Newest papers

Mastering Model Fit: Applications of the InterModel Vigorish for Interpretable Machine Learning

Jiani Yan, Pengyun Wang, Klint Kanopka, Ben Domingue, Charles Rahal · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.29960 · cited by 0 works (OpenAlex) · open-access PDF

Korean

정확성만으로는 어떤 기능이나 모델 구성 요소가 확률적 적합성을 향상시키는지, 어떤 결과가 혜택을 얻는지, 또는 훈련 중에 변화가 어떻게 적합한지에 대한 정보가 거의 없습니다. 우리는 InterModel Vigorish (IMV)를 기계 인텔리전스 공간에 소개하여 참조 모델에 비해 예상되는 베팅 이점으로 적합성의 차이를 표현합니다. SHAP-IMV는 Shapley 가중계 하위 하위 세트 비교를 통해 특성에 이득을 부여합니다. Multi-IMV는 one-vs-rest 및 pairwise 클래스 차별을 평가하고 Ablate-IMV는 아브레이션을 통해 건축적 구성 요소와 기능 그룹을 평가합니다. 우리는 테이블 데이터, 이미지, 센서 시퀀스, 그래

Machine translation (Meta M2M100); the English original follows.

Abstract

Accuracy alone gives little information about which features or model components improve probabilistic fit, which outcomes benefit, or how fit changes during training. We introduce the InterModel Vigorish (IMV) into the Machine Intelligence space in order to express differences in fit as an expected betting advantage over a reference model. SHAP-IMV attributes gains to features through Shapley-weighted subset comparisons, Multi-IMV evaluates one-vs-rest and pairwise class discrimination, and Ablate-IMV assesses architectural components and feature groups by ablation. We evaluate these applications over ten random seeds using tabular data, images, sensor sequences, and graphs. Feature rankings vary across estimators. High one-vs-rest accuracy can accompany negligible improvement over a null, while ablation can preserve or improve fit under a fixed training budget. On six graph-classification benchmarks, we time-profile IMV over 200 training epochs, reporting seed means and ranges. The open-source imvpy library and accompanying notebooks provide code to replicate our work and to use the IMV more widely.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Machine learning, Artificial intelligence, Computer science, Pairwise comparison, Feature (linguistics), Replicate

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange