Kai-Wei Chang

【個人資訊】

學校名稱：加州大學洛杉磯分校

實驗室名稱：UCLA NLP (LLM)

研究領域：自然語言處理 (NLP)、大型語言模型 (LLM)

職稱：副教授

個人經歷與網站：

• 現任UCLA電腦科學系副教授、UCLADataX AI 技術中心共同主任，以及Amazon AGI的Amazon Scholar
• 榮譽：Sloan Fellow (2021)、AAAI資深會員(2023)
• 獎項：獲EMNLP(2017)、KDD(2010)最佳論文獎及ACL(2023)傑出論文獎
• 個人網站：https://web.cs.ucla.edu/~kwchang/

研究室簡述

核心研究方向分為三類：

1. 可信賴NLP(Trustworthy NLP)： 專注於將模型與人類價值觀對齊（公平性與穩健性）
2. 多模態基礎模型(Multimodal Foundation Models)：開發如VisualBERT、GLIP等能透過語言描述辨識物體的視覺語言模型
3. NLP中的推理(Reasoning in NLP)：研究LLM遵循約束的能力，以及其常識、數學與邏輯推理能力

與臺大教授合作經驗：

• 曾為林智仁老師的研究生（合作Linear SVM研究）
• 與林軒田、王鈺強教授合作發表論文
• 與訪問學者林守德教授有合作項目
• 與李宏毅、陳縕儂及孫紹華教授研究領域相近，常於國際會議交流。

【研究實習計畫】

研究內容簡介：歡迎對Trustworthy ML、LLM Reasoning及Multimodal Reasoning有興趣且具相關經驗的同學加入

期待學生教育程度：

• 大三（含）以上大學生
• 碩士生
• 博士生

期待學生技能與特質：具備機器學習(ML)、強化學習(RL)或大型語言模型(LLM)之專業知識

其他要求：良好的英文能力

可接受期長：暑期

接受名額：2名。

簽證文件辦理：

• 由PI透過系所申請J1簽證文件
• 時長約 1-2 個月
• PI負擔DS-2019服務費

住宿：需學生自行尋覓（無校內住宿或補助）

財務證明與福利：

• 學生需證明每月有至少$2525 USD 的資金來源
• 其中51%須來自非個人/家庭來源，如台灣的獎學金
• 若單純訪問，PI可負擔部分學雜費（約數百美金）
• 對於背景極優秀者，若資金證明成為障礙，PI會協助尋求支持方案

學分提供：單純訪問則無。

‍

[ Personal Portfolio ]

University: University of California, Los Angeles (UCLA)

Laboratory: UCLA NLP (LLM)

Research Areas: Natural Language Processing (NLP), Large Language Models (LLMs)

Position: Associate Professor

Professional Background and Website:

• Associate Professor in the Department of Computer Science at UCLA
• Co-Director of the UCLA DataX AI Technology Center
• Amazon Scholar at Amazon AGI
• Honors: Sloan Fellow (2021), AAAI Senior Member (2023)
• Awards: Best Paper Awards at EMNLP (2017) and KDD (2010), and Outstanding Paper Award at ACL(2023)
• Personal Website: https://web.cs.ucla.edu/~kwchang/

Lab Overview

The lab’s core research directions fall into three categories:

1. Trustworthy NLP: Focuses on aligning models with human values,including fairness and robustness
2. Multimodal Foundation Models: Develops vision-language models such as VisualBERT and GLIP that can recognize objects through language descriptions
3. Reasoning in NLP: Studies the ability of LLMs to follow constraints, as well as their capabilities in commonsense, mathematical, and logical reasoning

Collaboration with NTU Faculty:

• Former student of Professor CJ Lin (worked onLinear SVM research)
• Co-authored publications with Professor Hsuan-TienLin and Professor Yu-Chiang Frank Wang
• Collaborated with visiting scholar Professor Shou-DeLin
• Has research overlap and frequent interactions atinternational conferences with Professors Hung-Yi Lee, Yun-Nung Chen, and Shao-HuaSun

[ Research Internship Program ]

Research Overview: Students with interest and relevant experience in Trustworthy Machine Learning, LLM reasoning, and multimodal reasoning are welcome to apply.

Target Student Level:

• Undergraduate students (junior year and above)
• Master’s students
• Ph.D. students

Expected Skills and Qualifications: Professional knowledge in machine learning (ML), reinforcement learning (RL), or large language models (LLMs)

Additional Requirement: Good English proficiency

Available Duration: Summer

Number of Positions: 2

Visa Processing:

• J-1 visa processed by the PI through the department
• Estimated processing time: 1–2 months
• DS-2019 service fee covered by the PI

Housing: Students must arrange their own housing (no on-campus housing or subsidyprovided)

Financial Requirements and Benefits:

• Students must demonstrate a minimum monthly funding of USD 2,525
• At least 51% of the funding must come from non-personal/ family sources (e.g., scholarships from Taiwan)
• For visiting students, the PI may cover part of the tuition/ fees (approximately a few hundred USD)
• For exceptionally strong candidates, if financial proof becomes a barrier, the PI may assist in identifying support options

Academic Credit: No credit will be provided for visiting students