Disentangled Feature Importance
Researchers have introduced two new methods for feature attribution and clustering analysis: Disentangled Feature Importance (DFI) and Cluster LOCO.
DFI is a population-level attribution framework that maps covariates to an independent latent representation and computes latent importance[1]. It attributes predictive signals across correlated measurement channels, addressing challenges in interpreting complex data. Cluster LOCO, on the other hand, is a family of model-agnostic feature importance scores for clustering[2]. Clustering is widely used for exploratory analysis and scientific discovery, but existing feature importance scores are often tied to specific algorithms and data assumptions. Cluster LOCO is built on feature occlusion and clustering generalizability, with two variants: Cluster LOCO-Split, which relies on data splitting, and Cluster LOCO-MP, an ensemble-based version for large-scale data. Researchers claim that Cluster LOCO more reliably recovers informative features than existing methods[2]. The DFI paper was first submitted on June 30, 2025, and last revised on June 6, 2026[1]. The Cluster LOCO paper was submitted on June 12, 2026[2].
research-paperinfrastructuretool-releasecommentary
Background sources we checked (4)
- arxiv.org ↗ When predictors are statistically dependent, the appropriate definition of feature importance depends on the operational goal. Conditional-incremental measures are well-suited for feature selection, acquisition, and compression, where shared predictive information is treated as r…
- en.wikipedia.org ↗ Machine learning (ML) is a field of study in artificial intelligence concerned with the development and study of statistical algorithms that can learn from data and generalize to unseen data, and thus perform tasks without being explicitly programmed. Advances in the field of de…
- en.wikipedia.org ↗ Present-day climate change includes both global warming—the ongoing increase in global average temperature—and its wider effects on Earth's climate system. Climate change in a broader sense also includes previous long-term changes to Earth's climate. The modern-day rise in global…
- en.wikipedia.org ↗ Artificial neural networks (ANNs) are models created using machine learning to perform a number of tasks. While the computational implementations of ANNs relate to earlier discoveries in mathematics, their creation was inspired by biological neural circuitry. The first implementa…