Liquid Neural Nets - what comes after Deep Neural Nets

Wanna meet a small+smart Neural Network that learns on the job, not only during training? Here you go. Liquid Neural Nets are still relatively unknown for various reasons. For one thing, they are simply different from other neural nets - they adapt to changing challenges “on the job” and require a little more mathematics to do so - and they do not yet quite fit the bill of the big players, whose revenues are essentially generated and stabilised by the “size” and scalability of the applications or the resources required to operate the applications. Big AI today is what Big Data used to be - expensive, unsustainable and often of unclear value to the business. Unlike their oversized relatives from the neural net zoo, Liquid NNs are one thing above all - small. And they are constantly learning. In this talk, I would like to explore how this can work and why these special neural networks could represent a promising, particularly resource-saving alternative for many areas of application.