Description
Learning and growing together: The poster session offers Lamarr members with a platform to showcase their research projects. Lamarr members across all research areas, working groups and levels of hierarchy and experience are asked to join the open exchange and discuss the presented work.
Trustworthy methods for medical image segmentation should come with a reliable mechanism to estimate the quality of their results. Training a separate component for confidence prediction is relatively fast, and can easily be adapted to different quality metrics. However, the resulting estimates are usually not sufficiently reliable under domain shifts, for example when images are taken with...
Using data-based approaches, accurate predictions of thermal deformations, which can significantly affect the quality of manufactured components, can be enabled. However, a sufficient amount of data with maximised information content is necessary for efficient training. In this paper, an approach for optimising sensor configurations for predicting thermal deformations is presented. From...
Accurate predictions of process characteristics in milling, such as tool vibrations, allow for identifying and avoiding unstable cutting conditions that can lead to excessive tool wear, surface defects or tool breakage. Therefore, considering vibrations during process design is essential to ensure dimensional accuracy, surface integrity and the longevity of cutting tools. Common methods,...
We address two critical capabilities required for autonomous robots operating in indoor environments, both centered around robust perception of unseen objects. This generalization can support various applications, but here we focus on mobile robotics.
The first focus is on robotic grasping, where 6D pose estimation is needed for successful manipulation. While 6D tracking is now reliable,...
Autonomous assembly is an essential capability for industrial and service robots, with Peg-in-Hole (PiH) insertion being one of the core tasks. However, PiH assembly in unknown environments is still challenging due to uncertainty in task parameters, such as the hole position and orientation, resulting from sensor noise. Although context-based meta reinforcement learning (RL) methods have been...
This work explores the applicability of synthetic data for training deep learning models aimed at real-time classification of astronomical radio signals. Building on previous research where lightweight convolutional neural networks (CNNs) using DM-time representations showed promising performance in detecting transient signals, we now turn to the question of whether synthetic datasets can...
The Boolean Satisfiability Problem (SAT) is a foundational problem in computer science with applications across a wide range of domains. Because SAT solvers exhibit varying behavior across different problem classes, the ability to generate synthetic SAT instances is valuable for benchmarking and solver-specific analysis. Recent methods have introduced Deep Learning approaches into this...
Tractography enables the reconstruction of white matter pathways from diffusion MRI and is a key tool for studying brain connectivity in both research and clinical contexts. Within the overall tractography pipeline, the parcellation step assigns individual streamlines to specific anatomical bundles, or discards them as false positive detections. We introduce PETParc (Parallel Efficient...
Robust generalization is a foundational requirement for trustworthy artificial intelligence (AI), underpinning the reliability, stability, and fairness of deployed systems. Two geometric phenomena are frequently correlated with generalization: neural collapse, where internal class representations converge to a maximally simple and symmetric structure, and flatness of the loss landscape, where...
Stochastically sampling word segmentations from a subword tokeniser, also called subword regularisation, is a known way to increase robustness of language models to out-of-distribution inputs, such as text containing spelling errors. Recent work has observed that usual augmentations that make popular deterministic subword tokenisers stochastic still cause only a handful of all possible...
Multi-Agent Path Finding (MAPF) focuses on determining conflict-free paths for multiple agents navigating through a shared space to reach specified goal locations. This problem becomes computationally challenging, particularly when handling large numbers of agents, as frequently encountered in practical applications like coordinating autonomous vehicles. Quantum Computing (QC) is a promising...
Measurements of neutral, oscillating mesons are a gateway to quantum mechanics and give access to the fundamental interactions of elementary particles. For example, precise measurements of violation in neutral mesons can be taken in order to test the Standard Model of particle physics. These measurements require knowledge of the -meson flavour at the time of its production, which cannot be...
Neural networks often suffer from plasticity loss, which limits their ability to adapt to evolving data distributions in continual learning settings. This results in degraded performance, poor generalization, and inefficient use of model capacity. While recent methods mitigate this by resetting underutilized neurons based on utility scores, the underlying mechanisms remain poorly understood....
Forecasting high-energy flares in blazars—active galactic nuclei with relativistic plasma jets oriented toward Earth—over extended temporal horizons presents a significant challenge due to the complex variability inherent in their light curves. In this study, we investigate the long-term predictability of flare activity using over 15 years of photon flux observations from the Fermi-LAT...
Emergent Misalignment (EMA) is a puzzling phenomenon where models finetuned on a narrowly misaligned task (e.g., including insecure backdoors in code) learn to be broadly misaligned. EMA is concerning, as models trained on superficially harmless data might become broadly misaligned. At the same time, the fact that alignment behavior across different domains is so strongly correlated during...
Hyperbolic representations are effective in modeling knowledge graph data which is prevalently used to facilitate multi-hop reasoning. However, a rigorous and detailed comparison of the two spaces for this task is lacking. In this paper, through a simple integration of hyperbolic representations with an encoder-decoder model, we perform a controlled and comprehensive set of experiments to...
