This course is a prerequisite before attending “AI on Intel: AI from the Data Center to the Edge – An optimized path using Intel architecture” workshop.
This course consists of four sections:
- Introduction to AI
- Machine Learning
- Deep Learning
- Applied Deep Learning with Tensorflow
The lessons consists mainly of presentations and some videos. They may also contain some activities like case studies and programming exercises using Jupyter notebooks.
Registration for Workshop
Upon completion of this course, an email will be sent to you with a registration link to sign up for a workshop session. You will then be notified if your registration is successful and further instructions will be provided. See below for more information.
Limited slots are available. Registration will be on a first come first serve basis.
AI From the Data Center to the Edge
Intel’s AI Portfolio
- Hardware: From training to inference with emphasis on 2nd Gen Intel® Xeon™ Scalable Processors
- Software: Frameworks, libraries and tools optimized for Intel® Architecture
- Community resources: Intel Developer Zone Resources
Exploratory Data Analysis
- Obtain a dataset
- Explore data visually to understand distribution
- Data Reduction and address imbalances
Training the Models
- Infrastructure: Intel® AI DevCloud, Amazon Web Services, Google Compute Engine, Microsoft Azure
- Process: Prepare and visualize the dataset, prepare for consumption into framework, hyper-parameter tuning, training, validate
- Check your scores
- Compare your results
- Hyper parameter tuning
- Pick the winner or go back to training
Deploy on the Edge/Inference
- Introduction to the Intel® OpenVINO™ Toolkit – Capabilities and benefits
- Usage Models
- Model Optimizer – Optimize model, generate hardware agnostic Intermediate Representation (IR) files for prebuilt and custom models
- Demo of Inference Engine – Deploy to CPU, integrated GPU, and Intel® Movidius™ Neural Compute Stick
If the ‘Mark Complete’ button does not enable at the end of the lesson, please refresh the page.