Optimizing Data-Flow in Binary Neural Networks

Explore the optimization of data flow in Binary Neural Networks through this insightful conference talk from tinyML EMEA 2022. Delve into the Hardware and Sensors session as Lorenzo Vorabbi, DL Labs Vision and Processing Methods Team Leader Support at Datalogic, presents key strategies for enhancing efficiency. Learn about binauralization, quantization, and the VTG Model, while understanding the importance of clipping operations and normalization optimization. Discover how to quantize the best normalization operation and examine accuracy results. Gain valuable insights into use cases, checkout models, and key points that can revolutionize your approach to Binary Neural Networks.

Introduction
Overview
Goals
Use Cases
Checkout Models
Binauralization
Quantization
VTG Model
Clipping Operation
Normalization Optimization
Quantizing the best normalization operation
Key points
Accuracy Results
Summary