FAQ to my dear admission committee

FAQ (Keep Updating Now)

Hi, thank you for your interest in clicking this page. I’d like to share my answers to common/funny questions and try to let you know me better, beyond just standard materials/scores. I understand you may review a single applicant in few minutes, so as long as you’d like to read more, I am glad to share more.

Introduce Yourself (Short!)

I am an undergraduate student interested in interactions and networks. I have abundant research/work experience in recommendation systems, data mining, networks, LLM, and HCI. I also built projects utilizing my hardware and web development skills.

Fun Fact About My Name

If you speak Japanese, the best way to pronounce my name in my dialect is first to pronounce “新世界／しんせかい Shinseikai“ (which means new world) and then exchange the first two characters, “世新界/Seishinkai.” Yeah, you did it!

Otherwise, say “Shin Jay” is also close enough for my first name in Chinese.

You can also call me Pablo. I use this English name from my favorite poet, Pablo Neruda. If you called me by this name, I could know you read this page!

Future Plan / Why PhD?

My ultimate goal is to be a good professor for students. I do enjoy teaching and expressing things in simple and vivid ways, like 3Blue1Brown. I also love the working environment of a professor, which often engages discussion with colleagues and students and demands for industry (and seems more free than working in industry). From another perspective, I wish I could use funding to fulfill my and my friends’/students’ curiosity.

I wish you could let me know if these thoughts above are different from your real situation. Appreciate! Click here.

I do think I seemed to own some gift like a vendor or consultant. My friends working as VCs, patent agents, and company CEOs actually believe I will be a great entrepreneur and state their willingness to invest in me. Flattered, wish to meet their expectation. I will stay prepared for possible opportunities.

In my SOP, and here, I’d like to stress my interest in interactions and relations, and the prediction and adaptation of the upcoming society will consist of comparable human and AI agents (LLM agents). But as you can see, my ultimate goal is to be a good teaching professor. I always remain passionate about teaching and education. Accessible to everyone, abundant, and intriguing, tailored education is my dream. I learn most knowledge from the internet or public community as well, and I always want to return. The AI Association I built in my university and the LLMQuant community are both my practice, while I’d like to do more. But I do not have “serious” experience in education, and I don’t think just saying I am passionate in my SOP works anyhow, so I didn’t include this ambition about education in my material.

Recent Ideas?

For networks, I am considering a new task in source tracking. Given a 1~t temporal graph where information is propagating, can we train a model to take the last t-n~t graphs to predict the source node(s) that is propagating in 1~t-n time frames? Besides from a recall task, we can also view this as a generative task. Could the same propagated networks be driven from different source combinations? What if built a condition loss on targeting generating “minimal energy” source node(s) combinations? This could work on news tracking and societal question answering. In my previous survey, I didn’t find similar tasks in the GNN era. But you can smell its meaning from the complex networks perspective. I have a primary model design and training plan.

I really want to share my recent ideas with you. But to avoid scooping, I’d like to invite you to email me, and then the full text will be delivered in reply.

For multimodal AI + HCI, I am considering using speed+text+emotion analysis model to build a real-time evaluation of talking, lecture. For example, this system could tell you if you didn’t stress enough or if two adjacent sentences are too jumpy in logic level.

For social simulation, I’d like to follow the paper NETEVOLVE, and further design is ..

For relational AI, …

Working Style?

1. I am extremely productive when I feel something is interesting. From self-estimation, I can lead, plan, and finish good research by myself. So it’s okay to be relatively hands-off. However, I always appreciate you are glad to meet or leave comments on my works.
2. Every time we meet (or I will update a tracking doc regularly), I will let you know my previous attempts (and their results) in systematic ways, for example, a table and notes about my summary insights and vivid comparison. I will also let you know my future to-do lists based on my current understanding. If needed, I will list high-level directions and long-term plans of work.
3. What I wish I could know from you is 1) any comments besides my current conclusion. 2) Any suggestions about my to-do lists 3) different understanding opinions 4) Something you think that may be related and intriguing. Again, it’s totally okay to be hands-off, but please do not acknowledge that it is natural since you’re “senior.”.
4. I am very interested in working form interdisciplinary background friends and hear about their views from different perspectives, and find them always helpful and intriguing. I think this hobby can help me understand ideas and things in a “smoother” space. I maintained this hobby not only for work and research but basically everything, maybe as certain reflection of curiosity as well.

