Could I change part of my paper after acceptance

Since you have not proof read the paper yet, you simply ask the editor to make the changes to the table as part of the proof reading process. This is perfectly fine as long as the changes you make are trivial and does not change the basis for your conclusions. The worst case would be if the paper was accepted due to the data you reported and your changes will change the results so that the conclusions no longer can be made. The proof-reading stage comes after an accept and it is not allowed to make substantial changes. By substantial, I mean both by volume (changing a lot of text) or by quality (changing data upon which the decision was made).

Supply the changes to the editor as part of the proof reading process but make sure you explain that the changes in the table does not influence the results and conclusions of the paper.

Once the paper is published, there's no changing it. This seems to be the case whether it is physically printed in a magazine or first displayed online. Some journals will publish retractions, which are usually listed as a separate publication explaining what should be different in the original publication, though I'm sure that's not what you want.

I can say with certainty for PLoS One, if you need to issue revisions in that stage between acceptance and publication, they will fix minor issues if it doesn't fundamentally change the direction of your paper. Editing is hard work - it doesn't make sense to send it back for revisions if the paper hasn't changed too much. You'll want to send a copy of your revised paper to the editor, along with an letter/email detailing exactly what has changed in the paper and especially how it doesn't affect the other parts of your paper. Remember that the editor is also one of your reviewers, so if you can make the case that the mistake is indeed minor and take the time to correct it yourself, the decision to send it off for a seemingly pointless re-review becomes less attractive.

Source: A co-author made a small mistake that was corrected after acceptance.