Search

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., Comment: 10 pages

We have explored different methods of improving the accuracy of a Naive Bayes classifier for sentiment analysis. We observed that a combination of methods like negation handling, word n-grams and feature selection by mutual information results in a significant improvement in accuracy. This implies that a highly accurate and fast sentiment classifier can be built using a simple Naive Bayes model that has linear training and testing time complexities. We achieved an accuracy of 88.80% on the popular IMDB movie reviews dataset., Comment: 8 pages, 2 figures

It is shown in this paper how a connection may be made between the symmetry generators of the Hamiltonian (or potential) invariant under a symmetry group $G$, and the subcasimirs that come about when the rank of the Poisson structure of a dynamical system drops by an even integer. This {\em kinematics-dynamics} connection is made by using the algebraic geometry of the orbit space in the vicinity of rank change, and the extra null eigenvectors of the mass matrix (Hessian with respect to symmetry generators) of the Hamiltonian (or potential). Some physical interpretations of this point of view include a control-theoretic prescription to study stability on various symplectic leaves of the Poisson structure. Methods of Invariant Theory are utilized to provide parametrization for the leaves of a Poisson dynamical system for the case where a compact Lie group acts properly on the phase space, which is assumed to be modeled by Poisson geometry.