Seminar on Internet Technologies (Winter 2024/2025)

From NET Wiki
Jump to navigation Jump to search

Details

Workload/ECTS Credits: 5 ECTS (BSc/MSc AI); 5 (ITIS)
Lecturer: Prof. Xiaoming Fu
Teaching assistant: Tong Shen[shen.tong@cs.uni-goettingen.de]
Time: Please read this introduction slide [1]. If there is any question, please contact teaching assistants.


Announcement

No open talk. You can contact your topic advisor to schedule a 1V1 meeting or talk.

Course description

This course covers selected topics on up-to-date Internet technologies and research. Each student chooses a topic, does a presentation, and writes a report on it. Besides the introduction meeting, there are no regular meetings, lectures, or classes for this course. The purpose of this course is to familiarize the students with new technologies, enable the independent study of a specific topic, and train presentation and writing skills.

The informational meeting at the beginning of the course will cover some guidelines on scientific presenting and writing.

Due to the topic advisors' workload limitation, we could only provide limited topics, and the topic assignment will be on the basis of the first come first serve principle. Please contact the topic advisor directly for the topic availability.

Note: Participants in the seminar only need to register for the exam before the end of the course.

Passing requirements

  • There will be 2 milestones before the presentations that the students should pass before they register for the course.
    • Intro milestone where the adviser makes sure that the student starts to work on the topic and follows an accepted methodology.
    • Midterm milestone. (ex. programming tasks are done etc... )
  • Actively and frequently participate in the project communication with the topic advisor
    • This accounts for 20% of your grade.
  • Present the selected topic offline (20 min. presentations + 10 min. Q&A).
    • This accounts for 40% of your grade.
  • Write a report on the selected topic (6-8 pages) (LaTeX Template:[2]).
    • This accounts for 40% of your grade.
  • Please check the #Schedule and adhere to it.

Schedule

  • TBD.01.2025: Deadline for registration to attend the final presentation
  • TBD.01.2025 : Final Presentations (Online, wait to decide)
  • TBD.02.2025(23:59) : Deadline for submission of the report (should be sent to the topic adviser!).

Topics

Topic Description Prerequisites Topic Advisor Readings Available
Privacy protection in video analytics In this topic, you will study how to do privacy protection in video analytics, e.g., video blur Basic programming knowledge, Basic machine learning knowledge, need coding work [Tingting Yuan, tingting.yuan@cs.uni-goettingen.de] Yes
How to do efficient offline training In this topic, you will study how to do efficient offline training for reinforcement learning Basic programming knowledge, Basic machine learning knowledge, need coding work [Tingting Yuan, tingting.yuan@cs.uni-goettingen.de] Yes
Disaster Monitoring In this topic, you will study methods to monitor disaster scenarios (e.g. floodings) with aerial images. Basic machine learning knowledge [Fabian Wölk, fabian.woelk@cs.uni-goettingen.de] Yes
Explainable AI(XAI) / graph neural network (XGNN) In this topic, student will study how AI models / GNNs are explained by SOTA papers. Basic AI / GNN knowledge [Zhengze Li, zhengze.li@cs.uni-goettingen.de] Yes
Anomaly Detection in Graphs In this topic, student will read papers to learn how to detect anomaly edge/graph/subgraph… with the help of GNN. Basic AI / GNN knowledge [Zhengze Li, zhengze.li@cs.uni-goettingen.de] Yes
Social Media Comments Network (Intern/Project/Thesis possible) In this topic, student will study methods to crawl the dataset from social networks(e.g. X, YouTube) and utilize social network analysis in any topic you are interested in (science/education/language…) to find out the network structure and compare the difference among different topics. Python(Data crawling, cleaning, statistical data analysis, modeling and visualization), basic graph knowledge would be a plus [Zhengze Li, zhengze.li@cs.uni-goettingen.de] No
Influence of LLM robots in social networks (Intern/Project/Thesis possible) In this topic, student will study methods to crawl the data of LLM robots from social networks(e.g. X, Facebook) and utilize NLP and SNA to evaluate the influence of LLM robots in a specific topic. Python(Data crawling, cleaning, statistical data analysis, modeling and visualization), basic graph knowledge would be a plus [Zhengze Li, zhengze.li@cs.uni-goettingen.de] Yes
The life-circle of vanished scientific journals (Intern/Project/Thesis possible) In this topic, student will mine the information of vanished/(ongoing)/top journals, try to find out the difference features(manually/ML-based method) between journals facing different destinies. Python(Data Crawling, Cleaning, EDA, Modeling). Basic graph, XAI knowledge is a plus. [Zhengze Li, zhengze.li@cs.uni-goettingen.de] Yes
Traffic prediction with GNN (Intern/Project/Thesis possible) In this topic, student will study how to use XGNN to predict traffic volumn. Python(Modeling and Visualization). Graph and XAI knowledge. [Zhengze Li, zhengze.li@cs.uni-goettingen.de] Yes
ML/DL based industrial equipment predictive maintenance (Intern/Project/Thesis possible) In this topic, student will study how to use cutting-edge machine learning models to predict when industrial equipment need to be maintained before crashing. Python(Cleaning, EDA, Modeling and Visualization). XAI knowledge is a plus. [Zhengze Li, zhengze.li@cs.uni-goettingen.de] Yes
AI for High-quality Image Restoration and Manipulation (Intern/Project/Thesis possible) Image restoration and manipulation are low-level vison problems aiming to either restore the degraded images for higher perceptual quality (such as better color, contrast brightness, etc.) or manipulate image styles content for better understanding or visual-appealing effects. Moreover, such problems also plays key role for many high-level computer vision tasks, including image detection, recognition and (semantic) segmentation... In this topic, students need to follow the new trends and advances in the area of many sup-problem and explore new methods for completive or superior opportunity for academic and industrial applications. Python & CV knowledge. [Zhengze Li, zhengze.li@cs.uni-goettingen.de] No
Lidar-based traffic flow analysis In this topic, you will study methods to analyze traffic usage on roads, e.g., in terms of traffic flow, speed, and density to identify patterns and trends. Basic point cloud processing & ML knowledge [Yanlong Huang, yanlong.huang@cs.uni-goettingen.de] Yes
Personalized chatbot based on ChatGPT In this topic, you will learn about ChatGPT and learn to use OpenAI ChatGPT API to create a personalized chatbot. NLP & ChatGPT [Wenfang Wu, wenfang.wu@cs.uni-goettingen.de] Yes
Multimodal Large Language Model Evaluation for Multimodal Tasks This topic focuses on assessing the performance and effectiveness of large language models in handling tasks that involve multiple modalities, such as text, images, and audio. It involves the evaluation of these large models using specialized multimodal datasets, considering both quantitative metrics and qualitative analysis. In this topic, you will gain insights into various large models, including GPT-4, and acquire knowledge about a wide range of multimodal tasks. Additionally, you will become proficient in implementing evaluations related to these tasks. Large Language Model & multimodal setting [Wenfang Wu, wenfang.wu@cs.uni-goettingen.de] Yes
Knowledge Graph Completion What are the Knowledge Graph (KG) requirements for future applications and scenarios? What is the task of Knowledge Graph Completion? What is the correlation between KGs and NLP? How to use popular large language models (LLMs) to assist in the implementation of knowledge graph completion? In this topic, you will learn about KGs and learn to use LLMs to perform a KGC task. Knowledge Graph & NLP [Tong Shen, shen.tong@cs.uni-goettingen.de] Yes
Emotional Support Conversation Generation Does the large language model have emotions? Can it provide emotional support to users? In this topic, you will learn about techniques of large language models, such as prompt engineering and instruction fine-tuning, and use the above approaches to implement the emotional support conversation. Large Language Model & Emotional Support [Jing Li, jing.li@cs.uni-goettingen.de] Yes

