The Pacific Laboratory of Artificial Intelligence (PLAI) is dedicated to the pursuit of general purpose artificial intelligence. Technology-rich spin-off entrepreneurial activity is encouraged as a by-product.
We are first and foremost an academic lab. To fit in at PLAI your career goal prioritization should be to be, in this order, an academic, entrepreneur, or industrial research scientist. Deviations from this prioritization or having a different goal means that you should probably not pursue a PhD because it would not be in your best interest to do so. From experience, pursuing a PhD is a waste of your time and money if these are not your goals, and may well be even if they are. This is because there are more PhD graduates than academic jobs, competition is fierce, and the lost income you could have made over the course of pursuing a PhD can now reasonably be expected to add up to $1M. That said, PLAI alumni have done well on the academic job market and will continue to do well. What is more, and you can ask the students who have worked with me at Columbia, Oxford, and here at UBC, I devote a huge amount of my time and attention to helping ensure my students achieve succesful outcomes. Even if they sometimes reprioritize and end up taking a job in industry at the end.
You are expected to complete your PhD in 5 years plus or minus one year. In order to be competitive on the world job- and fund-raising markets you should have at least 5 accepted papers by the time you graduate. The acceptance rate at major machine learning conferences is approximately 20%. PLAI acceptance rates are significantly higher, especially so if you count resubmissions. Do the math. This means you should plan to submit somewhere between 5 and 25 papers during your PhD. Note that we simply do not submit papers to lower tier conferences and venues (AISTATS, NeurIPS, ICML, and ICLR are the preferred conference venues; UAI, IJCAI, AAAI, ECML are acceptable -- so too top tier programming language, vision, and NLP conferences). Journal submissions can be discussed on an ad hoc basis but, with few exceptions, such submissions will be extended length treatments of conference papers already accepted. I have a strong preference for journals that do not impose a paywall, for instance JMLR, TMLR, and FTML.
Note that there are relatively few deadlines throughout the year. Missing deadlines is a cardinal sin and a serious detriment to achieving your aims as a PhD student.
No paper shall be submitted to any conference, workshop, or journal unless a 5-page paper draft with a full experimental section including preliminary figures is submitted to me 2 weeks before the deadline. Only in exceptional circumstances will exceptions to this rule be granted.
The rules for arXiv paper submissions are the same as conference paper submissions. If the paper is not ready to be submitted to a conference it is not ready to be put on arXiv. If conferences allow simultaneous arXiv upload I encourage it.
My goals: achieve AI, do world-class research, generate societally noticable impact, generate sufficient funding to sustain and grow my laboratory, get a boatload of citations, make money on the side, and attract new and excellent students and postdoctoral researchers.
Your goal: get a PhD by working on research ideas, writing papers, getting them published, garnering attention, etc.
I cannot achieve my aims alone and have to work through my students and postdocs. You cannot achieve your aim without partnering with a faculty member, in your case, me.
My promise to you is that I will work very hard to get you to where you want to be in life, whatever that ends up being. This always means getting you to the forefront of our research discipline and helping you to craft a research story and vision that, in totality, yields a world-class dissertation-worthy body of work. What you pay for this is lost income and having to put my name on our work. It falls on both of us to ensure that this never becomes me putting my name on your work. This is subtle but very important. It requires some discipline from both of us. Particularly in the choosing of research topic. You will inevitably do the coding, the reading, writing, and, ultimately, the idea generation. It is extremely important that there is sufficient input from me, particularly on the research direction side of things, to ensure both my interest and my contribution to your work are worthy of us calling it ours.
The most important thing I can and will do for you is, at the end, write you a letter of recommendation and expose you to my network. You will not understand and may not believe the value of that until well after you have become much older. Academia is one part brutal meritocracy and one part brutal political game. Industrial research positions of the kind most want are arguably even more brutally competitive.
There is a group calendar which you are responsible for watching, adding and removing meetings from, etc.
There will be a weekly group meeting which will consist of paper-reading, guest lectures, practice talks, brainstorming, etc. You are required to participate in the weekly group meeting.
The agenda of our one hour weekly meeting is to be 5% administrivia and 95% current research progress. I should have access to a current project repository in which paper-formatted LaTeX notes can be found, compiled, and (by me) read (or overleaf). I will pull said repository either in advance or, more likely, during said meeting and read/skim it in order to catch myself up on and context switch to your work. The remainder of our meeting can and should be spent at a board or looking at code working through problems related to your research, transitioning (at about a year into your degree) from me teaching you to you teaching me.
I am casual in my manner and style and am more lenient than I should be. I like to walk around and talk. I like to be engaged all the time, being updated and kept in the loop even if I can't or don't respond. I have traditionally strongly supported the weakest members of my group and left the strongest to their own devices. Starting now I will spend more time on the projects I find interesting and prioritize working with the students who are doing, at that time, the most interesting and best work. That said, I work hard on behalf of all of my students and no one will be left behind.
- Not being in the lab every day.
- Taking holiday at inappropriate times.
- Not working hard.
- Eating noises (I have a disorder: mysophonia)
- Having a poor attitude.
- Not playing hard and having fun.
- Not doing world-class work.
- Holding meetings that have no research content.
- Playing politics.
- Not working well with others.
- Poor writing.
- Not contributing positively to the group dynamic.
- Not volunteering and leading group activities.
- Not leaving me enough time to edit papers.
- Playing copy-editor rather than content-editor.
- Not removing meetings from calendars that won't happen.
- Wasting my time.
- Not helping when asked.
- Being asked about exceptional holiday.
- You not putting your holiday in the calendar.
- Keeping non-standard working hours.
If these points are unclear or you deem them to be unreasonable please bring this to my attention.
- Submitting work with my name on it without proper notification.
- Lying.
- Unethical professional behavior (stealing ideas, irreproducible research, etc.).
- Anything that violates the positive space ideals below.
It is your responsibility to understand, track, and meet departmental guidelines and shepherd yourself through the necessary bureaucratic procedures. I am here to help you with your research and personal development. I am and will remain at best forgetful and more realistically ignorant of most departmental and university processes. I refer you to the CS department PhD handbook. Let this be your friend and guide.
Positive space is UBC's phrase for spaces that don't allow bullying, racism, sexism, classicism, etc. PLAI is one such space and I take the underlying intent and message of positive space very seriously. Within those bounds, however, the group dynamic reflects and encourages a robust sense of humor. I hate toxic forms of political correctness which have emerged that have nothing to do with a place being a positive space. If this phrase offends you I politely suggest that PLAI may not be a good fit for you culturally. And I swear more than I should. If swearing offends you please tell me and I will try not to in your presence. The group rule is now that I must do a push-up every time I swear. I have done fewer push-ups than I imagined I would.
- Can I go to ICML/ICLR/AISTATS/UAI? a. Do you have a paper accepted there? Yes. b. Do you not have a paper accepted there? No.
- Should I write a workshop paper and submit it to ICML/ICLR/AISTATS/UAI? a. Do you have a paper accepted there? Yes. b. Do you not have a paper accepted there? Maybe. This depends on my mood and the laboratory funding level. Tread cautiously.
- Can I do an internship? a. Yes. One. Make it a good one. And do it early, when you are less useful to me and the laboratory, rather than late in your PhD.
- What is your vacation policy. UBC allows 15 days. Your PhD is a job; treat it like one. I strongly discourage holiday before conferences and other major deadlines. I strongly encourage holiday after deadlines and after attending major conferences.
- Can I fail out of the PhD program? Yes. I do not believe everyone is cut out to get a PhD, nor do I believe that everyone should get a PhD who attempts to do so.