Graduate Student Survival Guide Janardhan Rao Doppa School of EECS, Oregon State University doppa@eecs.oregonstate.edu http://web.engr.oregonstate.edu/~doppa 1 Disclaimer • This guide is based on what I learnt myself in the past 4 yrs from – reading, talking to people and my personal experience. • These are general guidelines. Please follow what works best for you – personalize it! • Take this guide with a pinch of salt – some bias! 2 School requirements vs. Your education • “Never let your schooling interfere with your education” – Mark Twain • Classes and program requirements are important, but what is more important is to become an expert in your area. • Optimal policy depends on your reward function quickly finish PhD and get a high paying job Take some more time to produce a high-quality thesis … 3 Taking classes • Take fewer(no more than two) classes per term and spread out your classes over yrs you will have more time to do research work • Prioritize your classes based on which courses will be useful for your research when they are offered and their frequency flexibility for special topics courses 4 Getting started: “exploration” phase • Explore different topics in your area of interest Talk to senior PhD students and learn about their research Attend reading groups, seminars and colloquium talks Attend project meetings of different professors Read recent publications of professors Talk to professors whose work interests you Use the first year(GTA period) as a buffer time for your exploration 5 Choosing your Advisor • Things you should look for mentoring skills – talk to his students “active” researcher – look at his publications funding – current projects at hand time for each student – depends on the number of students, his academic activities reviewing, editorship, conference/workshop organization and research fame! 6 Choosing your Advisor • Young Asst. Professors are dynamic and have lots of new ideas more hands-on and work closely with students sometimes can be impatient – remember they are also under pressure, i.e., tenure period • Take home message “win-win” situation for both -- If you can work hard and keep up with the pace, you can be highly productive! 7 Choosing your Advisor • Senior professors are experienced and leaders in the field most of them are very good mentors have the luxury to be relatively more patient slight advantage in terms of advertising your work and job prospects – connections acquired over years! mostly have less time for students – remember they are busy with reviewing, editorship, travel for PI meetings, conferences etc. • Take home message If you can work more independently, you can reap huge rewards in the end! 8 Skills needed to Succeed • Reading and understanding ability Grad school requires lot of reading Read and understand ideas -- quicker the better Critically analyze ideas, arguments and assumptions, and being constructive at the same time! Grows your bag of tricks: may not see immediate benefit, but helps in the long run 9 Skills needed to Succeed • Technical writing its completely different! start working on it from your first year, e.g., course project and research progress reports articulating your ideas and presenting them as a coherent story sell your ideas – conference papers accept/reject writing good papers comes from practice – learn by imitating good writers 10 Skills needed to Succeed • Presentation skills Making good slides – start working on your power point skills Giving good talks – advertisement to read your paper! Comes from practice – learn by imitating good speakers International students should work on their speaking skills 11 Skills needed to Succeed • Social networking No. I’m not talking about Facebook or MySpace Meeting your peers and seniors at conferences, workshops and symposiums Talk about research problems Bounce ideas off each other Follow up after discussion through email and possible future collaboration Requires a lot of effort 12 Skills needed to Succeed • Leadership skills In research, teaching and academic service, e.g., volunteer activities, community building .. • Time management To maintain a balance between work and life! Leading a stress-free happy life Time Management talk by Randy Pausch • Patience Grad school is different from undergrad! Ideas may not work as you thought Repeated failures, e.g., paper rejections .. This skill will be helpful more generally in life as well 13 Skills needed to Succeed • Math skills Turning intuitions into formal arguments • Working in a team See the big picture! Be a good team player Learn from your team-mates, e.g., different views, different disciplines, other skills .. Accept others mistakes as yours Be generous about credit assignment 14 Skills needed to Succeed • Some software To manage emails, e.g., pine, thunderbird, outlook – Organize to save your time! To prepare manuscripts, e.g., LaTex editor like Texniccenter and WinEdit (windows), LyX and Kile (Linux) To draw your plots, e.g., gnuplot To develop rapid prototypes, e.g., Matlab or R 15 How to read a research paper ? • Goal: understand the scientific contributions • Read critically Is it solving the right problem ? Are there any simple solutions ? Reasonable assumptions ? Clear justification under the given assumptions ? Breaking points of the solution ? 16 How to read a research paper ? • Read constructively Reading critically is much easier – tearing something is easy than to build it up! Involves harder, more positive thinking What are the good ideas ? Do they have other applications or extensions ? Can they be generalized further ? 17 How to read a research paper ? • Make notes Highlight the important points Write additional comments on the hard copy Maintain summaries of all the papers you read Will be useful later, e.g., writing paper or thesis • See the big picture and connections Compare it to other related works How does it advance the field ? 18 What to read ? • Major conferences ICML, NIPS, AAAI, IJCAI (AI and ML) SIGCOMM, MOBICOM, Hot-Nets (Networks) • Major Journals JMLR, MLJ, JAIR, AIJ (AI and ML) • Tech reports from Research groups Follow other groups who are working in your area 19 How to write a research paper ? • Keep your audience in mind, e.g., conference or symposium or journal. • Write a good story Describe and motivate your problem with real-world examples (What) Why is it important ? Short-comings of previous methods How are your solving the problem ? (How) Justify why it solves the problem (Why) Theoretical proof or Experimental evidence Comparison with state-of-the-art Lessons learned from your work Don’t give away your future work !! 20 How to write a research paper: A recipe • Top-down approach Section level outline Sub-section level outline Paragraph level outline o More or less like a presentation with bulleted points o Check for flow of ideas o Think about the plots and tables you want to include o Consistent terminology and symbols • Get feedback from peers or advisor and refine 21 How to give a presentation ? • Keep your audience in mind • Remember, talk is an advertisement of the paper • Tell a good story Describe the problem Motivate through real-world examples Key intuition behind your solution High-level solution Briefly talk about main results Use figures as much as possible and avoid text! • Practice your talk 22 General Advice on Research • Choosing the right problems to work on Work on important fundamental problems Keep you busy for few years • Vision What should your PhD thesis look like ? Impactful research during tenure period ? You should know your goal and take small steps to reach it! • Build your reputation as a researcher • Read books similar to auto-biographies by great researchers The art of doing science and engineering: learning to learn -- Richard Hamming Fermat’s enigma: The epic quest to solve the world’s greatest mathematics problem 23 Opportunities after a PhD • Academic positions Post-doctoral researcher Look for those NSF funded CI Fellowships Tenure-track Asst. professor Non-tenure Research Asst. professor Faculty at a teaching university • Industry positions Researcher Research Engineer Engineer 24 Questions ?? 25