After realizing that there was no wifi in a cafe I went to this afternoon, I decided to spend time drawing some stuff around instead of just leaving.
July 29, 2017
July 28, 2017
SIGGRAPH 2017 parties
I wish people invite me to attend parties as much as they invite me to review papers.
=D— Li-Yi Wei (@liyiwei) July 28, 2017
https://www.facebook.com/groups/leagueparty/
SIGGRAPH Taipei
9 to 11 pm, Sunday July 30 2017
L.A. Live Courtyard & Residence Inn, LA 2 & 3
NVIDIA
6:30 – 9:30 pm, Monday July 31 2017
(Invitation only)
Snap
6 to 8 pm, Monday July 31 2017
Disney (?)
Dreamworks (?)
light.co (?)
GVU (?)
Adobe
5:30 to 7:30 pm, Wed August 2 2017
4th floor (outside) Bonaventure Brewing Co. 404 S. Figueroa St
Canadian night
9 pm to midnight, Wed Aug 2 2017
Shoo Shoo, Baby, 717 West 7th Street, Los Angeles, CA 90017
July 27, 2017
John Urschel
Retiring NFL player who is also a current MIT math PhD student and his latest paper is about CVT.
😀
HPRT and CVML
Looking at the paper titles for HPG 2017, I wonder if it can be renamed to HPRT (similar to how CVPR can be renamed to CVML).
When we have real-time ray tracing on mobile devices (the day will come), all the legacy graphics algorithms (i.e. tricks) will become obsolete.
I believe real-time RT, instead of machine learning, will be the end of traditional rendering tricks.
July 24, 2017
OKR 2017
For OKR 2016, I rated myself 50 percent:
- published only 2 top papers (with 2 student first authors)
- shipped no product
- graduated my first HKU PhD student who landed as a postdoc in USC
- received slightly below 80 percent rating for the machine learning class
- sketching level remains primitive
- finished the art of war, got stuck in the beginning of the book of change
- began the baroque cycle, but read a whole bunch of other books that I did not plan to
- my heart rate remained above 40 beats per minute but I picked up something else
The default period is from July 2017 to June 2018 unless stated otherwise.
Publish top papers
Publish at least 4 papers in top graphics/HCI/vision/ML venues: ToG (including SIGGRAPH and SIGGRAPH Asia), CHI/UIST, NIPS/ICML, CVPR/ICCV.
Build top products
Ship at least one product with at least 1 million users, preferably one that can help content creation with workflows.
Develop top talents
Help at least 4 students publish in top graphics/HCI/vision/ML venues individually.
Graduate my second HKU PhD student.
Have fun, be awesome
Practice sketching to a professional level.
Finish the book of change, to renew my (now rusty) Chinese and crack the only Chinese classics that my grandfather did not during his life time.
Finish the baroque cycle, which I lost patience in the middle of the first volume (quicksilver).
Reduce my steady heart rate to the previous level – less than 40 beats per minute, and perfect all three core forms of the art I have started learning for about 1 year.
July 23, 2017
Where are the coders?
I saw this article (in Chinese) pondering why it is so hard to find good programmers for software companies in Taiwan given the large number of CS graduates every year. The answer is that the hardware industry, which dominated in Taiwan, has a stronger pull than the software industry for programmers. Specifically, with higher pay and lower challenge, programmers tend to pick the easy route for the short term gain (hardware), forgoing longer term benefits such as impact and growth (software).
Hubs such as Hong Kong where finance dominates also have a stronger pull for programmers than the software industry. The job conversations I heard from the HKU CS students are mostly about different sectors of finance, such as investment banks, commercial banks, hedge funds, insurance companies, rather than finance versus other industries.
Top programmers flock to the software industry only where it dominates, such as in the Bay Area and the internet companies (e.g. BAT) of China. Their formations were due to largely unique factors (e.g. geography and government policy) and thus hard to replicate elsewhere.
Instead, computer scientists, as well as top guys in other professions, cluster where they can continue to dominate.
July 21, 2017
Khan’s grave
He is neither nice nor stupid.
Why would he want to put himself in a situation where he has no control and subject to humiliation from his enemies?
You cannot find something that does not exist.
What you can find are probably some decoys or (worse) booby traps.
July 19, 2017
July 18, 2017
Picking research problems
This is the most important stage of conducting research. I do not claim to know the answer, and my experience is limited to specific fields in CS, but (as before) I write this down to get feedbacks and to save the trouble of repeating myself.
Listen to people facing real problems instead of trusting the future work sections in research papers. If there are good (to elaborate) users or product people around you, ask what they need.
Identify concrete problems first, before thinking about other parts like idea novelty, algorithm framework, etc. (Look for hammers after nails.)
If we can find a trivial solution for an important problem, great, solve it, and quickly move on to something else.
This beats the alternative of coming up with a fancy method that does not solve anything.
The problem should fit our interest and expertise well enough, so that we have sufficient willingness and capability to address it.
If the problem is not sufficiently clear or convincing, the research project is probably a waste of time.
If the solution to the problem is not clear or convincing at the beginning, it is not only quite normal but also a potential sign that the problem is sufficiently non-trivial.