Project Week 23/24
Project Week; AI-Powered Real-Time Support for Self-Paced Study
In this project, participants will engage in the development of innovative software leveraging artificial intelligence to provide real-time assistance and support to learners during self-paced study with pre-recorded online lectures or lecture slides. The primary goal is to enrich the overall learning experience by seamlessly integrating advanced technology (LLMs) into the educational environment. Through the implementation of generative AI solutions, particularly Large Language Models, the software empowers users to seek explanations related to video content through prompts such as screenshots or specified text. The integrated generative AI module is designed to possess a comprehensive understanding of the entire presentation, ensuring contextual awareness of the viewer's current slide when responding to queries. This software aims to deliver immediate answers to student questions during studying, mirroring the interactive nature of traditional classroom settings.
Project Week: Safeguarding Bavarian Beer Heritage: Immersive Virtual Beer Tour with AI
Our project aims to employ VR technology and AI in the creation of an innovative educational platform, enabling students and global audiences to conceptualize, develop, and engage with immersive experiences centered around Arts and Cultural Heritage. Focusing initially on Bavarian beer heritage, the platform's breadth will extend to encompass the diverse cultural aspects of Bavaria and other global cultural heritage topics.
Expanding the Scope to Other Cultural Heritage Subjects:
The scope of our project is not limited to Bavarian beer heritage; it will also encapsulate other vital aspects of Bavarian cultures, such as traditional cuisine, festivals, architecture, and crafts. The platform is built with adaptability in mind, allowing the integration of additional cultural heritage themes, such as:
• Exploration of historical and architectural marvels of castles and palaces
• Investigation into local folktales and legends
• Immersion into the rich traditions of Bavarian music and dance
Project Week: Comics Illustration Synthesis using Deep Generative Models
The objective of this project is to develop a generative model that can create comic illustrations from text descriptions. We aim to address the time-consuming and costly process of traditional comics illustration creation by using machine learning algorithms to learn from existing comics. The project will involve building a dataset of text-image pairs by extracting dialogue and illustration descriptions from Dilbert Comics, then using transfer learning and text-to-image generation techniques to create a text visualization model. The model will be evaluated through both automatic and human evaluations.
Project Week: Unleashing the Power of Large Language Models with In-Context Learning
"In-context learning, also known as conditioning on input-label pairs (prompts), has demonstrated significant promise for large language models (LLMs) in performing downstream tasks. In this project, students will explore how to adapt LLMs to new tasks using in-context learning techniques, without the need for parameter updates. Several benchmark datasets will be employed to demonstrate the effective utilization of prompts in achieving the desired outcomes, such as math problem reasoning, commonsense question answering, and sentiment classification. The primary objectives of this course are twofold:
(1) To gain expertise in state-of-the-art in-context learning techniques, thereby elevating model performance.
(2) To identify and address existing challenges, developing innovative solutions to further enhance the efficacy of in-context learning.
By the end of this project, participants will have a comprehensive understanding of in-context learning and its potential in advancing the field of LLMs."
Project Week: AI-assisted writing
There are already tools to support professional writing and augmenting the abilities of humans to get the best of both worlds, but there is still room for improvement in terms of education. Scholars receive only limited feedback for their essays or creative writing since teachers do not have the time to annotate texts in detail. That's where AI comes in handy. We aim to create a tool that supports students throughout the writing process by giving them timely feedback and proposing alternatives. To build this tool, we need an infrastructure which collects and structures essay data. The students set up an app specifically designed for data collection as well as ways organize it efficiently.