Multi-Task Extension of Geometrically Aligned Transfer Encoder
Abstract
Molecular datasets often suffer from a lack of data. It is well-known that gathering data is difficult due to the complexity of experimentation or simulation involved. Here, we leverage mutual information across different tasks in molecular data to address this issue. We extend an algorithm that utilizes the geometric characteristics of the encoding space, known as the Geometrically Aligned Transfer Encoder (GATE), to a multi-task setup. Thus, we connect multiple molecular tasks by aligning the curved coordinates onto locally flat coordinates, ensuring the flow of information from source tasks to target data to support the performance.
Cite
Text
Ko et al. "Multi-Task Extension of Geometrically Aligned Transfer Encoder." ICML 2024 Workshops: AI4Science, 2024.Markdown
[Ko et al. "Multi-Task Extension of Geometrically Aligned Transfer Encoder." ICML 2024 Workshops: AI4Science, 2024.](https://mlanthology.org/icmlw/2024/ko2024icmlw-multitask/)BibTeX
@inproceedings{ko2024icmlw-multitask,
title = {{Multi-Task Extension of Geometrically Aligned Transfer Encoder}},
author = {Ko, Sung Moon and Lee, Sumin and Jeong, Dae-Woong and Kim, Hyunseung and Lee, Chanhui and Yim, Soorin and Han, Sehui},
booktitle = {ICML 2024 Workshops: AI4Science},
year = {2024},
url = {https://mlanthology.org/icmlw/2024/ko2024icmlw-multitask/}
}