Deep Learning with F#: An Experience Report
Please log in to watch this conference skillscast.
This talk details the experience of using F# for DL and data science (DS) on a daily basis in two large corporations and compares F# with Python for DL/DS
With respect to DS/DL, there are two issues of concern with Python: (A) models are becoming tightly integrated with the front-end language (e.g. Python and PyTorch). This makes its harder to consume some recent models from other languages / ecosystems. And (B) today's data pipelines, can become large and complex resulting in large code bases surrounding the actual DL model(s). Python's lack of static typing and slowness is a major impediment.
F# is expressive, statically type-checked and much faster than basic Python at data handling. Combine that with tooling that supports interactive programming well an F# is ideally suited DS/DL for the .Net ecosystem.
YOU MAY ALSO LIKE:
- Functional Concurrency in .NET with C# and F# with Riccardo Terrell (Online Course on 7th - 10th December 2021)
- Getting Geospatial Data on the Web (Online Meetup on 10th February 2022)
- Haskell: A Conversion Experience (SkillsCast recorded in November 2021)
- Using F# to Structure an Island of Simplicity (SkillsCast recorded in October 2021)
Thank you to our sponsors and partners
Gold
Deep Learning with F#: An Experience Report
Faisal Waris
ML/AI researcher and data scientist for over 5 years of experience in data science and deep learning. PhD in Computer Science from Wayne State University.