{"id":13526266,"url":"https://github.com/ml-research/WhittleNetworks","last_synced_at":"2025-04-01T07:31:46.803Z","repository":{"id":94931251,"uuid":"372619505","full_name":"ml-research/WhittleNetworks","owner":"ml-research","description":"Code for \"Whittle Networks: A Deep Likelihood Model for Time Series\" @ICML2021","archived":false,"fork":false,"pushed_at":"2021-08-13T22:11:36.000Z","size":1099,"stargazers_count":9,"open_issues_count":1,"forks_count":0,"subscribers_count":5,"default_branch":"main","last_synced_at":"2024-11-02T11:35:03.702Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","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/ml-research.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}},"created_at":"2021-05-31T20:17:56.000Z","updated_at":"2024-03-03T17:58:37.000Z","dependencies_parsed_at":"2023-04-12T08:16:17.129Z","dependency_job_id":null,"html_url":"https://github.com/ml-research/WhittleNetworks","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ml-research%2FWhittleNetworks","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ml-research%2FWhittleNetworks/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ml-research%2FWhittleNetworks/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ml-research%2FWhittleNetworks/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ml-research","download_url":"https://codeload.github.com/ml-research/WhittleNetworks/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":246600738,"owners_count":20803472,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","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":[],"created_at":"2024-08-01T06:01:27.201Z","updated_at":"2025-04-01T07:31:46.155Z","avatar_url":"https://github.com/ml-research.png","language":"Python","funding_links":[],"categories":["Papers"],"sub_categories":["2021"],"readme":"# Whittle Networks: A Deep Likelihood Model for Time Series\n\nThis repository is the official implementation of Whittle Networks introduced in \n[Whittle Networks: A Deep Likelihood Model for Time Series](https://ml-research.github.io/papers/yu2021icml_wspn.pdf) by Zhongjie Yu, Fabrizio Ventola, and Kristian Kersting, published at ICML 2021.\n\n## Setup\n\nThis will clone the repo, install a Python virtual env (requires Python 3.6), and the required packages.\n\n    git clone https://github.com/ml-research/WhittleNetworks.git\n    ./setup.sh\n\n## Datasets\n\nDownload datasets from [TU Datalib](https://tudatalib.ulb.tu-darmstadt.de/handle/tudatalib/2887/), and unzip:\n\n    wget https://tudatalib.ulb.tu-darmstadt.de/bitstream/handle/tudatalib/2887/data.zip\n    unzip data.zip\n\n## Demos\n\n### Activate the virtual environment:\n\n    source ./venv_wnet/bin/activate\n\n### To train WSPN on some of the datasets:\n\n    ./run_WSPN.sh Sine\n\n\"Sine\" can be replaced with \"MNIST\", \"SP\", \"Stock\", or \"Billiards\".\n\nThis will train and evaluate WSPNs with 1d, pair, and 2d Gaussian leaf nodes. Details can be found in Table 1 in our paper.\n\n### To extract the conditional independence structure from a WSPN:\n\n    python script_graph.py --data_type=Sine --graph_type=bn\n\n\"Sine\" can be replaced with \"SP\", \"Stock\", or \"VAR\".\n\n```--graph_type``` can be either \"bn\" -- directed graph, or \"mn\" -- undirected graph.\n\nBayesian information criterion will be enabled with ```--BIC```\n\nPre-trained WSPN models are in ```results/```\n\n### To train and evaluate conditional WSPN for forecasting on the Mackey-Glass dataset: \n\n    python script_wcspn.py\n\n### To train and evaluate Whittle AE:\n\n    python train_WhittleAE.py\n    python test_WhittleAE.py\n\n\n## Citation\nIf you find this code useful in your research, please consider citing:\n\n\n    @inproceedings{yu2021wspn,\n      title = {Whittle Networks: A Deep Likelihood Model for Time Series}, \n      author = {Yu, Zhongjie and Ventola, Fabrizio and Kersting, Kristian}, \n      booktitle = { Proceedings of the International Conference on Machine Learning (ICML) },\n      pages = {12177--12186},\n      year = {2021}\n    } \n\n## Acknowledgments\n\n* This work is supported by the Federal Ministry of Education and Research (BMBF; project \"MADESI\", FKZ 01IS18043B, and Competence Center for AI and Labour; \"kompAKI\", FKZ 02L19C150), the German Science Foundation (DFG, German Research Foundation; GRK 1994/1 \"AIPHES\"), the Hessian Ministry of Higher Education, Research, Science and the Arts (HMWK; projects \"The Third Wave of AI\" and \"The Adaptive Mind\"), and the Hessian research priority programme LOEWE within the project \"WhiteBox\".\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fml-research%2FWhittleNetworks","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fml-research%2FWhittleNetworks","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fml-research%2FWhittleNetworks/lists"}