No, you shouldn't use "I." I used to be in graduate school and have never read a single paper which does that. Scientific efforts typically involve multiple researchers even if only one person is listed as the author, so that's why in most circumstances, it is "we." I don't think using "I" is proper even if there is really only you alone who is involved, because that sounds too personal for a scientific paper. (Not for political theory papers, where saying "I argue that ...." is totally fine.) However, note that it is extremely common to use the passive voice in scientific papers. Therefore, when you *have* to say "I did ....", better say "... is done." z.B. "I measured the bit error rate of this modulation scheme and found it to be 10e-3." should be written as: "The bit error rate of this modulation scheme is measured to be 10e-3." Of course, in the dedication part of a thesis, feel free to say things like "I would like to thank [my adviser, wife, etc. ] for ... " because that's the part where being personal is OK. But not in the real contents.

Hi this Italki site will also help you http://www.italki.com/knowledge/contribution-TIPS+FOR+WRITING+A+THESIS.htm

Thomas has the answer, but I'd just add a little, and a little softer. Avoid 1st person, (I, we) in the text. I agree the passive is a good way round it. It seems maybe less important than it once was. Is your audience progressive or traditional? The safe approach is to avoid the 1st person, by wording things differently. I suspect that using "we" is more common in the abstract. Usually there is a way to reword. I scan the abstracts from the last conference "We use a broadband variant ".... "Wideband acoustic signals are employed" ... "This paper compares the performance of "... "We describe recent developments in modem-based"... single author paper "This paper presents a quantitative analysis"... "We will provide an in-depth look"... In the text "Here we consider only our worst case with the low SNR type D"... Worded to include the reader? (Scotish paper) "To identify the featureless classifier to use, we consider how other applications with similar issues"... (US paper) It seems they make "we" more acceptable by not starting with it? "Once features are detected, they can be described".. (Canadian paper has no 1st person" & uses passive voice a lot. Richard is right in his first statement, consult the experts. But for the second statement: To use "I" would be in a very small minority of thought in this, I think.

If I were you, I would ask my teacher/professor this same question. However, if I were the author, I would simply use "I".