A. Teaching

A.1       Teaching Philosophy

I wrote the below as part of my teaching interests when I first sought a faculty position six years ago, to express my teaching philosophy then:

 

“When I first volunteered to teach our department’s Data Structures in C course, my advisor said that she was sure that I would be a good teacher since I explained concepts well and that I was diplomatic, but I was not as convinced. At the time, I knew that I could tutor or work with people one−on−one from my experiences in tutoring peers in high school and during college, but teaching was a different sort of ball game. In many aspects, I was right, it was different, but I still like to think of teaching as tutoring many people simultaneously.

 

I think this mindset is important because tutoring is done mostly on a one−on−one basis. This atmosphere allows the tutor to establish rapport with the student and to build a common experience. Teaching is no different. As a teacher, I have to motivate students to participate in class by asking them interesting questions, and to encourage them to ask me questions.

 

After considerably more teaching experience, I still maintain this statement as a central tenant: that teaching needs to be done one-to-one, in a convivial relationship akin to peers conversing.  The final clause is particularly important in NUS’ context, as much primary and secondary education has the instructor broadcasting knowledge to the students: a one-way channel.  Incoming students may expect this method, but it is detrimental to achieving the learning outcomes that we wish our graduates to possess: independent inquiry, critical thinking and communication skills. 

 

So how can this student-teaching bias be undone?  I ask my students to call me by first name and of course, this means I try to learn their first names as well.  This is to lessen the formality and distance between us. When time allows, I step through proofs and complex math by asking members of the class how to derive individual steps.  This helps to bring the class out of monotony, energize the students’ minds while ensuring the experience is more akin to one-on-one tutorial.  When I answer questions incorrectly, I acknowledge this publicly and try to correct the misunderstanding when I’m clear of my doubts.

 

I try to structure my lectures so that students can get a sense of what to pay attention to.  I explicitly signal this by connecting lectures together, referencing ideas and aims within and across different lecture sessions.  I follow this up, by closing each session of my graduate courses with issues that remain unsolved in the literature, providing a roadmap for the challenges ahead, and get them thinking about the course materials in the larger context (see Supplementary Teaching Portfolio, Appendix F), which are often questions that I don’t have an answer to.  These provide the basis for thinking of the course materials in relation to other real-life aspects.  They also provide argumentative practice for the student to practice for the final exams, which (even in the undergraduate courses) are not about rote learning.  I have gotten feedback from my exam moderators as well as students taking the class that validate these claims.  While the above remain techniques, the principle is to let each student know that you are a guide to self-learning and a role model, rather than an instructor.

 

Of course, as the teacher, I have the responsibility to ensure that the pedagogical goals of each module are achieved by the students.  However, each learner learns in different ways, and while the goals may be the same, there are different routes to it.  Gilligan (1987) spoke of using the double bind in teaching – where the teacher presents two or more alternatives to students, which all lead to the same desired learning outcome.  I have practiced this when possible, giving different options for a single assignment (hands-on vs. theoretical; or language vs. vision vs. robotics; or individual vs. team project; see Supplementary Teaching Portfolio, Appendix C) for which the student can choose on their preferences, but for which the underlying knowledge learned is the same.  I also put the Pygmalion effect (Tauber, 1998) to work in my classes, by telling my classroom students and PhD advisees that I am challenging them, one-on-one, with assignments as tough as those faced students at MIT, CMU or Berkeley, and that they can do them well with the required effort.  They responded very well; in some courses I have raised the difficulty and expectation of the assignments while boosting the scores in my student feedback compared to previous lecturers.  

 

I want to close with a final note on the role of technology in teaching.  Li and Bernoff (2008) dissect how Web 2.0 technology impact relationships and communication for businesses.  While seemingly unrelated, they understand the key to their value: “First, the technologies change rapidly. And secondly, the technologies are not the point.  The forces at work are.  … Concentrate on the relationships, not the technology.”  That is to say, technology is a relationship enabler.  Understanding what each technology’s characteristic is important in enabling a one-to-one tutoring relationship.  For example, I favor forums and wikis over blogs and announcements for interaction – as in the former two my interactions resembles students’, but in the latter two, the media acts as a broadcast, reinforcing a hierarchical relationship.  To echo my first statement, I find teaching works best when I can interact with each student individually, at their own level, on their ground.  When technology can serve to better these individual student relationships, then I have a reason to adopt it.

