Robustness

Adding Consistency Loss between AugMix Augmented Images to enhance the Generalization of your Image Classification ModelFigure 1 - Visualizing AugMix: Original Image (left) and Two Augmented Versions. | Image by authorIntroductionImage classification models are best able to predict data from the same distribution as the training data. However, in real-world scenarios, the input data may suffer from variations. When inferencing with different cameras for example, the lighting conditions, contrast, color distortions etc.…

source. Just as the world was turning to the crisp fall of 2022, one of the largest functional blockchains went down.This wasn’t the first time either — Solana has had its fair share of downtime over the years, particularly this year.The worrying aspect was: it was due to a single node misconfiguration — one that took out the chain for a whole 6 hours. As one can imagine, for a financial crypto giant, this is a big deal.For a protocol that popularized and has a respectable Nakamoto Coefficient score, it brings to light…

OpenAI’s Whisper will enable speech recognition apps to reach new levels of efficiency Speech recognition or voice recognition technology has come a long since the concept first emerged. But users continue to have only one persisting problem with voice recognition, which is accuracy. Over the past couple of years, researchers have been working on building AI algorithms that can accurately process voice input and consistently focus on the research and development of speech development. Recently, OpenAI’s Whisper is making…

What are adversarial attacks and How to protect your embedded devices from thoseImage by authorA Quick IntroWith very limited options for adversarially robust Deep Neural Networks (DNN) for Embedded Systems, this article attempts to provide a primer on the field and explores some ready-to-use frameworks.What is an adversarially robust DNN?Deep Neural Networks have democratized machine learning and inference. There are two primary reasons for that. Firstly, we do not need to find and engineer features from the target…