Declarative Probabilistic Programming for Undirected Graphical Models: Open Up to Scale Up
نویسنده
چکیده
We argue that probabilistic programming with undirected models, in order to scale up, needs to open up. That is, instead of focusing on minimal sets of generic building blocks such as universal quanti cation or logical connectives, languages should grow to include speci c building blocks for as many uses cases as necessary. This can not only lead to more concise models, but also to more e cient inference if we use methods that can exploit building-block speci c sub-routines. As embodiment of this paradigm we present Fast Froward, a platform for implementing probabilistic programming languages that grow.
منابع مشابه
A Chance Constrained Integer Programming Model for Open Pit Long-Term Production Planning
The mine production planning defines a sequence of block extraction to obtain the highest NPV under a number of constraints. Mathematical programming has become a widespread approach to optimize production planning, for open pit mines since the 1960s. However, the previous and existing models are found to be limited in their ability to explicitly incorporate the ore grade uncertainty into the p...
متن کاملFACTORIE: Probabilistic Programming via Imperatively Defined Factor Graphs
Discriminatively trained undirected graphical models have had wide empirical success, and there has been increasing interest in toolkits that ease their application to complex relational data. The power in relational models is in their repeated structure and tied parameters; at issue is how to define these structures in a powerful and flexible way. Rather than using a declarative language, such...
متن کاملTraffic Analytics using Probabilistic Graphical Models Enhanced with Knowledge Bases
Graphical models have been successfully used to deal with uncertainty, incompleteness, and dynamism within many domains. These models built from data often ignore preexisting declarative knowledge about the domain in the form of ontologies and Linked Open Data (LOD) that is increasingly available on the web. In this paper, we present an approach to leverage such “top-down” domain knowledge to e...
متن کاملLearning Acyclic Directed Mixed Graphs from Observations and Interventions
We introduce a new family of mixed graphical models that consists of graphs with possibly directed, undirected and bidirected edges but without directed cycles. Moreover, there can be up to three edges between any pair of nodes. The new family includes Richardson’s acyclic directed mixed graphs, as well as Andersson-Madigan-Perlman chain graphs. These features imply that no family of mixed grap...
متن کاملProbabilistic Inference in General Graphical Models through Sampling in Stochastic Networks of Spiking Neurons
An important open problem of computational neuroscience is the generic organization of computations in networks of neurons in the brain. We show here through rigorous theoretical analysis that inherent stochastic features of spiking neurons, in combination with simple nonlinear computational operations in specific network motifs and dendritic arbors, enable networks of spiking neurons to carry ...
متن کامل