Fuzzy Labels

 This will be a bit of a strange connection between queer culture and machine learning, but stay with me.

So a lot of machine learning is finding similarities and differences between data, whether that's numbers or pictures or something else. As a result, defining boundaries is a key concept. Now there's broadly two ways to define these boundaries: strict and fuzzy boundaries. Strict boundaries are hard limits, for example saying that anyone under 5'2" is short while 5'3" is medium. Fuzzy boundaries are looser, so saying that someone can be both short and medium height at the same time. Fuzzy boundaries were essentially developed to imitate the way humans typically think of boundaries. You can be short in some situations and medium height in others. We don't assign hard labels to anything, it all tends to be situational.

What this made me think of though were labels that we use to describe our identities and orientations. We can use one label in one situation, and one label in another. To some extent, we can be many different labels. Maybe we're kinda bi and maybe we're kinda lesbian. There isn't a strict box for any label, and there isn't a strict box for any person.

But many people don't think of labels this way. We put a lot of pressure on ourselves to fit into one definition and for that to be a perfect description of us. Or alternatively a lot of people can put that pressure on us and invalidate certain labels for groups of people. What's really interesting to me though is that this system of "fuzzy boundaries" was developed to fit more how people think of categories. And yet we sometimes reject this mentality in favor of being imposing and dictating just what a certain identity looks like.

Essentially, we all need to relax and embrace fuzzy labels a little more. Don't sweat the details so much, this is an established thing and method of thinking. We are all going to be just fine if we don't fit one category perfectly.