6 papers
Uncertainty-Aware Remaining Lifespan Prediction from Images
Tristan Kenneweg, Philip Kenneweg, Barbara Hammer
Predicting mortality-related outcomes from images offers the prospect of accessible, noninvasive, and scalable health screening. We present a method that leverages pretrained visio…
JEPA for RL: Investigating Joint-Embedding Predictive Architectures for Reinforcement Learning
Tristan Kenneweg, Philip Kenneweg, Barbara Hammer
Joint-Embedding Predictive Architectures (JEPA) have recently become popular as promising architectures for self-supervised learning. Vision transformers have been trained using JE…
Generating Synthetic Genotypes using Diffusion Models
Philip Kenneweg, Raghuram Dandinasivara, Xiao Luo +2
In this paper, we introduce the first diffusion model designed to generate complete synthetic human genotypes, which, by standard protocols, one can straightforwardly expand into f…
The SAME score: Improved cosine based bias score for word embeddings
Sarah Schröder, Alexander Schulz, Barbara Hammer
With the enourmous popularity of large language models, many researchers have raised ethical concerns regarding social biases incorporated in such models. Several methods to measur…
Evaluating Metrics for Bias in Word Embeddings
Sarah Schröder, Alexander Schulz, Philip Kenneweg +3
Over the last years, word and sentence embeddings have established as text preprocessing for all kinds of NLP tasks and improved the performances significantly. Unfortunately, it h…
No learning rates needed: Introducing SALSA -- Stable Armijo Line Search Adaptation
Philip Kenneweg, Tristan Kenneweg, Fabian Fumagalli +1
In recent studies, line search methods have been demonstrated to significantly enhance the performance of conventional stochastic gradient descent techniques across various dataset…