MARVEL improves the performance of deep neural networks with just-in-time classification
Deep Learning is a common machine learning method, but it takes a lot of computing power and is not always suitable for applications, where decisions need to be taken closer to the sensor. To address the challenge, MARVEL partners have found ways to improve the performance of deep neural networks with a method called just-in-time classification. This is the result of using multiple neural layers, each performing a high-dimensional transformation of its input. In their latest blog post, MARVEL describe the challenges and the solutions in detail - find out more at the project website!