| Presentation | 2018-02-19 De facto bankruptcy prediction by deep learning Tadaaki Hosaka, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In this research, deep learning is applied to the prediction of corporate de facto bankruptcy. We collect financial statements over four fiscal years for 102 companies delisted from Japanese stock markets due to de facto bankruptcy and 2062 continuing companies. In addition, the number of samples is increased by interpolating and extrapolating the financial statements of arbitrary two fiscal years. The key point in our method is to transform a set of financial ratios calculated from the financial statements of one sample into a gray-scale image, and treat it as one image sample. Eventually, we generate 7520 samples for each class and use them as learning data of a convolutional neural network based on GoogLeNet. Obtained network indicates higher accuracy in bankruptcy prediction than the traditional methods using CART, linear discriminant analysis, SVM, and AdaBoost. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | de facto bankruptcy / financial ratio / imaging / convolutional neural network |
| Paper # | PRMU2017-156,CNR2017-34 |
| Date of Issue | 2018-02-12 (PRMU, CNR) |
| Conference Information | |
| Committee | PRMU / CNR |
|---|---|
| Conference Date | 2018/2/19(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Shinichi Sato(NII) / Tetsuo Ono(Hokkaido Univ.) |
| Vice Chair | Hironobu Fujiyoshi(Chubu Univ.) / Yoshihisa Ijiri(Omron) / Masayuki Kanbara(NAIST) / Kazunori Takashio(Keio Univ.) |
| Secretary | Hironobu Fujiyoshi(AIST) / Yoshihisa Ijiri(NAIST) / Masayuki Kanbara(Hokkaido Univ.) / Kazunori Takashio(Panasonic) |
| Assistant | Masato Ishii(NEC) / Yusuke Sugano(Osaka Univ.) / Wataru Mito(SECOM) / Yuka Kobayashi(Toshiba) / Tatsuya Ishihara(NTT) |
| Paper Information | |
| Registration To | Technical Committee on Pattern Recognition and Media Understanding / Technical Committee on Cloud Network Robotics |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | De facto bankruptcy prediction by deep learning |
| Sub Title (in English) | |
| Keyword(1) | de facto bankruptcy |
| Keyword(2) | financial ratio |
| Keyword(3) | imaging |
| Keyword(4) | convolutional neural network |
| 1st Author's Name | Tadaaki Hosaka |
| 1st Author's Affiliation | Tokyo University of Science(Tokyo Univ. Sci.) |
| Date | 2018-02-19 |
| Paper # | PRMU2017-156,CNR2017-34 |
| Volume (vol) | vol.117 |
| Number (no) | PRMU-442,CNR-443 |
| Page | pp.pp.65-70(PRMU), pp.65-70(CNR), |
| #Pages | 6 |
| Date of Issue | 2018-02-12 (PRMU, CNR) |