Lifetime
- When I die, I wish to have no desire to live longer or live an afterlife. I wish to have no regrets, having lived a full life.
- I want to give my best to the people around me.
- Love to my family, friends, and co-workers.
- I want to succeed, without leaving others behind.
- I wish to live three professional lives - as a researcher and inventor (age 30-50), as an investor (age 50-70), and as a writer and artist (age 70-90).
- Researcher and inventor (age 30-50)
- It is an exciting time with the emergence of computer technologies that automatise and speed up tedious tasks. I wish to ride this big tide and be part of the state of art.
- I desire to guide the field, bring new perspectives to the problems, and invent solutions.
- I want to learn how to turn technology into products that change people’s lives.
- I will get myself prepared for the next stage:
- Expertise in 3+ fields
- Leadership experience
- Business experience
- Collect a sufficient amount of personal equity
- Investor and advisor (age 50-70)
- Contribute to technology with knowledge, experience, and money.
- Address the most pressing problem for mankind.
- Advise and invest in companies to blossom their technology into profitable businesses.
- Writer and artist (age 70-90)
- I am intrigued by the kinds of experiences and feelings that are very unique to individuals and yet are shared by many others.
- I like sharing such abstract thoughts and emotions with others, possibly borrowing artistic forms like writing, music, and painting.
- I hope to share such work with generations ahead.
- Stay fit throughout.
20-year goal (2022-2042): researcher and inventor
- Researcher and inventor.
- Get AI to the next level.
- Learn to run multiple research projects in parallel, efficiently.
- Expertise in 3 main areas of science: machine learning + two others (possibly medical science and economics?).
- Education of PhD students and junior researchers.
- Experience with running a tech business: from technology to products.
- Reputation as a key contributor to ML and AI.
- Get ready to become a tech investor.
- Leadership experience (running an organisation).
- Business experience (how the world works).
- Social capital (researchers, policymakers, business people).
- Collect “sufficient amount” of personal equity.
- Stay fit.
5-year goal (2022-2027)
- Get AI to the next level: Scalable Trustworthy AI (STAI) @ University of Tübingen
- People started to care about the trustworthiness of AI.
- But the techniques are, by and large, being verified at a smaller scale.
- How to bridge the gap between the large-scale needs and the current trustworthiness technology?
- Learn to run multiple research projects in parallel, efficiently.
- 10 PhD students and 5 postdocs/research scientists.
- Education of PhD students and junior researchers.
- Automatisation and efficiency for organising the team.
- Build expertise in STAI.
- Start the new field and lead the field.
- Continuously follow up on the state of the art.
- Work closely with students and postdocs to write papers regularly.
- Teach a relevant course & seminar.
- Reputation as a key contributor to ML and AI.
- One oral paper per year.
- Leadership experience (running an organisation).
- Advising researchers out of my expertise.
- Nurturing leadership in others, so that eventually the organisation runs without me.
- Thorough knowledge of financing a research lab.
- Experience the holistic pipeline of turning investment into invention and marketable solutions.
- Making policies and resolving conflicts.
- Stay fit.
1-year goal (2024)
- Manage the Scalable Trustworthy AI (STAI) group.
- Recruit XX PhD students and XX postdoc.
- Manage group activities like reading groups, seminars, retreats, and social events.
- Make sure everyone is healthy.
- Papers
- Make 5 oral-quality submissions on STAI.
- Advising
- Keep the average time to first paper below 1 year, without any shortcuts like writing the whole paper for the students.
- 1 hour per week per student to think about their research & career.
- Business
- Funding for ResearchTrend.AI.
- Hire XX engineers.
- Attract XX users.
- PR and teaching
- Organise 2 international events (e.g. workshops) related to STAI.
- Share ideas for data-centric trustworthy ML through external activities (social media, talks, etc.).
- Baby
- Stay fit: Walk every day (2 hours/day) and start playing Tennis.
2023 goals
- Manage the Scalable Trustworthy AI (STAI) group.
- Recruit 1 PhD student and 1 postdoc.
- Failed to recruit postdoc for the STAI group.
- Manage group activities like reading groups, seminars, retreats, and social events.
- We had reading groups, retreats, and some social events.
- Introduce a tie between STAI group and the Naver project.
- Not achieved yet. Thinking about possibilities to let Naver be involved in Tübingen.
- Recruit 1 PhD student and 1 postdoc.
- Papers
- Make 5 oral-quality submissions on STAI.
- This was an ambitious goal. I'd score 3/5.
- 2 groundbreaking research outputs from the Naver project.
- There was one cool research output. Score: 1/5.
- Make 5 oral-quality submissions on STAI.
- Advising
- Keep the average time to first paper below 1 year, without any shortcuts like writing the whole paper for the students.
- Not many samples to take the average yet. Will maintain this goal next year.
- 1 hour per week per student to think about their research & career.
- Not achieved.
- Keep the average time to first paper below 1 year, without any shortcuts like writing the whole paper for the students.
- PR and teaching
- Give 5 invited talks on STAI.
- Done.
- Organise 2 international events (e.g. workshops) related to STAI.
- Not done at all.
- Keep teaching a course on "Trustworthy ML".
- Give 5 invited talks on STAI.
- Stay fit: Run (2/week) and do morning routines (every day).
- 50% done. Was difficult to keep doing this all the time.
- Baby: was not part of the goals, but Shana took the majority of my daytime, especially between April and October 2023.
2022 goals
- Build a lab on Scalable Trustworthy AI (STAI).
- Recruit 2 PhD students and 1 postdoc (research scientist).
- 3 PhD students. Found a research scientist collaborator for the Naver project.
- Build automatisation tools and systems for the research lab.
- Notion page, web page, Slack, ...
- Secure funding for the large-scale trustworthy AI from Naver.
- Funding secured for the Naver project.
- Recruit 2 PhD students and 1 postdoc (research scientist).
- Papers
- Make 3 key submissions on STAI with existing collaborators.
- There was only one key submission.
- One oral paper at a top-tier conference.
- No oral paper. There's not so much one could do about this. What's important is to publish papers that are above my own research criteria.
- Make 3 key submissions on STAI with existing collaborators.
- Expertise on STAI
- Do not let the reading list queue beyond 20 items.
- It went far beyond 20 items. Not a good measure though.
- Teach a course on "Trustworthy ML".
- Do not let the reading list queue beyond 20 items.
- Advising
- 2 hours per week to talk to each student.
- 1 hour is often sufficient for individual meetings. This is also fulfilled through group meetings and lunch. Setting a quantitative goal is perhaps not a good idea.
- 1 hour per week per student to think about their research & career.
- Often fulfilled through reading recent papers on their topics. I feel the need to spend more time on this.
- Get the students and postdoc ready to write a paper towards ICCV 2023, ICML 2023, or the likes.
- This is hard. I'm studying the best way to speed up the time-to-first-paper as well as the actual acquisition of research skills that are going to be helpful for efficient research beyond the first paper.
- 2 hours per week to talk to each student.
- Stay fit: Tennis (1 hour x 2 / week), Run (5 km in 30 min x 2 / week), Yoga (30 minutes x 2 / week).
- I like tennis, but the courts are often only available in Summer in Germany. Running is always good. I don't enjoy Yoga.
- I'm staying fit.