Uncategorized

Reactive LDA Library

FavoriteLoadingAdd to favorites

In this post, we introduce a parallel LDA (Latent Dirichlet Allocation) library. It’s available on Github. The library is written in Scala with Akka actors (hence the name Reactive). The underlying algorithm is based on vanilla Gibbs sampling. We will show that this approach scales much better than training LDA with collapsed Gibbs sampler, due to the easiness of parallelism.…
Original post: Reactive LDA Library
Source: Kifi