# Transformation Importance with Applications to Cosmology

arXiv: https://arxiv.org/abs/2003.01926 · HTML: https://mlsanity.com/paper/2003.01926/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 4 Mar 2020.

## Built on by (most cited first, 8 of 8)

- [Beyond Explaining: Opportunities and Challenges of XAI-Based Model Improvement](https://mlsanity.com/paper/2203.08008/) (2022)
- [Augmenting Interpretable Models with LLMs during Training](https://mlsanity.com/paper/2209.11799/) (2022)
- [Adaptive wavelet distillation from neural networks through interpretations](https://mlsanity.com/paper/2107.09145/) (2021)
- [Does Your Model Think Like an Engineer? Explainable AI for Bearing Fault Detection with Deep Learning](https://mlsanity.com/paper/2310.12967/) (2023)
- [Machine Learning and Cosmology](https://mlsanity.com/paper/2203.08056/) (2022)
- [Matched sample selection with GANs for mitigating attribute confounding](https://mlsanity.com/paper/2103.13455/) (2021)
- [Interpreting and improving deep-learning models with reality checks](https://mlsanity.com/paper/2108.06847/) (2021)
- [Subgroup Discovery in Unstructured Data](https://mlsanity.com/paper/2207.07781/) (2022)

## Abstract (arXiv)

Machine learning lies at the heart of new possibilities for scientific discovery, knowledge generation, and artificial intelligence. Its potential benefits to these fields requires going beyond predictive accuracy and focusing on interpretability. In particular, many scientific problems require interpretations in a domain-specific interpretable feature space (e.g. the frequency domain) whereas attributions to the raw features (e.g. the pixel space) may be unintelligible or even misleading. To address this challenge, we propose TRIM (TRansformation IMportance), a novel approach which attributes importances to features in a transformed space and can be applied post-hoc to a fully trained model. TRIM is motivated by a cosmological parameter estimation problem using deep neural networks (DNNs) on simulated data, but it is generally applicable across domains/models and can be combined with any local interpretation method. In our cosmology example, combining TRIM with contextual decomposition shows promising results for identifying which frequencies a DNN uses, helping cosmologists to understand and validate that the model learns appropriate physical features rather than simulation artifacts.