The AI research ecosystem is a demanding, high-pressure environment that profoundly shapes the future of technology. Its effectiveness and sustainability depend not only on technical innovation but also on the people who sustain its progress. Investigating the psychosocial factors that link individual traits to work experiences and mental health is therefore essential for enabling sustainable,...
Current pruning methods for large language models (LLMs) achieve high compression post-training while preserving performance. However, most existing work focuses on calibration using English data, despite the multilingual nature of modern LLMs and their widespread use in non-English languages. This poster presents the first study on how the calibration language impacts pruning for multilingual...
We propose a sublinear algorithm for probabilistic testing of the discrete Fréchet distance - a standard similarity measure for curves.
We assume the algorithm is given access to the input curves via a query oracle that returns the set of vertices of the curve that lie within a radius $\delta$ of a specified vertex of the other curve.
The goal is to use a small number of queries to...
Traditional interpretability techniques such as rule-based models and feature attribution methods, each offer complementary strengths, however are often applied in isolation. Rule-based approaches are intuitive and logically structured, making them easy to understand, but they often struggle to scale effectively. On the other hand, feature attribution techniques like SHAP are well-suited to...
Service robots operating in dynamic human environments must reliably locate objects that are moved, concealed, or completely novel. Current frameworks often assume static environments, failing to address these real-world uncertainties.
We present an open-vocabulary framework that combines spatial, semantic, and geometric reasoning to overcome these challenges. By unifying spatial cues about...
This abstract outlines my current research for my PhD thesis, focusing specifically on creating a synthetic dataset for multi-camera multi-object tracking (MCMOT) within logistics applications.
Motivation: Tracking moving assets such as trucks, trailers, or containers in logistics yards is crucial for developing digital twins, measuring key performance indicators, and enhancing operational...
The advancement of artificial intelligence (AI) in intralogistics critically depends on the availability of realistic and diverse datasets. However, existing datasets in this domain often focus on narrow tasks such as object detection or activity recognition, lacking comprehensive three-dimensional (3D) representations of entire intralogistics systems. This paper addresses this gap by...
Understanding causal relationships in oncology is essential for improving treatment strategies and generating testable medical hypotheses. We present CaDSIm (Causal Discovery with Simultaneous Imputation), a new method for learning causal structures and associated Structural Equation Models from real world pan-cancer data, which is typically high dimensional, noisy, and incomplete.
Our...
The onset detection of epileptic seizures from multivariate Electroencephalogram (EEG) data is a challenging task. The variation in seizure patterns across patients and epilepsy types makes it particularly difficult to create a generic solution. Existing approaches indicate low recall due to their inability to capture complex seizure onset patterns. In this paper, we propose VISOR – a novel...
Dynamical systems governed by ordinary differential equations (ODEs) serve as models for a vast number of natural and social phenomena. In this work, we offer a fresh perspective on the classical problem of imputing missing time series data, whose underlying dynamics are assumed to be determined by ODEs. Specifically, we revisit ideas from amortized inference and neural operators, and propose...
Recent works for time-series forecasting more and more leverage the high predictive power of Deep Learning models.
With this increase in model complexity, however, comes a lack in understanding of the underlying model decision process, which is problematic for high-stakes application scenarios. At the same time, simple, interpretable forecasting methods such as ARIMA still perform very...
In machining processes, measurement data collected during operation can provide valuable insights into the final quality of the manufactured components. This enables both the reduction of unnecessarily long process chains and the real-time adaptation of process parameters. Machine learning models trained on this data can capture and predict complex process characteristics. However, acquiring...
Over the last decade, AI-algorithms have become a standard tool for data analysis in astroparticle physics. While these efforts were pioneered by the use of ensemble methods for event selection in IceCube, the capabilities of AI in the context of neutrino astronomy have been exemplified by the detection of neutrinos from the Milky Way (also by the IceCube collaboration). This detection was...
Anytime algorithms can be interrupted before completion while still delivering an intermediate result. This is a desirable property for embedded systems where timing is critical, such as object detection in cyber-physical-systems. However it is generally neither supported by models nor inference frameworks. To enable a model to be anytime, early-exits can be added to the network, which allow...
Chirality information (i.e., information that allows distinguishing left from right) is ubiquitous for various data modes in computer vision, including images, videos, point clouds, and meshes. Contrary to symmetry, for which there has been a lot of research in the image domain, chirality information in shape analysis (point clouds and meshes) has remained underdeveloped. Although many shape...
Despite advances in conversational systems, the evaluation of such systems remains a challenging problem. Current evaluation paradigms often rely on costly homogeneous human annotators or oversimplified automated metrics, leading to a critical gap in socially aligned conversational agents, where pluralistic values (i.e., acknowledging diverse human experiences) are essential to reflect the...
We explore what it means to build a scientific "theory" of a black-box model, drawing on van Fraassen's Constructive Empiricism (CE), and demonstrate how such a theory can be used for explainable AI (XAI).
A scientific theory is more than just an explanation: it not only has value in its own right, but also serves as a robust framework for answering different questions.
According to CE, a...
In recent years, machine learning and deep learning have revolutionized data analysis across various fields, including particle physics and medical imaging. However, their potential in radio interferometry—a technique used to study the universe through arrays of radio telescopes—remains underexplored. The radionets-project has been pioneering the use of deep learning methods to process data...
