Are encoders able to learn landmarkers for warm-starting of Hyperparameter Optimization?
This work addresses a specific challenge in meta-learning for hyperparameter optimization, but it is incremental as it builds on existing representation methods without achieving major improvements.
The paper tackled the problem of representing heterogeneous tabular datasets for meta-learning in hyperparameter optimization warm-starting by proposing two novel encoders designed to capture landmarker properties, but found that while these encoders learned representations aligned with landmarkers, they did not lead to significant performance gains in the meta-task.
Effectively representing heterogeneous tabular datasets for meta-learning purposes is still an open problem. Previous approaches rely on representations that are intended to be universal. This paper proposes two novel methods for tabular representation learning tailored to a specific meta-task - warm-starting Bayesian Hyperparameter Optimization. Both follow the specific requirement formulated by ourselves that enforces representations to capture the properties of landmarkers. The first approach involves deep metric learning, while the second one is based on landmarkers reconstruction. We evaluate the proposed encoders in two ways. Next to the gain in the target meta-task, we also use the degree of fulfillment of the proposed requirement as the evaluation metric. Experiments demonstrate that while the proposed encoders can effectively learn representations aligned with landmarkers, they may not directly translate to significant performance gains in the meta-task of HPO warm-starting.