| Presentation | 2019-12-19 Mathematical Representation of Emotion by Combining Recognition and Unification Tasks Using Multimodal Deep Neural Networks Seiichi Harata, Takuto Sakuma, Shohei Kato, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | To emulate human emotions in robots, the mathematical representation of emotion is important for all components of affective computing such as emotion recognition, generation, and expression. There are several methods to represent emotions by vectors of continuous values and mapping them from uni-modality data to low-dimensional space. However, the representation of emotions obtained by uni-modality data seems to depend on such modality. In this study, we proposed integrating multi-modalities on a DNN acquiring mathematical representation (emotional space) of emotion. We aim at the acquisition of emotional space which does not depend on modalities by combining recognition task and unification task. Experiments with audio-visual data have confirmed two things. First, there are differences in the emotional space acquired from a single modality. Second, the proposed method can acquire a modality independent emotional space. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Affective Computing / Deep Neural Networks / Multi-modal / Multi-task Learning / Metric Learning / Emotional Space |
| Paper # | HIP2019-65 |
| Date of Issue | 2019-12-12 (HIP) |
| Conference Information | |
| Committee | HIP |
|---|---|
| Conference Date | 2019/12/19(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | RIEC, Tohoku University |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Miyuki Kamachi(Kogakuin Univ.) |
| Vice Chair | Shuichi Sakamoto(Tohoku Univ.) / Yuji Wada(Ritsumeikan Univ.) |
| Secretary | Shuichi Sakamoto(NICT) / Yuji Wada(NTT) |
| Assistant | Atsushi Wada(NICT) / Hidetoshi Kanaya(Ritsumeikan Univ.) / Yuki Yamada(Kyushu Univ.) |
| Paper Information | |
| Registration To | Technical Committee on Human Information Processing |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Mathematical Representation of Emotion by Combining Recognition and Unification Tasks Using Multimodal Deep Neural Networks |
| Sub Title (in English) | |
| Keyword(1) | Affective Computing |
| Keyword(2) | Deep Neural Networks |
| Keyword(3) | Multi-modal |
| Keyword(4) | Multi-task Learning |
| Keyword(5) | Metric Learning |
| Keyword(6) | Emotional Space |
| 1st Author's Name | Seiichi Harata |
| 1st Author's Affiliation | Nagoya Institute of Technology(NITech) |
| 2nd Author's Name | Takuto Sakuma |
| 2nd Author's Affiliation | Nagoya Institute of Technology(NITech) |
| 3rd Author's Name | Shohei Kato |
| 3rd Author's Affiliation | Nagoya Institute of Technology(NITech) |
| Date | 2019-12-19 |
| Paper # | HIP2019-65 |
| Volume (vol) | vol.119 |
| Number (no) | HIP-348 |
| Page | pp.pp.1-6(HIP), |
| #Pages | 6 |
| Date of Issue | 2019-12-12 (HIP) |