Experience: From May 2022 to March 2024, the lab member Zhipeng Zhou conducted research internships at Tencent AI Lab Machine Learning Center (Shenzhen). During this period, he primarily focused on research areas such as long-tail learning and multi-task learning. As the first author, he produced three papers, two of which have been accepted by top-tier conferences in computer vision, CVPR 2023 [1], and machine learning, ICLR 2024 [2].
Reflection: I am grateful to my advisor for suggesting that I pursue an internship two years ago, which led to my internship experience in Shenzhen over the past two years. When I first arrived in Shenzhen, it was during the most severe period of epidemic prevention and control. I spent almost every day commuting and closely following real-time news. I was uncertain about when and where lockdowns would occur, whether my health code was green, and if the company building was closed. This internship was my first real entry into the professional world, so many things were unfamiliar to me, such as finding accommodation, commuting, and stockpiling supplies during lockdowns. This situation persisted for about six months. Once the epidemic was under control, I gradually adapted to life in Shenzhen and started exploring the city, going hiking, and having meals with fellow interns from other universities.
Perhaps different from the internship experiences of many others, during my time at Tencent AI Lab, I did not experience significant pressure in terms of business tasks, thanks to the protection and care provided by my industry mentors and leaders. This allowed me to focus on research and make substantial contributions. With the help of the company's extensive computing resources, I was able to thoroughly validate various ideas and continuously verify and understand the research principles advocated by my advisor. This experience helped me strengthen my motivation, solidify my understanding of research problems, and produce well-grounded papers. It marked a significant turning point and breakthrough in my research philosophy. The academic atmosphere in my research group was vibrant, with industry mentors at the forefront of research and a group of interns from various prestigious universities. We discussed problems, exchanged ideas, and made rapid progress together. I still remember that during these two years, our group of interns voluntarily worked overtime at the company on weekends, investing in ourselves. The collective effort to meet conference deadlines was like an adrenaline rush, intense, exciting, and tireless.
Looking back on these two years of internship, I am grateful for my advisor's advice and thankful for this extraordinary experience, which has given me more confidence to continue on the path of research!
Fig：the building where the internship department is located Fig：the photo of climbing a mountain
[1] Zhipeng Zhou*, Lanqing Li*, Peilin Zhao, Pheng-Ann Heng, Wei Gong. Class-Conditional Sharpness-Aware Minimization for Deep Long-Tailed Recognition. In CVPR 2023. (CCF-A)
[2] Zhipeng Zhou, Liu Liu, Peilin Zhao, Wei Gong. Pareto Deep Long-Tailed Recognition: A Conflict-Averse Solution. In ICLR 2024.