Theano is a numerical computation library for Python. It allows user to define, optimize, and evaluate mathematical expressions that involve multi-dimensional arrays. The actual language of Theano expressions is symbolic.
It compiles mathematical expressions in Python. It picks up structures and turns them into very efficient code. The code optimizations extract as much performance as possible from the hardware. It uses NumPy, native libraries like BLAS and code (C++) to run CPUs or GPU as fast as it can. Specifically, expressions are defined in an abstract sense, compiled and used to create calculations.
Theano was designed to handle computations required for large neural network algorithms used in Deep Learning. In 2007, Theano's development was a pioneer of libraries of its kind. It is an industry standard for Deep Learning research and development. The MILA group at the University of Montreal developed the computation library and named it Theano, the Greek mathematician and philosopher.
In 2017, the MILA group stopped active development on the Theano project, citing the availability of other actively developed frameworks being made available.
Introduction to the Python Deep Learning Library Theano
Synkhronos: a Multi-GPU Theano Extension for Data Parallelism
Adam Stooke, Pieter Abbeel
Theano: A Python framework for fast computation of mathematical expressions
The Theano Development Team
Documentaries, videos and podcasts
Theano - Ep. 17 (Deep Learning SIMPLIFIED)
12 january 2016
- Python (programming language)A general purpose high-level programming language
- MXNetA deep learning framework and multi-language machine learning library
- KerasKeras is an open source neural networks API written in Python built on top of Tensorflow, CNTK or Theano.
- TensorFlowA machine learning software library for numerical calculation
- Microsoft Cognitive Toolkit
- Show More