finding

active

finding:linear-regression-of-ood-iia-on-training-emd-yields-coefficient-0-3424-r-2-0-729-f-1-28-75-28-p-001

Linear regression of OOD IIA on training EMD yields coefficient -0.3424, R^2=0.729, F(1,28)=75.28, p<.001

Statistical evidence that training divergence (EMD) predicts lower OOD intervention performance

Source paper

extracted_from

Addressing divergent representations from causal interventions on neural networks

(2025) · Satchel Grant · Simon Jerome Han · Alexa R. Tartaglini · Christopher Potts

Neighborhood — ranked by edge-count

Claims (1)

claim

Representational divergence (as measured by EMD) can predict lower out-of-distribution intervention performance
associated_withsupports
Practical utility of reducing divergence demonstrated through regression analysis

Hypotheses (1)

hypothesis

We hypothesized that divergence could influence IIA when transferring the DAS alignment to OOD settings
associated_withsupports
Motivating hypothesis for the OOD experiment testing practical utility of divergence reduction

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

DAS behavioral loss produces EMD along feature dimensions of 0.032±0.003 on synthetic 10-class datasetfinding0.745
Quantitative baseline for divergence using behavioral DAS loss on synthetic dataset
DB-MTL training losses decrease smoothly and gradient norms are lower than EW on NYUv2, indicating training stability.finding0.742
Training stability analysis.
Modified CL loss produces EMD along feature dimensions of 0.007±0.001 on synthetic 10-class datasetfinding0.739
Quantitative improvement in divergence reduction using the modified CL loss on synthetic dataset
DAS behavioral loss achieves IIA of 0.997±0.001 on synthetic 10-class dataset training/test setsfinding0.737
IIA baseline for DAS behavioral loss on synthetic dataset
For small CL loss weights epsilon, IIA is maintained (potentially improved) while EMD decreases in Boundless DAS on a 7B LLMfinding0.730
Empirical result showing the CL loss can reduce divergence without sacrificing interpretability accuracy
Linear combination ρd – dr – log k yields a predictive correlate of success.hypothesis0.729
Implicit hypothesis behind S form.
NLA explanations grow more informative over training with FVE increasing from 0.3-0.4 to 0.6-0.8 roughly linearly in log(training steps)finding0.721
Quantitative evidence that NLA training produces increasingly informative explanations despite optimizing only for reconstruction.
If EI maximization is used as a regularization in representation learning, then OOD generalization will improve beyond current invariant risk minimization methods.hypothesis0.717
Proposed conjecture in §4.3.1.