Goodbye 2024, hello 2025.

Posted on In Views: Waline: Word count in article: 5.4k Reading time ≈ 5 mins.

This is an automatically translated post by LLM. The original post is in Chinese. If you find any translation errors, please leave a comment to help me improve the translation. Thanks!

Today is the last day of 2024, not particularly special, just like the past few days, running experiments - unsatisfactory results - running experiments again. Less than a month left until the deadline for IJCAI submission, hoping to finish writing the article smoothly.

I don't have the habit of writing a diary personally, I only write an article and put it on my blog when I feel the need to write something. Because I feel that a day is too short to form something worth thinking about and recording. However, it is necessary to reflect on things every once in a while, so here is this annual summary. If nothing unexpected happens, there will be such an article at the end of each year to summarize the work and life of the year.

Research

This year's research was not particularly smooth. The first paper submitted to ICML received a low score of 334 and was rejected, planning to submit to Neurips but missed the deadline due to a mistake in submission time. Later, when resubmitted to Cybernetics, it was rejected due to mismatched themes, and when resubmitted to NN, it was rejected due to some incorrect images. In short, it was a very bumpy road.

In addition, after returning from Beijing in April, I almost started the development of the lh project from scratch and had little time for research. It wasn't until the end of October, after the project was completed, that I had time to read papers and plan for future research. From the statistics of paper reading, this trend is clearly visible (laughs).

Even though it wasn't very smooth, this year was very important to me. Firstly, under the guidance of Professor Zhang Xi from the Institute of Automation, I completed the entire process from an idea to the final paper submission. I am not entirely a novice anymore, and what I need to do next is to continuously refine the various aspects of this process and submit more papers. Secondly, this year the lh project was completed, and the team's project-related matters came to an end, allowing me to focus on reading papers, conducting experiments, and writing articles. Everything is moving in a positive direction.

Hope to submit more articles next year o(￣▽￣)ブ

Project

After returning to school in April, I started working on the development of the lh project. This project was signed by me at the end of last year and lasted from October of last year to October of this year. Shortly after signing the project, I went to the Institute of Automation and did not pay much attention to the development progress of the project. The progress of the project was mainly handled by two junior brothers. The main content of this project is algorithm design and reinforcement learning training on a simulator developed by another company. When I returned, the progress of the project was in a mess: there were many bugs in the simulator that were not fixed, the algorithm code was coupled with the simulator code and difficult to manage, the development was still at the interface verification stage, and the required scene and algorithm development in the contract were not carried out.

With such a development progress, one junior brother was going for an internship in May, and the other junior brother had limited development capabilities. It would require a lot of handover costs for other team members to continue the development. In this situation, I took over the subsequent development work of this project, mainly considering the following aspects:

  • I had studied some books on software project management before, but I had never had the opportunity to put the content into practice. This project development was a good platform to practice project management.
  • Studying at the Institute of Automation was a favor to the mentor, completing this project was also a return to him. At the same time, I had discussed with the mentor that after completing this project, I would focus on research and not participate in other projects.

After the development of this project, I summarized some experiences about project development:

  • Consider what features are necessary and which ones are not needed. No code is better than unnecessary code.
  • Communicate more with the client. In open projects, the client usually does not know what they want. Communicate with the client using the minimum prototype to clarify the subsequent development arrangements (agile development).
  • In a long-term project, code maintainability is far more important than code performance. Because it is much easier to find an algorithm master to optimize a module's performance than to modify functionality in a pile of code.

Overall, I gained a lot during the development of this project. In the past, I focused on code performance and readability, but now, after encountering more projects, I focus on code maintainability and usability, stepping into the DevOps path.

Life

At the beginning of the year, I bought a Switch in Beijing (no computer to play games and can't not play games/(ㄒoㄒ)/~~). My friend gave me a lot of game cartridges for my birthday.

I went to the Grand View Garden with Chuli and encountered live streaming for product sales in the garden.

I built a white computer for Chuli, with all white accessories and a transparent case.

Before returning from Beijing, I went to the Old Summer Palace.

After returning to the lab, we set up a new server, with the configuration I chose, 2 AMD EPYC 9654, 192C384T@3.7GHz, AMD YES!

In May, I rode the Xi'an Metro's Xihu Line with Jiege, a rare subway with black smoke in this era!

I participated in the Young Doctoral Scholars' Trip to Taiyuan. (The magnificent Taiyuan young doctors journey. | Kezhi's Blog)

I climbed Huayang Grassland, Zhuge Forest Park.

I climbed Lushan in Wuhan, recited poems all night to enter for free but went to the wrong gate.

I bid farewell to many graduating friends. (Happy graduation in 2024. | Kezhi's Blog)

I completed the game Black Monkey using RDP. (Use RDP to remotely play games on the intranet. | Kezhi's Blog)

I visited Laozi at Hangu Pass.

I bought a 3D printer on Double Eleven to start a new business.

This year, Song got married, my cousin got engaged, Guoguo and Panting got together, Zhou Xiaotian and Teacher Tao also got engaged.

I hope everyone can find the right partner in the future, get married more, let me share the joy, and write more articles!