume (vol) | vol.122 |
| Number (no) | PRMU-404,IBISML-405 |
| Page | pp.pp.371-377(PRMU), pp.371-377(IBISML), |
| #Pages | 7 |
| Date of Issue | 2023-02-23 (PRMU, IBISML) |