A bad feeling about first published papers

You should do nothing other than continue growing and learning. The most common reason would be #1. I can't say the most likely reason is #1, not having seen the papers.

But a lot of authors, even poets, look back on their early work with a sense of wonder about how they could have been so naive. It is a sign of growth.

Yes, #2 is probably also a factor to some extent as CS has been a fast moving field overall, not being much over 60 years old, compared to, say math and such.

And you haven't said that the papers are actually wrong, just a bit naive. Hopefully we all learn some things as we grow older. It normally happens because we are actively thinking about things.

Others at the time (reviewers, editors, advisors) thought that the papers were fine for the time. They were fine.

Just ignore the bad papers. Don't include them in your curriculum vitae. Instead of using the title "Papers" in your CV, just use the title "Selected papers." That implies that you have some papers that you are not proud of.

If indeed:

• "my assumptions were naive" and
• "the conclusions ... there is a high chance that they are not valid for other datasets"
• "I could explain my approach differently"

(which, I should say, might just be your misjudging your past work, as other answers suggest; but if that is actually the case)

then consider writing a new paper about the subject. Regardless of whether it is sufficiently important for a top-tier publication, you can at least share your further-developed you of the subject with the community.

It is not uncommon for papers on some subject to begin with partial insights, simplistic assumptions and limited applicability, and develop - often by the same authors - into something deeper, more mature and more useful.

Note: If you do write a paper, don't waste your time and the readers berating your previous work; rather, focus on the new presentation of the previous method, a discussion of its assumptions (as opposed to other, less-naive assumptions) and dispassionate explanation about its limits of applicability. If you can add some new result that doesn't make this assumption - even if it's a proof of impossibility or a concrete useful/insightful counter-example, that's even better. But don't write a "my last paper sucks" paper.