Synchronized original link: https://www.dreamwings.cn/atec2022-variety-show/5902.html
"The Genius Programmer" Season 3 is finally airing! 🎈
Below are the updated video links for each episode:
- "The Genius Programmer 3" Episode 1: The strongest brain appears, 16 contestants face the cruelest competition format
- "The Genius Programmer 3" Episode 2: Contestants form teams, the real competition is about to begin
- "The Genius Programmer 3" Episode 3: Hunting moment, betrayal easter egg triggers team trust crisis
- "The Genius Programmer 3" Episode 4: Double reversal, who can embrace the dawn of victory
Preview and promotional text:
-
Scheduled! "The Genius Programmer" Season 3 will be available online on August 28th!
-
"The Genius Programmer" Season 3 Exciting Preview!
Haha, it's been a long time since I shared some technical topics. I've been lurking for a long time, so this time I'll make a big splash 😄😄😄
Hello everyone, I am one of the members of the problem-setting team for ATEC 2022. I am mainly responsible for the online competition problem one, offline competition problem one/problem three, and the design and security of the player submission process. This is also my first time transitioning from an ML competition participant to a problem setter. The whole process has been very rewarding, and I am very grateful for the efforts and dedication of the entire project team 💪
Both Zero and I have backgrounds in ACM. In the early stages, we thought about how to incorporate some unknown factors that could affect the players' decision-making into the boring competition. Therefore, we borrowed the sealed leaderboard mechanism from ACM (in variety shows, players are not fully/entirely aware of their current rankings). This made the entire competition intense and exciting throughout, and it is probably the first time in the history of machine learning competitions with such ups and downs! 👀
Speaking of problem setting, what I deeply felt this time is that setting problems for ML competitions is far from simple as solving problems 😣:
- In terms of problem design, we need to consider how to guide the players to the direction of the examination. Therefore, we need to do a lot of post-processing and testing on the score ladder to ensure that the expected solution can bring greater benefits. However, the competition time is limited, so this is also a consideration that exploring may lead to greater opportunities for score improvement.
- In terms of problem data, we should try to avoid the occurrence of abnormal data that may cause ambiguity under limited information, such as abnormal logs like script requests, or cases similar to coupon deduction amount greater than the threshold amount.
- In terms of problem rules, we need to ensure fairness to the greatest extent possible. Here, there is actually a sense of game between the players and the problem-setting team. The problem-setting team hopes that the players can come up with solutions that are suitable for industrial scenarios, but the players may use data leaks or loopholes in the problem rules to improve their scores.
Finally, I hope everyone will enjoy this program~ Looking forward to it~ 🤩🤩🤩