Shogun, a language agnostic machine learning library
summary
A C++ machine learning toolbox that exposes the same API across languages via SWIG-supported multi-language bindings (Python, R, Java, Octave, Lua, C#, Ruby) and auto-generates examples with a meta-example framework so one example becomes runnable in every binding. Focused on large-scale kernel methods it includes efficient SVMs, Multiple Kernel Learning, kernel hypothesis testing and Krylov solvers, and records a 10% runtime reduction on a kernel test. Runtime switches between CPU and GPU, and public cloud.shogun.ml Jupyter notebooks host runnable demos.
AI-written summary · about
chapters (12)
- 0:00Introduction to Shogun
- 1:04Library origins and scope
- 2:28Target users and licensing
- 3:24Platform and architecture
- 4:17Contributions and GSoC
- 5:04Interface philosophy
- 7:46Multi-language support
- 9:48Usability challenges
- 11:17Education and demo showcase
- 13:40Cloud resources for Shogun
- 14:44Real-world applications
- 16:13Future developments
related talks
05mAn Introduction To Shogun Machine Learning Toolbox One For All
30mSnap ML: Accelerated, Accurate,Efficient,Machine Learning
27mInterpretability and explainability in machine learning
45mOptimizing the Performance of Machine Learning in Enterprise Java SaaS with GraalVM, Python and CUDA
42mSnap ML: a highly-accelerated, scalable software library for machine learning
01h 03mScaling up machine learning without tears (and what do programming languages ha...
25mLeveraging AI Ecosystem for Portable and Sustainable GPU Kernels in HPC
22mProductive Machine Learning and Deep Learning Projects