Roadmap Research area and example thesis topic What

Roadmap § Research area and example thesis topic § What to expect § Factors for successful research § Financial aid Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 1

Why do you do research? § Do you feel that many technologies in our society can be improved by your smart ideas? § Doing research gives you the opportunity to impact the industry and change the world § Doing research gives you the opportunity to have other researchers follow your great ideas, and amplify the influence of your ideas § Overall, you will be able to make impact § Are you curious about the real answers to various open research questions today? § Doing research gives you the opportunity to satisfy such fundamental curiosity as a human being, and to expand the knowledge base of the mankind Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 2

Why do research in my area? § My research area: Distributed computing, distributed algorithms, and applied algorithms in networking. § Overall, focus is on applied algorithmic issues arising from real systems. § “Algorithmic issues” means my research has a lot to do with algorithms. “Applied” means that those algorithmic issues must be well motivated/justified by real systems. § Doing research in such a research area gives a great opportunity to make impact and at the same time, to satisfy your curiosity § Future computer systems (or computational devices) will almost always be in networks (e. g. , Internet of Things) § These systems pose new challenges to my research area, and introduce numerous exciting open research questions Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 3

An example thesis topic of one of my current Ph. D. students (Zhao Yuda) § We want to monitor the possibility of landslide after a large earthquake, using wireless sensors § See http: //www. winetsrl. com/landslide_e. html for an example commercial system doing this § To do so, we can drop many wireless sensors into a certain region from the airplane, to form a wireless sensor network § See http: //en. wikipedia. org/wiki/Wireless_sensor_network for what is a wireless sensor network § Each sensor reports a motion reading, and we want to know the average motion reading in a certain area to monitor possible landslide Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 4

An example thesis topic of one of my current Ph. D. students (Zhao Yuda) § Let’s re-cast the earlier setting into a research problem… § Goal: Total N nodes. Each node has a binary value, root node wants to compute the sum of all values § This is the base case -- Once we can do this, we will be able to easily compute the average among all values, where each node has a floating-point value § Performance measure: We want to minimize the number of bits sent by each node when computing the sum root 1 0 1 1 1 Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 5

Example thesis topic: Existing approaches for computing sum § Approach 1: Send/relay all N values to the root § A node near the root may need to send/forward over N bits § Approach 2: Use a spanning tree and do aggregation § Illustrated in the following animation § Each node sends at most O(log. N) bits sum = 0+3+2 =5 1 root 0 1 1 2 1 3 0 0 1 1 1 Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 6

Example thesis topic: Failures make things interesting… § Tree-based aggregation cannot tolerate failures sum=0+3=3 failed 1 X root wrong result! 1 1 3 0 0 § 0 1 1 1 Can you design an efficient protocol for computing Sum so that each node only sends O(log. N) bits and can tolerate arbitrary failures? Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 7

Example thesis topic: So what is thesis topic? § Design better fault-tolerant protocols for computing Sum § Prove lower bounds on the number of bits that a node needs to send § In case you are curious § We have proved that to compute Sum in a fault-tolerant way, some node will need to send § So it is impossible to design a fault-tolerant protocol where each node only sends O(log. N) bits… Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 8

Roadmap § Research area and example thesis topic § What to expect § Factors for successful research § Financial aid Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 9

Style of Research in Our Group § For our research, novelty is more important than sheer amount of engineering § My group works in a smart and creative way § For our research, 1 hour of creative thinking can worth 10 hours of drilling § Your creative thinking will be a key asset in my group Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 10

What to Expect from Me § I usually work with my students on research problems for which we already have unique insights and clear advantages over other research groups in other universities § I do research TOGETHER with the student, instead of just supervising them § I spend a lot of time with my students, prove many theorems together, solve hard steps if they cannot solve those, and write our papers together § Of course, if the student prefers to do most the work alone, I won’t disturb him/her either § The above two factors ensure that my students have high success rate of solving the research problems § So far (at the time of my writing), the vast majority of my papers get accepted upon the very first time we submit them for publication Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 11

What I Offer to Students § Opportunity to do cutting edge research and achieve results that nobody in the world (not just in Singapore or Asia) has achieved § Example 1: My students and I proposed a new concept of Error Estimating Coding (EEC) for wireless networks. EEC offers a new functionality/overhead tradeoff that is different from the classic error correcting codes (which had been around for more than half a century). The work won the Best Paper Award in SIGCOMM 2010. SIGCOMM is the top conference in networking. This was the first time that this prestigious award was given to researchers outside of the United States. § Example 2: My students and I proved strong results on fault-tolerant community complexity of certain aggregate functions. Our results close an open question over the past decade, and at the same time, open up a new topic ripe with future research opportunities. The work was published in JACM is often considered as THE TOP publication venue on principles of Computer Science. JACM explicitly states that it publishes “the most significant work on principles of computer science”, and only accepts papers that are “among the best papers of the year in its area”. Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 12

What I Offer to Students § As evidenced by our top-tier publications, the top-notch research opportunities you get here will be at the same level as the opportunities that you would get in any top university that you can think of in the world. § In fact, our ultimate goal is for our work to be pioneering and to lead the way for other universities and research groups in the world. § Finally, from a practical perspective, you should understand that when you are looking for a research job, it is your CV (especially your research/publication record) that determines your competitiveness against other people – The Ph. D. diploma, purely by itself, has almost no value. Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 13

Roadmap § Research area and example thesis topic § What to expect § Factors for successful research § Financial aid Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 14

What makes you do successful research here § Factors for you doing successful research here § Your logical and creative thinking § Your strong background in mathematics – this is the foundation of all natural sciences § Your strong background in algorithms § General factors § Have strong motivation and dedication to do cutting edge research § Be mature, honest, and responsible (my group has a low tolerance threshold for dishonest or irresponsible behavior, especially related to academic and research integrity) § I can take students with an undergraduate degree in mathematics or physics as well Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 15

Rough Gauge for Students from China § If you are from China, following are some more specific measures for you to judge whether you are “strong in mathematics”. § Sufficient condition #1: Your math background will be sufficiently strong if back in high school, you were able to get 全国高中数学联赛 一等奖 (provincial level) or 全国高中物理学联赛一等奖 (provincial level) or 全国高中信息学联赛一等奖 (provincial level) § Sufficient condition #2: You have got prizes at similar levels as the prizes in #1. § Sufficient condition #3: You feel that you would be able to get those prizes, if you had tried hard enough and if you had got good training from your high school on these competitions. § Sufficient condition #4: You feel that your mathematics ability is comparable to those people who have those prizes (e. g. , by comparing with your classmates). § If you are not sure, you can always come and ask me. Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 16

Financial Aid from the University (At the time of my writing) § Research Scholarship (http: //www. comp. nus. edu. sg/graduates/scholarship. html): § “The Research Scholarship will carry monthly emoluments, plus full tuition fee subsidy”. “Each scholarship holder will receive a non-taxable monthly allowance of S$2, 000 (for international students) and S$2, 500 (Singapore Citizens). Graduate students who passed their Ph. D Qualifying Examination will be awarded an additional top-up of S$500 per month (for two years) during their Ph. D candidature. ” § There are other scholarships and fellowships (http: //www. nus. edu. sg/admissions/graduate-studies/scholarshipspgf. php and http: //www. nus. edu. sg/ngs/NGSS. html) with month stipend up to S$3, 200 Associate Professor YU Haifeng (http: //www. comp. nus. edu. sg/~yuhf/) 17