Workflow

1. Select a topic

Each student needs to choose a topic from the list. You can start to work on your selected topic at any time. However, please make sure to notify the advisor of your selected topic in advance, because you might be refused by the advisor if someone has registered on the same topic.

2. Get your work advised

Each topic has an advisor, who will help you to solve problems regarding the topic. Please do not hesitate to contact your advisor. It is recommended (and not mandatory) that you can schedule a skype or zoom meeting with your advisor right after you select your topic. Your advisor will give you some useful guidance and suggestions, which will help you to gain more from this course.

3. Approach your topic

  • By choosing a topic, you will get a direction of elaboration.
  • You may work in different styles, for example:
    • Survey: Basic introduction, an overview of the field; general problems, methods, approaches.
    • Specific problem: Detailed introduction, details about the problem, and the solution.
  • Based on the research, you should have your own ideas on your topic.

4. Prepare presentation

  • Present your topic to the audience (in English).
  • The final presentation will be conducted offline.
  • 20 minutes of presentation followed by 10 minutes of discussion.

You need to present your topic to an audience of students and other interested people (usually the NET group members). Your presentation should include your general idea of your topic and highlight interesting problems and solutions. You must finish your presentation within a limited time. You have 20 minutes to present your topic followed by 10 minutes of discussion. It is highly recommended to send your slides to your topic advisor in advance, he/she will give you help with your presentation.

Hints for preparing the presentation: If your topic includes many aspects, and 20 minutes is too short for you to introduce them all, it is recommended to focus on one certain important aspect. Besides, you can discuss with your advisor, he/she will help you to reduce the content. Please make sure to finish your presentation in time.

Suggestions for preparing the slides: No more than 20 pages/slides. Get your audiences to quickly understand the general idea. Figures, tables, and animations are better than sentences. Don't forget a summary of the topic and your ideas.

5. Write a report

  • Present the problem with its background.
  • Detail the approaches, techniques, and methods to solve the problem.
  • Evaluate and assess those approaches (e.g., pros and cons).
  • Give a short outlook on potential future developments.

The report must be written in English according to common guidelines for scientific papers, between 6 and 8 pages of content (excluding bibliography, etc.). Please note that you can not directly copy content from papers or webpages, as this will be considered plagiarism. All quoted images and tables need to indicate their source.

6. Course schedule

There are no regular meetings, lectures, or classes for this course. The work is expected to be done by yourself with the assistance of your topic advisor. Please follow the #Schedule to take appropriate actions.