After the double line below, you will find some general guidelines that apply to any technical talks; adapt them for your particular case. Some specific remarks for the SysID project: * A reasonable structure would be: introduction/problem statement, methodology/algorithm (focusing here in particular on the generation of the regressors, since that is the most interesting part; but also any other nontrivial points in e.g. the tuning procedure,etc.), tuning results, simulation and prediction graphs with the best model, and discussion/conclusions. You can adapt, you don't have to use these exact titles, etc. Don't skip the required results though. * Everyone solves the same problem, so you can keep the intro/problem statement part short. Don't skip it though! Present the problem as YOU understand it, which will tell the evaluators to what extent you understand it. * You can assume that your audience understands system identification concepts in general, and linear regression; but you still have to give some equations in order to introduce your particular notation! And you should explain the nonlinear and polynomial ARX concept, because everyone understands things differently, and this will let us evaluate how much you know. DESIGNING AND DEVELOPING THE SLIDES ### Consider carefully the time that you have to give your presentation. If the time is not enough to present all the research that you had in mind, scale down. Select one or a few representative points and focus on them. It's better to get part of the point across, than to try to tell everything and fail to reach the audience. ### Think about the structure of the presentation in advance. Don't just start writing material on the slides, you'll have to backtrack and remove, and that's going to take much more time than a well-structured approach. Design the presentation is top-down: think about the overall structure first, and then go hierarchically down into the details, filling each section with content. Organize the material: - section-level: e.g., a 15-minute presentation will typically have at most 5 sections, preferably up to 4 (excluding the conclusions) - sub-section level, if applicable - down to the slide level - perhaps even each idea or figure that you want to put on each slide Write this structure down on paper. Examine it critically, make changes if necessary. Think how much time you will take to present a slide, on average. Cut down from the slides if they are too many for the alloted time slot. Using pen and paper at the design stage is generally very helpful. Try to find a good balance between high-level explanations and technical detail; use examples where necessary. ### Include an introduction/motivation, a clear problem statement, and -- at the end -- a concise summary and some conclusions: the `take-home message' that you want to leave the audience with. ### Make sure the level of the introduction (and of the entire talk) is appropriate for the audience. It's one thing to present to a room full of specialists that know exactly what you're talking about; and another to try to make laymen understand a difficult topic. ### Avoid flashy slides, animations, fade-ins etc unless absolutely necessary. NEVER add them just for their own sake. They distract the attention from the content, to the form. ### Use a clean slide template, white background and dark colors for text and images. NEVER the other way around. Beamers have poorer contrast than CRT and LCD screens (which means your presentation will look worse than what it does on the computer screen), and if you have a dark background it's more than likely your audience won't be able to read your slides. ### Limit the use of color to a few, very different colors (different from one another). Beamers typically distort colors, and if they are not different enough who knows what people will see. E.g., never use yellow on white background, it'll be nearly invisible. ### Add checkpoints to the slides, e.g., insert section-level Table of Contents with the current section highlighted, at the beginning of every section. Make it clear when you talk, that you have reached a checkpoint, e.g., say: "So, now we found out how to do X and Y. Next, I will describe our technique to do Z, that uses X and Y." Instead of this, you may also simply just take a short break (slowly count to five in your mind). Such checkpoints offer "breathing-space" to the audience, where they can collect their thoughts, and let ideas "sink in". ### Use pictures. A picture is sometimes worth a thousand words. Don't be afraid to use words when necessary, though, and don't make your entire presentation a sequence of pictures (sometimes, five words are worth twenty pictures...). Such a presentation is also difficult to follow. You need to find a balance. ### While writing the words: keep sentences very short; actually, words on a slide should almost never form entire sentences. They should be brief sequences expressing each a particular idea, preferably never longer than 10 words, if possible less. Typically, articles such as "a", "the" can be entirely dropped from the presentation. Use & instead of and, and similar tricks, to shorten the text and make it easy to digest. ### Rehearse your presentation beforehand, making sure that you do not exceed the allotted time interval, but also that you do not fall significantly short of filling it. Rehearse at least once. Possibly twice. Three is already too much, because if you rehearse it too many times you'll learn it by heart, and speak too fast without adding emphasis in the right places. If you find at this point that you exceed your time, cut from the slides, rather than speeding up. HOLDING THE TALK ### Since not everyone will be acquainted with the presenters, it is very important to introduce yourself clearly to the audience when starting to present. ### Speak clearly, not too fast, don't rush through things. Don't be slow though, time is always limited. Make sure to add emphasis (using your voice) on important concepts, new ideas, important words, while speaking. Don't use a flat tone, that is boring and will prevent people from following what you have to say. ### Face the audience at least 50% of the time. Don't always stare at the slides while talking. Laptops can display image on the screen and also output to the beamer, use this feature to see your slides while facing the audience. In the instances where you need to use a pointer, half-face the audience, and half the slides. If you're right-handed, sit towards the right of the slides, and the other way around. Like that you'll be able to point without making wide moves of the hands, which can be distracting. Don't take the "face the audience" advice too far either. If you need to explain some detailed thing on the slide, or write on the board, you will of course have to look there. As always, it's not about absolutes but a rough guideline. ### Check the time once every few minutes, to see whether you are on schedule. If you're late, you may speed up a little, but not too much! better to skip stuff. NEVER EVER skip the conclusions. HANDLING QUESTIONS ### Let the person finish their question, always! Do not start answering half-questions or assume what their question will be. ### If somebody questions your results and they are wrong, don't be afraid to tell them. It doesn't matter that they are Prof.Dr. Whoever. But if somebody does question, consider it carefully, and if you think they may be right acknowledge it! Don't get defensive, simply acknowledge that they may be right and you'll investigate further. ### Don't say stuff like: "Hmm, that's an interesting question..." to gain time. Only say it if you mean it. Better, take a small break (at most 5 seconds) to collect your thoughts before you answer. Don't use 'well', 'uh', and stuff like that to fill-in the gaps between your words. Well, at least try not to. ### Don't turn questions into a long, one-on-one discussion with the person that asked the question. Other people may have questions and time is always limited. If the discussion gets too long, say you'll continue it after the talk, and take questions from other people. ### What if they don't ask questions? Well, then you didn't do some of the things above right... either that, or they are simply too bored to pay attention to your talk... it happens...