Let me know if you may have further questions about this section.

My Lab Uses Mixed Methods to Research. Can You Fit That?

Sure! I am glad to develop and use new skills, I am a good learner. As mentioned in my homepage, I never limited my self in certain skills or technology, as long as they can help us to achieve certain purpose. You can also see from my Projects section, I also have experience about hardware and web development.

We are Not Matched in Research Area, Sorry…

I understand that you may have a specific research area with funding to explore and may not have a direct match with my current research. However, as an undergraduate student, I believe my actual main job is to warm up or pre-train for potential research. Even though I only have publication in specific areas, I actually always keep eyes on border terms of AI and HCI and ready for them. You may find some clues of my diverse experience and my slogan “to get general insights and seek for interdisciplinary chance” in my Linkedin profile. Anyway, I am glad you have read until here, and I am always open to any suggestions or comments. Thank you!

Any Interesting Works Recently Read?

• NETEVOLVE: Social Network Forecasting using Multi-Agent Reinforcement Learning with Interpretable Features

• In-Context LoRA for Diffusion Transformers

• Exploring Collaboration Mechanisms for LLM Agents: A Social Psychology View

Other Offer?

I actually joined the early admission of CUHK and received an offer from Professor Weiyang Liu. However, with my consideration and Professor Liu’s suggestions, I rejected the offer and tried to apply to a more prestigious school, looking forward to an interdisciplinary and fantastic environment.

I am also grateful that Professor Yujun Yan from Dartmouth College has agreed to write me a strong recommendation letter, and offered me a position in her lab (return offer, yeah!).

Any Questions for Me/My Lab?

1. I’d like to know your advice style, your (lab’s) future direction, and how you help students onto the job market (applying funding, connecting with great researchers, etc.).

2. Sadly, I can’t drink coffee. Does your lab/department provide tea?

Any Belief of the Life?

1. It’s okay and even should be blindly optimistic in the beginning. And be cautious in the progress.

2. Experience is important. Data points build really insights and further ideas. So we should increase our experience. (One example is that I interned to experience four first-line city in China, LOL)

3. I’d like to stress that always looking at the acceleration rather than counting on how many things someone did.

   For example, peers or anyone may show great achievement that is even impossible to imagine. It’s usually greatly dependent on their strong background that may have no business with themselves. In other words, just like Bayes conditional probability, you should understand what exactly someone himself/herself credits, conditioned by their prior environment. And then, if needed, do fair comparison.

   But also understand that owning and such prior environment and resources are inherent advantages in comparison. I do not mean you should ignore or overlook them. But what I stress may help you to see some potential things.

   Hope I don’t bring you confusion :)

4. (I still get a lot to share, please go back to check)

What are you working on now?

I am preparing some slides for you to answer common questions and introduce myself (While I wish this page has already answered several of them). Meanwhile, I am writing one paper to SIGIR (Wish to see you at Italy) and learn leadership principles from the AWS Shanghai AI Lab and OpenSearch team now as an Applied Scientist intern. Amazing team, recommend! One picture here.

BTW, I am appreciated that my manager Charlie Yang and Lab Director Zheng Zhang are supportive of my PhD application and willing to provide recommendation letters!

My Questions Are Not Included!

I’d like to answer your question. I would appreciate it if you could let me know of these approaches, 1) fill this form 2) email me 3) click here to arrange a meeting. Let me know if you want to schedule any time that does not appear in the available selection in the above links.

Not Good Enough?

Check This Applicant Instead!

I’d like to recommend Chang Liu, a passionate and expert HCI student focused on understanding and leveraging the mutual influence between interaction contexts and user behavior. Please check his homepage for more information. From the perspective of a close friend, co-author, and collaborator, I can attest that Chang Liu consistently impresses with his resilience and dedication.