Contents

Accelerated Machine Learning with Intel (Morning Sessions)

• Introduction
• Welcome & presentation of the day's agenda and speakers.
• Hardware acceleration for AI and Intel® oneAPI AI Analytics Toolkit
  In this session, we will first introduce the hardware features that power and accelerate AI on Intel hardware.
  We will then take a first look at the software stack that leverages them, namely the Intel® oneAPI AI Analytics Toolkit.
• Intel Developer Cloud (IDC) - A sandbox for AI & Software development
  Intel Developer Cloud is a development environment that gives access to cutting-edge Intel hardware and software innovations to build and test AI, machine learning, HPC, and security applications for cloud, enterprise, client, and edge deployments. Learn about Intel's advanced CPUs, GPUs, and accelerators, along with open software tools, to optimize your AI products and solutions.
• Introduction to the Intel Distribution for Python, Data Parallel Extensions for Python, and Accelerating Classical Machine Learning on Intel Hardware
  In this session we will show you how to boost your classical machine learning models on Intel hardware. We will explore the use of Intel Extension for Scikit-learn, and libraries like daal4py to accelerate XGBoost and LightGBM models. Learn about leveraging the Intel Distribution for Python and optimizations for numpy, scipy, and parallel extensions (dpnp, dpctl, numba-dpex) for running Python on Intel CPUs and GPUs.

Accelerated Deep Learning with Intel (Afternoon Sessions)

• Generative AI Powered by Intel
  In this session, we present to you the most recent advancements in Generative AI, covering Large Language Models and Diffusion Models. We will explore how Intel plays a crucial role in powering this technology, from training and fine-tuning to inference across a spectrum of Intel hardware platforms.
• Introduction to Distributed Deep Learning across Intel Hardware
  In this session, we will talk about Distributed Deep Learning and how it can be effectively implemented on Intel's cutting-edge hardware platforms. We will cover topics like why Distributed training is required, explore various techniques of neural network parallelism, will cover various tools like such as Intel oneCCL, Distributed Data Parallel (DDP), Horovod, Fully Sharded Data Parallel (FSDP), and DeepSpeed which are essential for scaling of AI model.

The presentations will be accompanied by demos to demonstrate the performance improvements.
The course is organised by LRZ in cooperation with Intel.

Prerequisites

• Basic knowledge of Python
• AI Training Series: Orientation Session (or comparable previous knowledge)
• Good understanding of Deep Learning and Classical Machine Learning (courses such as Introduction to Deep Learning (I2DL) (IN2346) are provided by TUM - material also available for non-TUM students)

Language

English

Lecturers

Sneha Chattopadhay, Akash Dhamasia, Stefana Raileanu, Severine Habert (all Intel)

Prices and Eligibility

The course is open and free of charge for academic participants from Germany.

Registration

Please register with your official e-mail address to prove your affiliation.

Withdrawal Policy

See Withdrawal

Legal Notices

For registration for LRZ courses and workshops we use the service edoobox from Etzensperger Informatik AG (www.edoobox.com). Etzensperger Informatik AG acts as processor and we have concluded a Data Processing Agreement with them.

See Legal Notices