As a world-leading data science team at the Advanced Analytics Institute (AAI), Faculty of Engineering and Information Technology, we motivate ourselves for quality and highly collaborative research through a team approach. This has been fostering a quality and impact-oriented, vibrant research culture and environment, which result in many successes in our paper submissions to publications in top-tier venues. A recent success example is the acceptance of four papers and one tutorial in data science by the most prestigious artificial intelligence conference, The Thirty-Second AAAI Conference on Artificial Intelligence (AAAI-18), for which, as shown below, all the lead authors are PhD students in AAI with supervision of Professor Longbing Cao. Here, we would like to share the story of our success, with the focus on a driving factor –teamwork.

• Trong Dinh Thac Do, Longbing Cao. Coupled Poisson Factorization Integrated with User/Item Metadata for Modeling Popular and Sparse Ratings in Scalable Recommendation. AAAI-18, New Orleans, Louisiana, USA.

• Guansong Pang, Longbing Cao, Ling Chen, Defu Lian and Huan Liu. Sparse Modeling-based Sequential Ensemble Learning for Effective Outlier Detection in High-dimensional Numeric Data. AAAI-18, New Orleans, Louisiana, USA.

• Shoujin Wang, Liang Hu, Longbing Cao, Xiaoshui Huang, Defu Lian and Wei Liu. Attention-based Transactional Context Embedding for Next-Item Recommendation. AAAI-18, New Orleans, Louisiana, USA.

• Songlei Jian, Liang Hu, Longbing Cao, and Kai Lu. Metric-based Auto-Instructor for Learning Mixed Data Representation. AAAI-18, New Orleans, Louisiana, USA.

• Liang Hu, Longbing Cao, Jian Cao, Songlei Jian. When Advanced Machine Learning Meets Intelligent Recommender Systems, AAAI-18 Tutorial, New Orleans, Louisiana, USA.

The lead authors of our AAAI-18 accepted papers. From the left to the right, they are Thac, Songlei, Shoujin, and Guansong.

Our teamwork spirit is rooted in all steps of a research process, including research planning, formulation of research designs, paper writing, and research presentation. All these teamworks are mainly performed in our weekly research seminars, in which HDR students take turns to present their research progress on Thursday each week. Additionally, we also emphasize wide collaborations with internal and external researchers to deepen our research and increase our research impact.

Research Plans

At our team, research plans are extremely important, which guide our research throughout each semester. The research plans can be divided into short-term and long-term plans.

With short term plans, we have a board that reminds us the potential upcoming conferences with the deadlines on it.  The board is placed where everyone can see. This motivates us to ensure the deadlines at a daily basis. To prepare a submission to a high-quality conference (CORE A* conferences such as AAAI), we need plan it from three to six months ahead. Before the deadline, we all present the idea and our plan to team for discussion in the weekly research seminar. Frequently feedbacks from team make sure that the idea is on the right direction and on time. Hence, team members rarely give up an idea and it is even extremely rare that someone cannot finish the paper.

With long term plans, we prepare for research reflections in our weekly seminars after each short-term period and semester. The reflections help us to look back what we have done in the last period. In the meeting, we have a chance to talk about what we have done and what we can learn from that. Every team member then discusses about how to make it better in a constructive way. It ensures everyone has a better plan for the next periods and help him/her with the future research direction.

Snapshots of the Gantt chart-based research planning

Formulation of Our Designs

After the research plans, we then focus on the formulation of our research designs, which include the designs of our algorithms, theoretical justifications, and empirical justifications.

Each HDR student first presents every detail of his/her algorithmic idea for each piece of work. The team member then attempts to quickly understand the motivation and novelty of the proposed idea and raise critical comments. This process is challenging to our supervisor and team members, since each of us focuses on very different research areas. However, we work hard to help each other. The presenter is, therefore, expected to receive a large number of constructive comments from diverse perspectives. Those comments often help the presenter significantly refine the research design. The refined design will be presented next time and receive comments for further refinements in our weekly research seminars. In the later stages of this process, the discussions will focus more on the theoretical and empirical justification of the proposed design. This process is iteratively performed until the design, the theoretical and empirical results are thought to be ready for publications in top-tier venues. By doing this, we contribute ourselves to enhancing our peers' research, and in turn, the above process also substantially trains our thinking and improve our critical thinking ability.

Extensive discussions for sharpening our designs

Paper Writing

Once the research design and our justifications are done, we will mainly focus on the writing work. A strong training culture on academic paper writing has been built in our team, which can be divided into supervisor-training, self-training and team-training.

The supervisor-training is mainly given by our supervisor. First, he has given a seminar on paper writing when most of us are in the junior stage, in which he comprehensively demonstrated how to organize a good research paper from both the structure and the content perspectives.

The self-training and team-training are mainly conducted by ourselves and other team members respectively. We trained ourselves in the form of self-revision, self-reflection and so on. Each draft may need to be modified multiple times by ourselves before sending it to our supervisor for polishing. Further, we have built closely collaboration within our team. Once one's draft is ready, he/she delivers it to the team in our weekly seminar for pre-review to seek suggestions to further enhancing the submission in a quite different perspective.

More importantly, a reflection on the paper writing is done for each member in our team after each submission. We summarize how the submission could be further improved and what we have learnt from the submission.  Also, a typical exercise is to compare the very first version to the last version of a paper to think why it should be improved in this way. 

When all members contribute to the discussion

Presentation of Our Work

After the acceptance of the papers, we need to attend the conference and do an oral presentation in front of other researchers in this area. The ability of giving effective presentations is important for us to communicate with other researcher and spread our ideas to the world.  Our weekly seminars offer us very valuable chances to learn presentation skills and build our scientific oral communication ability.

Every Thursday, all our team members will attend the weekly meeting and half of us will do weekly report. Each report focuses on research progress in the previous two weeks, e.g., reading papers, new ideas, theoretical or empirical evaluation results, or writing papers. The presenters need to carefully prepare their report slides to make their reports as self-contained as possible. During our seminar, the team members not only give suggestions about the content of the report, but also comment on the presentation quality, such as the slides arrangement, the grammar, and the talking speed. Frequently presenting our work at the research seminars offers great opportunities to improve our presentation ability, such as how to engage with the audiences properly with questions, eye contacts and body languages. Besides, we also have rehearsal talks before attending the conference. Thanks to this practice, we can communicate our work effectively and give fluent talks in top conferences. 

Our team member, Thac, is presenting his work

Internal and External Collaboration

In addition to our weekly seminars, we have wide collaborations with Australian and international researchers. For the example of our AAAI-18 papers, we collaborate with researchers affiliated with Center of Artificial Intelligence of UTS (e.g., Dr. Ling Chen), Arizona State University (e.g., Professor Huan Liu), University of Electronic Science and Technology of China (e.g., Dr. Defu Lian), and National University of Defense Technology (e.g., Professor Kai Lu). Such collaborations help broaden our research areas and stimulate our seminar discussions.