Develop Efficient AI Solutions with Accelerated Machine Learning

Tailor your approach to developing efficient AI solutions with accelerated machine learning.

This expert-level workshop focuses on techniques for maximizing AI solution acceleration using components of the AI Tools. Workloads performed by CPUs and GPUs from Intel are surveyed, and implementations using key components, such as Intel® Extension for Scikit-learn* and Data Parallel Essentials for Python* language, are presented.

Topics covered include:

• Enabling patching and unpatching of scikit-learn—from fine-grained to global—for optimizing functions.
• Applying "compute follows data" principles via several algorithms including K-means, pairwise distance, and principal component analysis (PCA).
• A demo of data parallel Python with high-performing code targeting Intel CPUs and GPUs.
• Numba-dpex (a Numba* data-parallel extension), including examples of data-parallel code inside @numba.jit decorator and @kernel decorator functions readied to offload to a SYCL* device.
• How to write Python native extensions more easily using data parallel control (dpctl), a companion library based on SYCL.