Addressing the Garbage-In, Garbage-Out Problem in Deep Learning Models. NVIDIA GTC21

Watch Roey Mechrez, CTO and Co-Founder of BeyondMinds, speak about “Addressing the Garbage-In, Garbage-Out Problem in Deep Learning Models” at the NVIDIA GTC21 event.

Roey speaks about the challenges of building a stable AI system in production where data scenarios are complex.

During the presentation, Roey explains how the need for a solution that functions well in every data scenario has led BeyondMinds to develop a method to control, monitor, and screen the data at test-time. Roey continues to explain the considerations for building the method, how it combines in systems and how recent advancements can contribute. Roey then elaborates on the effort to reduce computation burden and demonstrates the contribution of the method to the overall stability of the system.