A.2       Teaching History

Teaching comes in many forms.  I have excelled at classroom teaching, and at direct supervision of both undergraduate and graduate students.  This has resulted in many students who have returned to me for supervision in projects and in graduate programme work.  I have already successfully finished mentoring three Ph.D. students, two of which who have gone to the U.S. to leading institutions to do research work.  I see all of these measures taken together as evidence of excellent teaching practice.

(a)                    Classroom Teaching

I have taught a good mix of courses at all levels of the spectrum (2 x 1000s, 1 x 3000, 3 x 5/6000s).  In the majority of my classroom teaching, I have exceeded the average performance of my peers within the department.  In the below tables, I show my average feedback scores on Question 8 (“Overall, the teacher is effective”) and the percentage of “5”s (excellent) rating I received per semester.  I also give the departmental average score for the same activity (where possible), so a fair comparison can be made.

 

Aside from these courses, I also assist in CS2305/6S, our Special Programme for talented students, in mentoring interested students in the methodologies of Information Retrieval (SIG.IR) for the past three years.


Modules Scheduled to Teach This Academic Year:

 

Modules Taught:

Class size: 56 students

Feedback Respondents: 30 students

Metric

My score

Dept. average for 3xxx modules

Average Q8

4.16

3.88

# of ‘5’ ratings for Q8

10 students (33%)

20.11%

Best Teacher Nominations

7 students

N/A

 

Class size: ~40 students (data not given by the system)

Feedback Respondents: 37 students

Metric

My score

Dept. average for 1xxx modules

Average Q8

3.97

4.075

# of ‘5’ ratings for Q8

6 students (16%)

29.6%

Best Teacher Nominations

11 students (29%)

N/A

 

Class size: 25 students

Feedback Respondents: 18 students

Metric

My score

Dept. average for 5xxx modules

Average Q8

4.28

4.06

# of ‘5’ ratings for Q8

8 students (44%)

25.6%

Best Teacher Nominations

6 students (33%)

N/A

 

Class size: 15 students

Feedback Respondents: 8 students

Metric

My score

Dept. average for 5xxx modules

Average Q8

4.35

4.03

# of ‘5’ ratings for Q8

3 students (37%)

31.2%

Best Teacher Nominations

4 students (50%)

N/A

 

Class size: 40 students

Feedback Respondents: 30 students

Metric

My score

Dept. average for 3xxx modules

Average Q8

4.13

3.72

# of ‘5’ ratings for Q8

9 students (30%)

13.7%

Best Teacher Nominations

8 students (26%)

N/A

 

Class size: 27 students

Feedback Respondents: 21 students

Metric

My score

Dept. average for 5xxx modules

Average Q8

4.19

4.17

# of ‘5’ ratings for Q8

8 students (38%)

36.7%

Best Teacher Nominations

7 students (33%)

N/A

 

Class size: 53 students

Feedback Respondents: 36 students

Metric

My score

Dept. average for 3xxx modules

Average Q8

4.22

3.65

# of ‘5’ ratings for Q8

15 students (41%)

16.7%

Best Teacher Nominations

17 students (47%)

N/A

 

Class size: 30 students

Feedback Respondents: 23 students

Metric

My score

Dept. average across all modules

Average Q8

4.13

3.75

# of ‘5’ ratings for Q8

5 students (21%)

29.2%

Best Teacher Nominations

9 students (39%)

N/A

 

Class size: students N/A – not given by system           

Feedback Respondents: 154 students

Metric

My score

Dept avg over all modules

Average Q8

4.10

3.85

# of ‘5’ ratings for Q8

41 students (26%)

23.2%

Best Teacher Nominations

45 students (29%)

N/A

 

Class size: 15 students

Feedback Respondents: 15 students

Metric

My score

Dept avg over all modules

Average Q8

4.40

3.77

# of ‘5’ ratings for Q8

6 students (40%)

36.4%

Best Teacher Nominations

5 students (33%)

N/A

 

Class size: N/A - not given by system     

Feedback Respondents: 205 students

Metric

My score

Dept. avg over all modules

Average Q8

3.77

3.76

# of ‘5’ ratings for Q8

22 students (10%)

16.8%

Best Teacher Nominations

12 students (5%)

N/A

·               Response to Negative Feedback:

I still have much to learn about teaching from my students.  An analysis of feedback I have received reveals that although my teaching is good in general, I still need to improve on my control of the time during lecture.  When structuring materials for modules that are new to me, I tend to pack too much material during a single lecture.  This happens as I sometimes do not anticipate the amount of student interaction there are to some of my questions. 

