Collaborative Prediction: To Join or To Disjoin Datasets
This work addresses the need for high-quality dataset selection in machine learning, particularly for improving model performance through effective data merging, though it appears incremental as it builds on existing prediction models and algorithms.
The paper tackles the problem of selecting and merging datasets from different sources to minimize the population loss of prediction models, proposing a practical algorithm with theoretical guarantees that reduces loss with high probability, as demonstrated in numerical experiments on linear regression and broader applications.
With the recent rise of generative Artificial Intelligence (AI), the need of selecting high-quality dataset to improve machine learning models has garnered increasing attention. However, some part of this topic remains underexplored, even for simple prediction models. In this work, we study the problem of developing practical algorithms that select appropriate dataset to minimize population loss of our prediction model with high probability. Broadly speaking, we investigate when datasets from different sources can be effectively merged to enhance the predictive model's performance, and propose a practical algorithm with theoretical guarantees. By leveraging an oracle inequality and data-driven estimators, the algorithm reduces population loss with high probability. Numerical experiments demonstrate its effectiveness in both standard linear regression and broader machine learning applications. Code is available at https://github.com/kkrokii/collaborative_prediction.