On the Rules of Computer Engineering and How to Break Them, November 2024

[@Tesfamichael.G]

This is a fantastic video! It offers a refreshing perspective on the importance of challenging established norms in computer engineering. Your insights on pushing boundaries and exploring unconventional approaches are truly inspiring. This video encourages a mindset of innovation and reminds us that groundbreaking advancements often come from questioning the status quo. Thank you for sharing!

[@mehdis.7404]

As an example of rules creating dogma I'd like to mention the NNs case. If you pick any old (before 2012) neural nets book, you are being told that the rule of "no more than three layers of neural nets are needed", because of some math proof that three layers can approximate any function possible and that discouraged stacking up more layers as those may cause overfit or unncessary computing. And if we had followed that rule we would never have Deep Neural Networks we have today. I think that 3 layers rule was for long idleness in AI before AlexNet come along.