 

Another point is in the quality of oral presentation: students have felt that my pace of lecture is too fast and that my volume is too low.  While I have done well so far, partially because I have an American English accent, I still need to improve on my voice control to ensure all students can hear me.  These problems have lessened in repeat teachings of the same course modules, partially perhaps due to smaller class sizes overall, but also due to the fact that experience helps me become more confident of the lecture materials, positively affecting both my timing and voice control.

 

I do expect a lot from students, and I believe that letting them bridge the gap between the theory in lecture and the practice embodied by the assignments is a purposeful way for them to learn.  In all of my classes, my assignments are really mini-projects that are open-ended and cannot be solved by simply following a recipe from lecture.  That said, I feel NUS students are very creative and do help each other with assignments when properly encouraged. I believe my teaching style encourages this form of higher order learning.  The downside is that students who may be overloaded from other classes do not anticipate the higher level of difficulty in the assignments.  For such assignments, I often give out a “baseline” solution that meets the minimal requirements of the assignment objectives, to help weaker students have a look at a working solution.)

·               Excerpted Comments (From Best Teacher Nominations):

-       This is my second class from this lecturer. My first was Digital Libraries. The reason i want to nominate him is that he explains concepts clearly and concisely. He articulates well, therefore its easier to catch what he means. Moreover, i can see the passion that he has in the subject matter. He is able to demonstrate his knowledge in the subject matter and more importantly create interest. Lastly, for every topic, he always leaves us with questions that has not yet been answered by the research community. This allows us, students to know where we can go and what is yet to be done. Sometimes, the questions are more important than the answers.

-       Min is a very patient tutor, who will always explains concepts which i'm not clear about. Besides that, he's very approachable, making it easier to seek consultation from him. I begin to understand why he likes to set tough assignments. It's through these assignments that we'll truely learn things from this course. The assignments given, are very challenging, but nonetheless interesting and fun, and very relevant to this course. Overall, Prof Kan had taught me well with respect to AI and he did arouse interest in me to pursue more on this area.

-       He is very creative man that doesn mind slogging out the saturday with us. He has his own monitor chart outside his door so that we know when we can find him and when we do not. So far, i haven seen any lecturers or tutors that allows us to know where he/she is so that we can approach him/her when we have problems. His way of doing up his presentation was very interesting too with his unique accent that caught our attention when we went for the very first lecture.

-       Class discussions are actively stimulated during lecture times. He is very approachable for consulation. Once, I missed a lecture because I was held back during an extended discussion with another lecturer. When I approached Dr. Kan the next day to discuss some stuff, he readily gave a concise summary of the lecture which I've missed. He used online forums to hold further discussions + monitor the forums regularly too. Gave us opportunity to practice writing a critical survey paper.

-        He is one of the nicest lecturers I've met so far in my 2 years university learning period. I nomite him for his professionalism and patience to us students. Another remarkable point is that he really THINKS how this module can improve, we can see that through the way he designs projects and so on. Thanks!

(b)                    Student Supervision

As an integral part of my teaching, I supervise both graduate and undergraduate students, mentoring them on a wide spectrum of research topics in the media area of computer science.  I have served as a mentor to 7 Ph.D. students, 3 of which have already finished and have gone on to find outstanding jobs that further their research career.  Among my colleagues, I have supervised an above average number of students at both undergraduate and graduate levels.

 

My first alumni chose topics more aligned to their co-supervisors and could not fulfilled research objectives in my own grants.   As such, I have been supervising Honors Year Projects (HYPs) as a means to get manpower for my research projects.  Currently, I have graduated 35 (shown below) HYPs/UROPs/interns and am currently supervising an additional 5.  This number is close to the maximum allowable number of undergraduate student supervision allowed by my department. 

