Cloud Application Acceleration for AI Integration

AI (Artificial Intelligence) is a world of newly opened possibilities for application developers.
By leveraging machine learning or deep learning, it is possible to achieve much more precise user profiling, personalization, and recommendations. Applications can be improved in various ways such as implementing smart search, voice interfaces, or intelligent virtual assistants.  

Currently, the most cost-effective way to implement AI is to use cloud services such as Google, AWS, or Microsoft. This seminar introduces a seminar on building the cloud AI software and hardware environments necessary for AI and machine learning developers in these clouds. 

Key Content
This session explains how to deploy machine learning applications using the Machine Learning Deployment Stack from Xilinx, a machine image provided by Xilinx, utilizing the F1 instance, an FPGA acceleration card from Amazon Web Service (AWS).
 
Detailed Content
- Overall design flow, software stack, xfDNN inference toolbox 
- Architecture and specifications of the machine learning process engine
- Compiling and quantizing required for deploying user networks using TensorFlow on AWS
- Placing YOLO network downstream of streaming images using xfDNN Python API and performing object detection
- Introduction to multiple network implementation process