Selected Topics in Data Science
Die Seminarteilnehmer werden aktuelle Forschungstrends und etablierte Ansätze im Bereich der Data Science behandeln. Die Teilnehmer können wählen, ob sie im Rahmen dieses Seminars den theoretischen oder den praktischen Forschungsprojektteil absolvieren möchten
Seminararbeiten aus vergangenen Semestern
SoSe 2025
Angewandte Arbeiten
- Supply Chain Resilience using Reinforcement Learning
- Neuroscientific Ablations
- Applying XAI Methods: Which Explanation wins?
- Implementation of specific scheduling problems
- Adversarial Attacks for Image Recognition Manipulation
- XAI for camera-based object detection
Theoretische Arbeiten
- Application of ML and DL in the World of Football
- Reinforcement Learning for Sustainable Production Optimization
- Literature Review on Supply Chain Resilience using Reinforcement Learning
- Bridging Data Across Diverse Sources with Schema Alignment
- Evaluating XAI Methods: Which Explanation wins?
- Research on scheduling problems
- Adversarial Attacks for object detection
- Model Compression for 3d Object Detection: Literature Survey
WiSe 2024-2025
Angewandte Arbeiten
- Data Analysis to Protect Against Climate-Driven Extremes
- Recreate 6DoF estimation setup in virtual domain
- Comparison of manual and automatic keypoint estimation
- Deep Neuroevolution - Optimization with Genetic Algorithms
- Adversarial Attacks for Image Recognition Manipulation
- Visualization Methods of Semantic Models
- Deep Learning Architectures for Efficient Scheduling
- Deep reinforcement learning for multi-objective scheduling
Theoretische Arbeiten
- Applications of ML and DL in the World of Football
- Adversarial Attacks for Image Recognition Manipulation
- Model Compression for 3d Object Detection: Literature Survey
- Evaluating XAI Methods: Which Explanation wins?
- Semantic standards for the digital product passport
SoSe 2024
Angewandte Arbeiten
- "Why you so good? – Comparison of two State-of-the-Art Agents"
- Accelerated 3D-Model Texturing form 2D-Images
- Adversarial Attacks for Image Recognition Manipulation
- Develop a TensorFlow (Lite) Model for Edge Devices
- Evaluating TensorFlow Lite Models for Edge Devices
- Fitting Educational Images to Target Groups using Generative AI
- Improving Anomaly Detection with Synthetic Data
- Comparative work in XAI - What defines a good explanation
Theoretische Arbeiten
- Evaluating the quality of LLM generated simplified text
- Literature Research in XAI – What defines a good explanation?
- Overview of E-Commerce Forecasting Tasks
- Prompt-engineering of LLMs for Text simplification
- SotA for Visual Sensors
Arbeiten aus vergangenen Semestern
| Semester | Titel | Art |
| SoSe 22 | QR-Code Reconstruction from Low-Resolution Video Frames | Applied |
| SoSe 22 | Dataset Enrichment using Simulations | Applied |
| SoSe 22 | Water Level Prediction of the Wupper using PyTorch Forecasting | Applied |
| SoSe 22 | Implicit vs. Explicit Information for Neural Networks | Applied |
| SoSe 22 | Detecting valid robot poses with neural networks | Applied |
| SoSe 22 | How to measure similarity of robot poses | Applied |
| SoSe 22 | Setup of a Machine Learning Semantic Labeling Algorithm | Applied |
| SoSe 22 | Strengths and Weaknesses of Dispatching Rules in Production Scheduling | Applied |
| SoSe 22 | An experimental comparative study of time series databases | Applied |
| SoSe 22 | Literature Review for Flood Prediction Using Deep Learning | Theory |
| SoSe 22 | State of the art in optical character recognition (OCR) | Theory |
| SoSe 22 | How to detect fake news in social media | Theory |
| SoSe 22 | In data we trust? The gap between reality and algorithmic predictions | Theory |
| SoSe 20 | Robustes Reinforcement Learning durch Domain Randomization | Applied |
| SoSe 20 | Was wäre wenn? - Hindsight experience replay for Reinforcement Learning | Applied |
| SoSe 20 | Intrisische Motivation zur effizienten Exploration in Reinforcement Learning | Applied |