 

I should state that mentoring so many students is an art.  It is by no means merely a way of recruiting manpower.  Students need to be taught the art of science and how to be critical and skeptical of their own findings.   I find this easiest to do by letting them drive our weekly meetings when possible (I meet all of my students an hour a week; usually amounting to two full days of my week committed to supervision: ~5 PhD students, ~5 undergraduate students), and playing devil’s advocate to question their results.

 

Running a larger group also has some unique teaching opportunities.  I run a monthly research group meeting with my students, where they get a chance to interact and question each other about their research.  Initially, this meeting was very awkward: a student would present their research, and since the other students were doing other unrelated research, they had nothing to contribute.  The meeting was widely regarded as a waste of time.  To remedy this, we brainstormed and decided to introduce some group activities.  These included tutorials on different technologies related to all students (distributed computing, current tools, resources and computing infrastructure, as well as critical paper reading sessions) that I supervised.  As a result, meetings have gotten livelier and I feel more camaraderie among the group members.  

 

Students also get a chance to peer teach in my group.  PhD students serve as mentors and another pair of eyes and ears for the undergraduate students, and hold their own independent meetings with the undergraduate students.  All students in the group have a chance to read and critique each other’s work at the thesis writing stage (cf communicating effectively as a learning outcome, “Teaching Philosophy”, A.1).  At times, they are resistant at first – “more work that’s unrelated to my topic” – but come around when they see how they can apply it to their own writing.

 

I also make sure that the students know what is going on at my end (what proposals and grants I’m working on, the status of my tenure application, what I’m teaching), because I want to treat them as a peer and fellow stakeholder (a reprise of tutoring one-on-one, “Teaching Philosophy”, see A.1) and also because I’m training my students to eventually serve in my role as an assistant professor.  As they see all the aspects that I’m involved in, and how I can manage it, they have a better understanding of their role in the academy and have can make a better informed decision of their job.

 

Careful mentoring of undergraduate students has served me well.  I believe that, due to this mentorship style in part, it has also served as an extremely good method for recruiting graduate students; all four of my Ph.D. newest students are returning NUS undergraduates.  I have thus used none of the new graduate student pool that have to NUS applied outside of Singapore, and freed these resources for other professors.

·               Graduate Research Students (To be supervised): 1

1.         Ng Jun Ping              Ph.D. (NUS undergrad, 1st class, applied for January 2009)
Topics in full document question answering

·               Graduate Research Students (Currently Supervising): 4

1.         Jesse Prabawa Gozali       Ph.D. (Year 2)
Human-Computer Interaction Issues in Digital Libraries

2.         Lin Ziheng                             Ph.D. (Year 2)
co-supervised w/ AP Ng Hwee Tou, Multidocument Summarization

3.         Zhao Jin                                Ph.D. (Year 3)
Equation processing and retrieval in Digital Libraries

4.         Tan Yee Fan                         Ph.D. (Year 4)
Disambiguation problems in Digital Libraries

·               Graduate Research Students Supervised (Alumni): 3

5.         Hendra Setiawan    Ph.D. (Submitted Thesis; Postdoctoral researcher at Univ. of Maryland, College Park)
co-supervised w/ I2R, Dr Li Haizhou, Phrase-based Machine Translation

6.         Qiu Long                   Ph.D. (Submitted Thesis; Researcher at the Institute of Infocomm Research),
co-supervised w/ Prof. Chua Tat-Seng, Scenario Template Generation

7.         Cui Hang                   Ph.D. (Graduated; Senior Relevance Scientist, Yahoo! Engineering, won Best Thesis Award)
co-supervised w/ Prof. Chua Tat-Seng, Soft Patterns for Question Answering

·               Undergraduate / Interns / Honors Students Supervised

As stated above, many of my undergraduate supervisees have returned to rejoin me in supervision in some form (as a PhD student, or re-taking undergraduate project supervision with me.  This is not because I ask them to; on the contrary, I often explicitly ask them to consider other professors so that they can get a wider range of contacts and recommendation letters. 

