Deep learning frameworks

Last revised by Candace Makeda Moore on 3 May 2021

Citation, DOI, disclosures and article data

Citation:

Toumpanakis D, Moore C, Murphy A, Deep learning frameworks. Reference article, Radiopaedia.org (Accessed on 20 Apr 2024) https://doi.org/10.53347/rID-88651

DOI:

https://doi.org/10.53347/rID-88651

Permalink:

https://radiopaedia.org/articles/88651

rID:

88651

Article created:

15 Apr 2021, Dimitrios Toumpanakis

Disclosures:

At the time the article was created Dimitrios Toumpanakis had no recorded disclosures.

View Dimitrios Toumpanakis's current disclosures

Last revised:

3 May 2021, Candace Makeda Moore

Disclosures:

At the time the article was last revised Candace Makeda Moore had no recorded disclosures.

View Candace Makeda Moore's current disclosures

Revisions:

3 times, by 3 contributors - see full revision history and disclosures

Sections:

Artificial Intelligence

Tags:

deep learning, machine learning

Synonyms:

Deep learning libraries

Deep learning frameworks are instruments for training and validating deep neural networks, through high-level programming interfaces.

Widely used deep learning frameworks include the libraries PyTorch, TensorFlow, and Keras. A programmer can use these libraries of higher functions to quickly define the architecture and the parameters of an artificial neural network without the need of writing many more lines of code in Python (or another language) or explicitly defining processes such as backpropagation.

These particular frameworks rely on other libraries for GPU-acceleration (such as cuDNN, NCCL and DALI) to deliver relatively easy-to-implement high-performance multi-GPU accelerated training.