Service robots operating in cluttered human environments such as homes, offices, and schools cannot rely on predefined object arrangements and must continuously update their semantic and spatial estimates while dealing with possible frequent rearrangement. Identifying all objects in cluttered, occlusion-heavy environments, such as shelves, requires selecting informative viewpoints and...
The post-surgical gauze retention can lead to serious complications and necessitate additional surgery for its removal. Due to data scarcity, the research on gauze segmentation on real-world surgical data remains underexplored. This work presents first investigation of gauze segmentation on real-surgical data. We use prevalently used segmentation architectures, including CNN-based,...
In this work, we address unsupervised temporal action segmentation, which segments a set of long, untrimmed videos into semantically meaningful segments that are consistent across videos. While recent approaches combine representation learning and clustering in a single step for this task, they do not cope with large variations within temporal segments of the same class. To address this...
Large Language Models (LLMs) remain vulnerable to adversarial jailbreaks, yet existing attacks rely on handcrafted priors or require white-box access for gradient propagation. We show that token-level iterative optimization can succeed without gradients and introduce RAILS (RAndom Iterative Local Search), a simple yet effective method using only model logits with a query budget comparable to...
Traditional Task test and Motion Planning (TAMP) systems integrate physics simulators for motion planning with discrete symbolic models for task planning. However, because these symbolic models are not derived from data, they must be meticulously handcrafted, requiring manually designed classifiers to bridge the gap with the physics simulator. This process is both resource-intensive and...
In the healthcare domain, sensitive patient data is inherently decentralized across institutions and cannot be centralized due to strict privacy regulations. Federated learning offers a collaborative model training without explicitly sharing patient data by communicating model parameters or soft labels. These approaches, however, are still vulnerable to privacy leakage and often limit model...
Social sciences define values as preferred behaviors or outcomes that motivate an individual's actions or judgments.
While LLMs often reflect biases from their training data, it remains unclear what values underlie their generation processes, and whether such internal value systems can be measured or modified.
In this paper, we investigate whether fine-tuning can steer a model’s internal...
In this paper, we present MCBench, a benchmark suite designed to assess the quality of Monte Carlo (MC) samples. The benchmark suite enables quantitative comparisons of samples by applying different metrics, including basic statistical metrics as well as more complex measures, in particular the sliced Wasserstein distance and the maximum mean discrepancy. We apply these metrics to point clouds...
Reinforcement learning traditionally learns absolute state values, estimating how good a particular situation is in isolation. Yet in both biological systems and practical decision-making, what often matters is not the absolute value of a state, but how it compares to alternatives. Motivated by empirical findings in neuroscience, we introduce \textbf{Pairwise-TD}, a novel framework that learns...
In this article, we propose a novel quantum regression model by extending the Real-Part Quantum SVM. We apply our model to the problem of stability limit prediction in milling processes, a key component in high-precision manufacturing. To train our model, we use a custom data set acquired by an extensive series of milling experiments using different spindle speeds, enhanced with a custom...
Designing stable milling operations is crucial to ensure a high surface quality of the machined workpieces and reduce rejects during production. Stability lobe diagrams can be used to identify stable conditions. Analytical approaches or simulation techniques can be used to reduce the experimental effort for stability evaluation for different process parameter values. However, complex...
Providing clear explanations is crucial in interdisciplinary
research fields like bioinformatics, where non-experts in machine learning
must understand model decisions to foster trust in the system. This work
introduces an explainable AI approach for compound potency prediction that
combines decision tree models with rule exploration and topic modelling.
The method demonstrates its...
Detecting temporal abnormal patterns over streaming data is challenging due to volatile data properties and the lack of real-time labels. The abnormal patterns are usually hidden in the temporal context, which cannot be detected by evaluating single points. Furthermore, the normal state evolves over time due to concept drifts. A single model does not fit all data over time. Autoencoders are...
While many have analyzed the resource efficiency of trained models, an important question remains: How can one be sustainable and resource-aware during AI development, or in other words, when looking for a suitable model to train on a specific learning task? AutoML can help with finding well-performing models on given data, however these frameworks overly focus on predictive quality and...
Pallets are one of the most important load carriers for international supply chains. Yet, continuously tracking activities such as driving, lifting or standing along their life cycle is hardly possible. As part of a preliminary project, it was shown that it is possible to develop a prediction model for pallet activities using data from inertial measurements units mounted on a pallet. A...
Automatic medical coding has the potential to ease documentation and billing processes. For this task, transparency plays an important role for medical coders and regulatory bodies, which can be achieved using explainability methods. However, the evaluation of these approaches has been mostly limited to short text and binary settings due to a scarcity of annotated data. Recent efforts by Cheng...
Human motion is of interest to industrial simulations for process optimisation, ergonomic evaluation and visualisation of digital-twin environments. Furthermore, it is of interest for simulation-based reinforcement learning in human-robot interaction and humanoid robotics for industrial scenarios. Human motion data created algorithmically lacks the variability and naturalness of real human...