 

Details about their current affiliations can be found on my group’s website at http://wing.comp.nus.edu.sg/portal/contentview/15/29.   In the table below, ^ refers to UROP, * refers to HYP, + refers to intern, # refers to non-credit projects and bold refers to students who returned to do RA or Ph.D.s with me.

 

Abhishek Arora+

Edwin Lee*

Zhicheng Liu#

 

Emma Thuy Dung Nguyen*

Thien An Vo^

 

Ezekiel Eugene Ephraim*

Chee How Lee*

Wei Lu*

 

Viet Bang Nguyen*

Tinh Ky Vu^*

 

Jesse Prabawa Gozali*

Malcolm Lee*

Anubhav Madan^*

Vasesht Rao*

Litan Wang^

Yijue How*

Mingfeng Lee*

Alex Ng*

Siru Tan*

Fei Wang^*

Kalpana Kumar^*

Thiam Chye Lee*

Meichan Ng*

Yee Fan Tan*

Xuan Wang^*

Sandra Lai*

Guo Min Liew*

Yong Kiat Ng*

Yung Kiat Teo*

Yue Wang*

Aik Miang Lau*

Ziheng Lin^*

Hoang Oanh Thi Nguyen^

Hoang Minh Trinh*

Jin Zhao*

·               Participation in Theses and Oral Examination Committees:

I have participated in 33 GP students’ defenses and examinations, leading to 10 MScs and 6 PhDs.  Of these, so far only five of my own students are included on this list; I service about ((33 students / 5 of my own students) / 2 reviewers per student) = 3 times more students than I send out to other examiners.  The list below is sorted in reverse chronological order of first examination.  My own student’s names are in bold.

 

2008

            Sun Jun                     GRP (3 Jun 2008)

            Wang Kai                  GRP (23 May 2008)

2007

            Wang Xiangyu         GRP (15 Nov 2007)

            Zhao Jin                    GRP (14 Nov 2007)

            Zhong Zhi                  GRP (2 Nov 2007)

            Zhang Dongxiang   GRP (30 Oct 2007)

            Wang Wenting         GRP (29 Oct 2007)

            Derry Tanti Wijaya   MSc Defense (26 July 2008)

                                                GRP (25 Oct 2007)

            Upali Kohomban     PhD Defense (30 Jan 2007)

            Ng Hong I                  Pre-submission PhD Defense (11 Jan 2007)

2006

            Wang Dong              MSc Defense (1 Aug 2007)

                                                GRP (13 Nov 2006)

            Chen Jinxiu              PhD Defense (13 Oct 2006)

            Tan Yee Fan             Proposal (23 Oct 2007)

                                                GRP (23 May 2006)

            Hendra Setiawan    Pre-submission PhD Thesis Defense (14 Aug 2008)

                                                Thesis Proposal (19 Jan 2006)

2005

            William Ku                 Thesis Proposal (17 Nov 2005)

            Sun Renxu                MSc Defense (11 Jul 2006)

                                                GRP (27 Oct 2005)

            Karthikeyan Vaiapury         MSC Defense (7 Feb 2007)

                                                            GRP, 2nd Attempt (25 Apr 2006)

                                                            GRP (26 Oct 2005)

            Li Linlin                      Thesis Proposal (6 Feb 2007)

                                                GRP (25 May 2005)

            Steven Halim                        Thesis Proposal (15 Oct 2006)

                                                GRP (17 May 2005)

            Zhao Shanheng       Thesis Proposal (30 July 2007)

                                                GRP (10 May 2005)

            Low Jin Kiat              MSc Defense (25 Jan 2006)

                                                GRP (6 May 2005)

2004

            Lan Man                    PhD Thesis Defense (16 Apr 2007)

                                                Pre-submission PhD Thesis Defense (30 Oct 2006)

                                                Thesis Proposal (21 Jul 2005)

            Yin Hainan                MSc Defense (18 Apr 2005)

                                                GRP (16 Nov 2004)

            Wang Gang              Pre-submission PhD Thesis Defense (14 Aug 2008)

                                                Thesis Proposal (11 Aug 2005)

                                                Qualifying Exam (26 Apr 2004)

