# Scalable Source Selection for Machine Learning Tasks

DevFeed: [Scalable Source Selection for Machine Learning Tasks](<https://devfeed.tech/articles/scalable-source-selection-for-machine-learning-tasks-118101.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2507.22186>)

Author: Ambarish Singh, Romila Pradhan

Published: 2026-11-05T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Databases](<https://devfeed.tech/sources/arxiv-databases.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Model Development](<https://devfeed.tech/topics/model-development.md>)

Tags: [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data-lake](<https://devfeed.tech/tags/data-lake.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [improvement](<https://devfeed.tech/tags/improvement.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [meta](<https://devfeed.tech/tags/meta.md>), [regression](<https://devfeed.tech/tags/regression.md>), [selection](<https://devfeed.tech/tags/selection.md>), [sources](<https://devfeed.tech/tags/sources.md>), [utility](<https://devfeed.tech/tags/utility.md>)

## AI overview

Selecting every table found through schema similarity and joinability can add noise or shift distributions, so the paper frames source selection as choosing a subset that maximizes downstream task utility. Splice is target- and cost-aware and seeks sources that add signals missing from the combined sources. Its faster variants estimate marginal utility using data profiles with meta-learning (Splice-DP) or gradient alignment (Splice-Grad). On two real-world classification benchmarks, Splice improved binary F1 by 3-5 percentage points over using all discovered tables. Across five additional classification and regression tasks, it improved utility by 1.8-13.7 points; the surrogate variants matched that quality with 3-7 times fewer model evaluations, and up to 77 times fewer at scale.

## Source excerpt

arXiv:2507.22186v3 Announce Type: replace-cross Abstract: Given a query table, existing data lake discovery techniques seek to identify data sources related to the query table through schema similarity and joinability. The discovered sources, however, are not always useful for downstream analyses; while some improve task generalization, some have the potential to inadvertently introduce harmful noise or distribution shifts. We formalize source selection as the problem of finding a subset of the discovered sources that maximizes downstream task utility. We propose Splice, a target- and cost-aware source selection system that adopts the idea of gene splicing to identify sources which provide additional signals that the union of all sources misses. To speed up the process, we develop two surrogate variants, Splice-DP and Splice-Grad, that estimate the marginal contribution of subsets to task utility by leveraging data profiles in a meta-learning approach and gradient alignment, respectively. On classification tasks over two real-world benchmarks derived from SOTA table union search tools, Splice shows 3-5 percentage points improvement in binary F1 scores compared to the case when all discovered tables are considered. On five additional real-world ML tasks spanning classification and regression, Splice improves task utility over the all-sources baseline by 1.8-13.7 points, while Splice-DP and Splice-Grad match this quality with 3-7x fewer model evaluations (up to 77x at scale).