Quantum mechanics

Context (archaeology)

Control (management)

Scale (ratio)

Biology

Dataflow

Telecommunications

Set (abstract data type)

Scalability

Parallel computing

Computer science

Control flow

Distributed computing

Paleontology

Physics

Reinforcement learning

Database

Mathematics

Flow (mathematics)

Flow control (data)

Inference

Artificial intelligence

Geometry

Online machine learning

Programming language

Artificial neural network

Dataflow architecture

Machine learning

Many recent machine learning models rely on fine-grained dynamic control flow for training and inference. In particular, models based on recurrent neural networks and on reinforcement learning depend on recurrence relations, data-dependent conditional execution, and other features that call for dynamic control flow. These applications benefit from the ability to make rapid control-flow decisions across a set of computing devices in a distributed system. For performance, scalability, and expressiveness, a machine learning system must support dynamic control flow in distributed and heterogeneous environments.

Dynamic control flow in large-scale machine learning