Data Science Retreat Demo Day #38

Data Science Retreat presents 11 Machine Learning prototypes and projects by Batch 38 participants. The event is free to attend and thanks to KI-Servicezentrum by Hasso-Plattner-Institut für Digital Engineering GmbH for hosting us.

Agenda:
17:30 - Drinks and Networking

18:00 - Welcome & Introduction

Followed by Project Presentations

Project Ideas:

1. IntelliTrack - AI-Powered Real-Time Inventory Tracking System
Project by David Tumma
This project improves inventory tracking, traditionally reliant on barcode scanners or RFID readers, by addressing their limitations like manual scanning and line-of-sight requirements.

2. Classification of types of sleep apnea based on audio-recordings of breathing sounds.
Project by Lucas Aresin
Using the PSG-Audio dataset, this project employs a convolutional neural network to classify sleep apnea events based on ambient audio recordings, allowing for at-home analysis using smartphone microphones.

3. EVA - Encrypted Visual Appearance
Project by Rahul Swaminathan
This project aims to create privacy-preserving video conferencing tools by altering the appearance and voice of the speaker through deep learning and cryptographic mechanisms, ensuring the identity remains hidden.

4. Optimization of Structural Design using Reinforcement Learning
Project by Arian Ghaemi
Reinforcement learning is used to optimize design parameters for reinforced concrete beams, aiming to minimize material costs while ensuring structural safety. The agent will also configure beam layouts for given floorplans.

5. AI-Driven Drum Machine with Custom Dataset Generation
Project by Filippo Guida
An AI network uses CNNs and a diffusion model with U-Net architecture to generate new drum patterns from a base pattern. The customizable dataset allows for diverse and innovative drum patterns.

6. Rhetorical Analysis Tool for News and Opinion Content
Project by John Heusinger
This tool analyzes news and opinion content to identify tactics like sensationalism and manipulation, promoting a critical approach to media consumption and awareness of content designed to elicit strong emotional responses.

7. AI-Powered insulin dosing assistant
Project by Amin Zayani
An AI-driven system provides real-time insulin dosing suggestions based on continuous glucose data, aiming to enhance insulin management by eliminating personal biases and improving accuracy.

8. Gentrification: Predicting Inclusive Growth or Displacement in Atlanta Neighborhoods
Project by Elyas Munye
This project identifies gentrifying neighborhoods in Atlanta and predicts whether they will experience inclusive growth or displacement. The goal is to guide neighborhood change to promote inclusive growth and mitigate displacement.

9. Use of Large multimodal models to assist blind or people with low vision
Project by Manish Lohani
This project presents the use of large multimodal Models to enhance accessibility and independence for blind or individuals with low vision. We will explore different use cases to assist performing daily tasks.

10. Study of the effects of EGNNs on the predictive power on the point cloud data
Project by Igor Zelenin
This project explores how Equivariant Graph Neural Networks (EGNNs) enhance predictive accuracy in point cloud data analysis. The results aim to advance the processing of unstructured 3D data for autonomous vehicle perception and forecasting.

11. Predicting Dangerous Areas in Urban Environments Using AI and Satellite Data
Project by Merve Demirtas Güzel
This project develops a predictive model to identify dangerous city areas using police records and satellite data. Advanced AI techniques like semantic segmentation, UNet, and ImageNet will analyze historical crime data and satellite imagery to highlight high-risk zones.

20:30 - Wrap up

See you all at the event.