About 简介

Hello~ I’m Long Qian (aka. Neymar QL), a PhD student in Artificial Intelligence at the School of Future Technology, University of Chinese Academy of Sciences (UCAS), working with the Zidong Taichu Foundation Model Research Center, Institute of Automation, Chinese Academy of Sciences.

My current research is centered on VLM/LLM MoE pre-training and post-training. I am particularly interested in how large models can become more scalable, more modular, and more effective in real-world and scientific settings.

I also work on AI4S, anomaly detection, and related directions, with the broader goal of building foundation models that are not only strong, but genuinely useful.

Current keywords
VLM / LLM MoE Pre-Train VLM / LLM MoE Post-Train AI4S Anomaly Detection

Hello~ 我是钱隆（你可以叫我 Neymar QL），现为 中国科学院大学未来技术学院人工智能方向博士研究生， 并在中国科学院自动化研究所紫东太初大模型研究中心开展科研工作。

我当前最核心的研究重心是 VLM/LLM MoE 的 pre-train 和 post-train。 我尤其关注如何让大模型在复杂真实场景与科学问题中， 变得更具扩展性、更具模块化能力，也更有效。

此外，我也关注 AI4S、异常检测 等方向；总之，我关心的是如何构建真正“可用”的基础模型，而不只是 benchmark 上更强的模型。

当前关键词
VLM / LLM MoE 预训练 VLM / LLM MoE 后训练 AI4S 异常检测
Timeline 时间线
2026.01

ICASSP2026(Oral) Accepted

ICASSP2026(Oral) 论文接收

MathPhys-Guided Coarse-to-Fine Anomaly Synthesis with SQE-Driven Bi-Level Optimization for Anomaly Detection. MathPhys-Guided Coarse-to-Fine Anomaly Synthesis with SQE-Driven Bi-Level Optimization for Anomaly Detection。
2025.11

1st Place · ICAIS 2025 (Literature Q&A Track)

ICAIS 2025 文献问答赛道第一名

The 1st International Conference on AI Scientists (ICAIS 2025).
国际 AI 科学家大会（ICAIS 2025）。
2025.11

AAAI 2026 Accepted

AAAI 2026 论文录用

Quality-Aware Language-Conditioned Local Auto-Regressive Anomaly Synthesis and Detection. Quality-Aware Language-Conditioned Local Auto-Regressive Anomaly Synthesis and Detection。
2025.06

“Three-Good Student” Award @ UCAS

国科大“三好学生”

Recognition for excellence in overall performance.
综合表现优异获得荣誉称号。
2024.10

1st Place · ECCV 2024 BRAVO Challenge (Track 2)

ECCV 2024 BRAVO Challenge Track 2 第一名

The competition is held as a workshop challenge; the results paper is officially published at ECCV 2024.
比赛以 Workshop Challenge 形式举办。
2024.09

ECCV 2024 Published

ECCV 2024 正式发表

The BRAVO Semantic Segmentation Challenge Results in UNCV2024. The BRAVO Semantic Segmentation Challenge Results in UNCV2024。
Current

PhD Student @ UCAS

博士研究生 @ 国科大

School of Future Technology, University of Chinese Academy of Sciences (UCAS) · Zidong Taichu Foundation Model Research Center, Institute of Automation, Chinese Academy of Sciences (CASIA). 中国科学院大学未来技术学院 · 中国科学院自动化研究所紫东太初大模型研究中心。
2020.09 – 2024.07

B.Eng. in Computer Science and Technology @ UCAS

本科：计算机科学与技术 @ 中国科学院大学

Major: Computer Science and Technology. 专业：计算机科学与技术。
Highlights 重要事件
01/26
ICASSP 2026(Oral) Accepted ICASSP 2026(Oral) 接收 MathPhys-Guided Coarse-to-Fine Anomaly Synthesis with SQE-Driven Bi-Level Optimization for Anomaly Detection. MathPhys-Guided Coarse-to-Fine Anomaly Synthesis with SQE-Driven Bi-Level Optimization for Anomaly Detection。
11/25
1st Place · ICAIS 2025 (Literature Q&A Track) ICAIS 2025 文献问答赛道第一名 The 1st International Conference on AI Scientists (ICAIS 2025). 国际 AI 科学家大会（ICAIS 2025）。
11/25
AAAI 2026 Accepted AAAI 2026 论文录用 Quality-Aware Language-Conditioned Local Auto-Regressive Anomaly Synthesis and Detection. Quality-Aware Language-Conditioned Local Auto-Regressive Anomaly Synthesis and Detection。
10/24
1st Place · ECCV 2024 BRAVO Challenge (Track 2) ECCV 2024 BRAVO Challenge Track 2 第一名 The competition is held as a workshop challenge; the results paper is officially published at ECCV 2024. 比赛以 Workshop Challenge 形式举办；结果论文为 ECCV 2024 正式发表。
06/25
“Three-Good Student” Award · UCAS 国科大“三好学生” Recognized for overall excellence. 综合表现优秀获得荣誉称号。
Selected Publications 论文精选
AAAI 2026 · Accepted
Quality-Aware Language-Conditioned Local Auto-Regressive Anomaly Synthesis and Detection
Long Qian, Bingke Zhu, Yingying Chen, Ming Tang, Jinqiao Wang
📄 Paper 💻 Code 🎓 Scholar
AAAI 2026 VLM Anomaly Synthesis Quality-aware
ICASSP2026(Oral) · Accepted
MathPhys-Guided Coarse-to-Fine Anomaly Synthesis with SQE-Driven Bi-Level Optimization for Anomaly Detection
Long Qian, Bingke Zhu, Yingying Chen, Ming Tang, Jinqiao Wang
📄 Paper 🎓 Scholar
ICASSP2026(Oral) Anomaly Detection AI4S priors Bi-level optimization
arXiv 2024
Friend or Foe? Harnessing Controllable Overfitting for Anomaly Detection (COAD)
Long Qian, Bingke Zhu, Yingying Chen, Ming Tang, Jinqiao Wang
📄 Paper 🎓 Scholar
Anomaly Detection Controllable Overfitting
ECCV 2024
The BRAVO Semantic Segmentation Challenge Results in UNCV2024
Tuan-Hung Vu, Eduardo Valle, …, Long Qian, …, et al.
📄 Paper 🎓 Scholar
ECCV 2024 Semantic Segmentation BRAVO Challenge (Workshop)
For the complete publication list and citation stats, please visit my Google Scholar profile.
完整论文列表与引用统计请查看我的 Google Scholar 主页。
Projects 项目
QARAD / ARAS
Quality-aware, language-conditioned anomaly synthesis and detection via local auto-regressive generation. 质量感知、语言条件的异常合成与检测（局部自回归生成）。
💻 Code
VLM GenAI Anomaly
InternImage-OOD
Exploring OOD robustness and generalization for vision models based on InternImage, with a focus on real-world distribution shifts. 基于 InternImage 的视觉模型 OOD 鲁棒性与泛化研究，聚焦真实分布偏移场景。
💻 Code
OOD Robustness Vision
ICAIS PaperQA
An open-source scientific paper QA system for the ICAIS 2025 AI Scientists Challenge, supporting paper Q&A, literature review, ideation, and paper review with streaming LLM responses. ICAIS 2025 国际 AI 科学家大会文献问答赛道冠军项目的开源系统，支持论文问答、文献综述、科研想法生成与论文评审，并提供流式 LLM 响应。
💻 Code
Paper QA LLM Agent
Contact 联系

Open for research collaborations and academic discussions. 欢迎科研合作与学术交流。

Say Hello 发邮件 GitHub Scholar