Should I add references to conference presentations?

If the slides you're using are going to have "independent life,"—in other words, if you're going to make them available separately from the conference paper (on your website, for instance), then the citations should be included as part of the presentation. I would follow posdef's example and place the citations on the same slide as where it's needed; this will save the reader from having to flip back and forth between different parts of the presentation or between the presentation and the paper.

Not including the citations is a bad idea, because it means you are potentially failing to give people the credit they deserve for ideas that were originally theirs. Even though it's "just" a conference presentation doesn't mean that the rules of crediting people for their work should be ignored. (Citing the work of others is also the right thing to do from the perspective of "playing nice with others." Taking credit for other people's work can make them leerier of working with you.)

I don't know if there is a specific way within the CS community but the way most established seniors seem to do in my field is to note down the reference at the bottom of the slide where they refer to someone's results/figures.

I think this is a better approach than to list them all in the end, because the audience gets the reference together with the content, that way you don't have to puzzle the references and the content 6 months after you attended the presentation.

If the people you are referring to are people you have had collaborations or communication with, it would not hurt to have them listed in a "thanks to" or more formally "acknowledgements" slide.

Hope it helps

Applied mathematician here; my solution is putting them on the same slide as the material. I use formats such as [Someone '99], [Lin WW, '00] (initials are almost mandatory for some common surnames), [Doe et al, book '04], [P and SomeoneElse, preprint '12] (my name is always abbreviated to an initial, which is a common convention). I find it a good compromise between clarity and shortness: I don't need to include a full sentence, but only the names in brackets.

You can use a different color or font to differentiate them visually from the text --- preferably something light but readable, a color that does not attract much attention.

I use them sparingly nevertheless --- overall I have typically less than 10 such citations in a 15-20 slide talk.

This makes immediately clear whether I think that a theorem is new/mine or not. Its original authors could be in the audience, so I think it's important to acknowledge them properly.

If your slides are already so cramped that these citations won't fit, then you have a much bigger problem. :)