Task Addition in Multi-Task Learning by Geometrical Alignment
This work addresses computational bottlenecks in multi-task learning for molecular property prediction, offering an incremental improvement for researchers in computational chemistry and drug discovery.
The paper tackles the challenge of scaling multi-task learning for molecular property prediction by proposing a task addition approach for the Geometrically Aligned Transfer Encoder (GATE) method, which improves performance on target tasks with limited data while maintaining computational efficiency, as demonstrated by superior results over conventional methods.
Training deep learning models on limited data while maintaining generalization is one of the fundamental challenges in molecular property prediction. One effective solution is transferring knowledge extracted from abundant datasets to those with scarce data. Recently, a novel algorithm called Geometrically Aligned Transfer Encoder (GATE) has been introduced, which uses soft parameter sharing by aligning the geometrical shapes of task-specific latent spaces. However, GATE faces limitations in scaling to multiple tasks due to computational costs. In this study, we propose a task addition approach for GATE to improve performance on target tasks with limited data while minimizing computational complexity. It is achieved through supervised multi-task pre-training on a large dataset, followed by the addition and training of task-specific modules for each target task. Our experiments demonstrate the superior performance of the task addition strategy for GATE over conventional multi-task methods, with comparable computational costs.