courses
Deep Learning with FPGAs
The penetration of FPGAs into the DSP communication market is fuelled by the following factors – the growing requirements for processing speeds of the order of 10-100 billions of operations per second, the need for rapid prototyping and Software definable architectures. FPGAs are embedded in microcontrollers with the aim to build complex and low cost products that are reconfigurable. This has led to huge demand for developing complex embedded system based application on FPGAs. The Hardware-software co-simulation and debugging is one of the most important aspects of developing embedded applications on FPGAs. FPGAs provide low latency and flexible architecture for accelerating deep learning solutions. The primary aim of this course is to build practical knowledge on convolution neural network and other similar networks for FPGA implementation.
What do you gain from this course?
- The industry skill sets required to design and model digital systems using HDL
- Hands on expertise in designing complex signal processing applications on FPGAs
- Theoretical and practical knowledge on FPGAs for embedded systems
Modules and it’s duration of the course are as mentioned below:
- HDL and Processor Design: 15 days
- FPGAs for DSIP: 15 days
- FPGAs for Embedded Systems: 15 days
- Mini Project: 20 days
The Syllabus covered in detail:
- Classical Machine Learning vs. Deep Learning
- Training vs. Inference
- Common Data Sets
- Image Net Classification Competition
- Neural network fundamentals
- Convolution neural network
- Network activation functions
- FPGA architecture
- DSP block sets
- Configurable logic
- FPGA interfaces
- FPGA advantages for deep learning
- Computer Vision (CV) Primer
- CV algorithm basics
- Components of Vision Systems Architecture
- Building a deep learning computer vision application
- Different programming languages, hardware and tools
