Monday, February 15, 2021

What is the main problem with Autoencoders that convert text and numbers to HTML for publishing online?

 I can speak here only for what I do here at my site and not for others. There likely are many more problems than this that can occur in other functions in regard to autoencoders but this is what is a problem for me personally here.

The problem is that all programmers have different styles of programming even in HTML and compatible languages like Java and others.

So, since autoencoders are somewhat set up mathematically you can get many different results when you are quoting potentially infinite programming styles you will now encounter online from literally every country on earth.

For example, not everyone speaks English as their first language which is usually the language autoencoders are programmed by in the first place. However, there are always exceptions to this rule. But, it's sort of like how all airline pilots are supposed to be able to speak English so they can all understand each other in emergencies worldwide as they arrive in emergency moments at random times.

So, for my site there is the programming of the people who host my site (blogspot or blogger.com) which is one set of programming. Then there is who I want to quote (which might be a site literally anywhere on earth). Then what I'm quoting might have already been run through Google Translate from whatever language it was originally written in like the French Wikipedia stuff I quoted recently was all originally written in French en Franc.

All these factors can play havoc when running all this stuff through an autoencoder which is basically set up in an mathematical or formulaic way using something similar to algebraic formulas.

So, you might get literally anything out the other side which is often what you see at my blog from quoting sources all over the world when using an autoencoder.

No comments: