We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Horizon: Facebook's Open Source Applied Reinforcement Learning Platform

Abstract: In this paper we present Horizon, Facebook's open source applied reinforcement learning (RL) platform. Horizon is an end-to-end platform designed to solve industry applied RL problems where datasets are large (millions to billions of observations), the feedback loop is slow (vs. a simulator), and experiments must be done with care because they don't run in a simulator. Unlike other RL platforms, which are often designed for fast prototyping and experimentation, Horizon is designed with production use cases as top of mind. The platform contains workflows to train popular deep RL algorithms and includes data preprocessing, feature transformation, distributed training, counterfactual policy evaluation, optimized serving, and a model-based data understanding tool. We also showcase and describe real examples where reinforcement learning models trained with Horizon significantly outperformed and replaced supervised learning systems at Facebook.
Comments: 10 pages
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Cite as: arXiv:1811.00260 [cs.LG]
  (or arXiv:1811.00260v5 [cs.LG] for this version)

Submission history

From: Edoardo Conti [view email]
[v1] Thu, 1 Nov 2018 07:02:45 GMT (37kb)
[v2] Tue, 5 Mar 2019 22:49:50 GMT (385kb,D)
[v3] Wed, 1 May 2019 05:57:02 GMT (228kb,D)
[v4] Thu, 30 May 2019 20:47:42 GMT (226kb,D)
[v5] Wed, 4 Sep 2019 19:30:00 GMT (231kb,D)

Link back to: arXiv, form interface, contact.