            Shen Dan                  MSc Defense (20 Apr 2004)

            Xiao Juan                  MSc Defense (29 July 2005)

                                                Qualifying Exam (8 Apr 2004)

            Chia Tee Kiah          Thesis Proposal (15 Oct 2006)

                                                Qualifying Exam (8 Apr 2004)

            Qiu Long                   Pre-submission Defense (28 Jul 2008)

                                                Thesis Proposal (10 Aug 2005)

                                                Qualifying Exam (6 Apr 2004)

2003

            Zhao Yunlong          PhD Defense (30 Oct 2003)

            Cui Hang                   PhD Defense (25 Apr 2006)

                                                Proposal (10 Aug 2004)

                                                Qualifying Exam (14 Aug 2003)

            Yang Xiaofeng         PhD Defense (9 Mar 2006)

                                                Pre-submission PhD Thesis Defense (21 Oct 2005)

                                                Proposal (10 Aug 2004)

                                                Qualifying Exam (11 Aug 2003)

            Chua Teng-Chwan MSc Defense (8 May 2003)

            Wu Wen                     MSc Defense (2003)

A.3       Teaching Performance Indicators

(a)                    Leadership roles in the development of modules

(b)                    Contributions to teaching materials

(c)                    Innovative Methodology

 

As all students are different and learn best in different ways, I coupled the use of the robots with other options for the same assignment.   Not all students learn best through hands-on application.  The students could instead do an assignment concerning computer vision or natural language processing (dealing with competencies in vision and language, as opposed motion, and translating them into programs).

 

 

 

 

To this end, I have opted to rely on student manpower to maintain my smallish computer cluster.  While most PIs in my department rely on Ph.D. students or paid central facilities to maintain their hardware, I employ motivated students to manage my systems for me.  This approach has its inherent risks; students sometimes can inadvertently do catastrophic damage to our machines.  However, by asking them to participate in this programme, they can get paid to self-train themselves independently in the real world (albeit on a small scale) helpdesk for my team.  They learn how to communicate and hand down knowledge about system administration while contributing to my group’s infrastructure – the same team and communication skills that represent higher order learning outcomes. “Senior” undergraduate system administrators also help out with purchasing and long term forecasting of the group’s needs.  I use their recommendations to decide where to spend my equipment grant monies and actualize their work to fulfill my group’s needs and these student’s confidence.

 

We maintain a spare machine in the group cluster for system admins to tune and try various set-ups.  In the upcoming future, I plan to network with Red Hat Linux (a provider of system administration certification) to see whether they would be willing to offer discounted or free examinations to students from this group.  These certificates often qualify the candidate for real systems administration positions, with potential earnings well over our average graduate.


A.4       Future Plans

As far as course teaching, I plan on further broadening the scope of my module teaching on the undergraduate levels.  There is nothing better to motivate one to appreciate interdisciplinary research than to teach it!  I may petition to teach a core area of my research, Natural Language Processing (CS 4248), or Human Computer Interaction (CS 3240).  Also, I still do have a lot of fine tuning in my teaching as evidenced by the feedback from students (see “Responses to Negative Feedback”, A.2.a), in making sure that the lecture end on time.  By planning and prioritizing the lecture notes more effectively, I’m confident that I can improve on these attention areas.

 

As far as advising, I plan to have my Ph.D. students and future postdocs co-supervise undergraduates with me; this will free up a nominal amount of time; but more crucially, will allow such senior students to experience supervision and responsibility.  As such, both the supervisor and the supervisee will benefit. 

 

A key difference in my advising trend is to contact the best undergraduate students early on and attract them to my research program through regular advising.  To this end, I have also supervised undergraduate students in the Special Programme (for gifted 1st and 2nd year students).  This helps to build early training for students in my area, of which a fraction will come to my group for undergraduate projects, and a fraction of those will return for graduate level studies.  This builds a steady pipeline of students that have strong camaraderie with my group, complete with prior research experience. This in turn helps the department as well, as my presence does not eat in to the general Graduate Programme pool of students that can be slated for other PIs.  As stated earlier, all four of my current Ph.D. students are graduates of our own SoC undergraduate programme.