{"id":28749104,"url":"https://github.com/acerbilab/normalizing-flow-regression","last_synced_at":"2026-02-21T22:32:12.521Z","repository":{"id":287841552,"uuid":"961806449","full_name":"acerbilab/normalizing-flow-regression","owner":"acerbilab","description":"Normalizing Flow Regression for Bayesian Inference with Offline Likelihood Evaluations","archived":false,"fork":false,"pushed_at":"2025-04-22T11:56:27.000Z","size":8850,"stargazers_count":2,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-10-30T21:24:30.667Z","etag":null,"topics":["approximate-inference","bayesian-inference","normalizing-flows","surrogate-models"],"latest_commit_sha":null,"homepage":"https://acerbilab.github.io/normalizing-flow-regression/","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/acerbilab.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2025-04-07T07:38:06.000Z","updated_at":"2025-05-03T08:24:19.000Z","dependencies_parsed_at":"2025-04-14T08:51:25.641Z","dependency_job_id":null,"html_url":"https://github.com/acerbilab/normalizing-flow-regression","commit_stats":null,"previous_names":["acerbilab/normalizing-flow-regression"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/acerbilab/normalizing-flow-regression","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/acerbilab%2Fnormalizing-flow-regression","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/acerbilab%2Fnormalizing-flow-regression/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/acerbilab%2Fnormalizing-flow-regression/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/acerbilab%2Fnormalizing-flow-regression/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/acerbilab","download_url":"https://codeload.github.com/acerbilab/normalizing-flow-regression/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/acerbilab%2Fnormalizing-flow-regression/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29695781,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-02-21T18:18:25.093Z","status":"ssl_error","status_checked_at":"2026-02-21T18:18:22.435Z","response_time":107,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["approximate-inference","bayesian-inference","normalizing-flows","surrogate-models"],"created_at":"2025-06-16T19:40:14.230Z","updated_at":"2026-02-21T22:32:12.515Z","avatar_url":"https://github.com/acerbilab.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Normalizing Flow Regression\n\nThis repository provides the implementation and code used in the AABI 2025 (proceedings track) article *Normalizing Flow Regression for Bayesian Inference with Offline Likelihood Evaluations* (Li et al., 2025).\n\n- See the paper [web page](https://acerbilab.github.io/normalizing-flow-regression/) for more information.\n- The full paper is available [on arXiv](https://arxiv.org/abs/2504.11554) and as [Markdown files](https://github.com/acerbilab/normalizing-flow-regression/tree/main/docs/paper).\n\n## Overview\n\nBayesian inference with computationally expensive likelihood evaluations remains a significant challenge in many scientific domains. We propose normalizing flow regression (NFR), a novel offline inference method for approximating posterior distributions. Unlike traditional surrogate approaches that require additional sampling or inference steps, NFR directly yields a tractable posterior approximation through regression on existing log-density evaluations. \n\n## Set up\n\n```bash\nconda env create -f environment.yml\nconda activate nfr\n# install kernel for jupyter notebook\npython -m ipykernel install --user --name nfr\n```\n\nSee `demo.ipynb` for an example of using NFR.\n\n## Citation\nTo appear in 7th Symposium on Advances in Approximate Bayesian Inference (AABI 2025, proceedings track).\n\n\u003e Li, C., Huggins, B., Mikkola, P., \u0026 Acerbi, L. (2025). Normalizing Flow Regression for Bayesian Inference with Offline Likelihood Evaluations. In 7th Symposium on Advances in Approximate Bayesian Inference.\n\n### BibTeX\n```bibtex\n@inproceedings{liNormalizingFlowRegression2025,\n  title = {Normalizing Flow Regression for {{Bayesian}} Inference with Offline Likelihood Evaluations},\n  booktitle = {Proceedings of the 7th Symposium on Advances in Approximate Bayesian Inference},\n  author = {Li, Chengkun and Huggins, Bobby and Mikkola, Petrus and Acerbi, Luigi},\n  year = 2025,\n  month = apr,\n  series = {Proceedings of Machine Learning Research},\n  volume = {289},\n  pages = {91--130},\n  publisher = {PMLR}\n}\n```\n\n## Acknowledgements\n\nThis repository includes code adapted from the `nflows` library: https://github.com/bayesiains/nflows, originally developed by Conor Durkan, Artur Bekasov, Iain Murray, and George Papamakarios.\n\nWe have modified `nflows/transforms/autoregressive.py` such that:\n- When neural network parameters are zeros, the flow becomes the identity transform.\n- The scale and shift transformation is constrained to a specified range.\n\n\u003c!--\n## Notation\n\n- target original/constrained space: the space where the target posterior is defined, potentially constrained by Cartesian product of intervals\n- target inference/unconstrained space: obtained by applying a transformation (e.g., probit transform) to the target original space\n- flow base space: the space where the flow base distribution is defined, unconstrained. `flow.transform`: target inference space -\u003e flow base space, `flow.inverse_transform`: flow base space -\u003e target inference space --\u003e\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Facerbilab%2Fnormalizing-flow-regression","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Facerbilab%2Fnormalizing-flow-regression","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Facerbilab%2Fnormalizing-flow-regression/lists"}