arXiv subjects

Lars Krupp

Publications and source records attributed to Lars Krupp.

17 recordsLinked to original sources

arXiv2026

This Is Taking Too Long -- Investigating Time as a Proxy for Energy Consumption of LLMs

The energy consumption of Large Language Models (LLMs) is raising growing concerns due to their adverse effects on environmental stability and resource use. Yet, these energy costs remain largely opaque to users, especially when models are accessed through an API -- a black box in which all information depends on what providers choose to disclose. In this work, we investigate inference time measurements as a proxy to approximate the associated energy costs of API-based LLMs. We ground our approach by comparing our estimations with actual energy measurements from locally hosted equivalents. Our results show that time measurements allow us to infer GPU models for API-based LLMs, grounding our energy cost estimations. Our work aims to create means for understanding the associated energy costs of API-based LLMs, especially for end users.

cs.PF↗

arXiv2025

Promoting Sustainable Web Agents: Benchmarking and Estimating Energy Consumption through Empirical and Theoretical Analysis

Web agents, like OpenAI's Operator and Google's Project Mariner, are powerful agentic systems pushing the boundaries of Large Language Models (LLM). They can autonomously interact with the internet at the user's behest, such as navigating websites, filling search masks, and comparing price lists. Though web agent research is thriving, induced sustainability issues remain largely unexplored. To highlight the urgency of this issue, we provide an initial exploration of the energy and $CO_2$ cost associated with web agents from both a theoretical -via estimation- and an empirical perspective -by benchmarking. Our results show how different philosophies in web agent creation can severely impact the associated expended energy, and that more energy consumed does not necessarily equate to better results. We highlight a lack of transparency regarding disclosing model parameters and processes used for some web agents as a limiting factor when estimating energy consumption. Our work contributes towards a change in thinking of how we evaluate web agents, advocating for dedicated metrics measuring energy consumption in benchmarks.

cs.AI↗

arXiv2025

Beyond the Pocket: A Large-Scale International Study on User Preferences on Bodily Placements of Commercial Wearables

As wearables become smaller, more powerful, and increasingly embedded in everyday life, their integration into diverse user contexts raises important design challenges. Despite this, their placement is still largely informed by lab-based assumptions not grounded in real-world, context-specific use. It remains unclear whether the designs evaluated in controlled studies reflect users everyday needs, routines, and habits. To address this gap, we collect empirical data on how people carry wearables in their daily lives, beginning to systematically examine user preferences for wearable placement across contexts and routines. We developed a multilingual questionnaire to capture real-world wearable placement practices. Responses from n=300 participants recruited through typical research channels, reveal how wearable usage patterns vary with users. We propose a set of user-centred guidelines for sensor placement and discuss how they fit in assumptions seen in related work. This study contributes to ongoing efforts to design more inclusive, adaptable, and context-aware wearable systems.

cs.HC↗

arXiv2025

Visualization enhances Problem Solving in multi-Qubit Systems

Quantum Information Science (QIS) is a vast, diverse, and abstract field. In consequence, learners face many challenges. Science, Technology, Engineering, and Mathematics (STEM) education research has found that visualizations are valuable to aid learners in complex matters. The conditions under which visualizations pose benefits are largely unexplored in QIS education. In this eye-tracking study, we examine the conditions under which the visualization of multi-qubit systems with the Dimensional Circle Notation (DCN) in addition to the mathematical symbolic Dirac Notation (DN) is associated with a benefit for solving problems on the ubiquitously used Hadamard gate operation in terms of performance, Extraneous Cognitive Load (ECL) and Intrinsic Cognitive Load (ICL). We find that DCN increases performance and reduces cognitive load for participants with little experience in quantum physics. In addition, representational competence is able to predict reductions in ECL with DCN, but not performance or ICL. Analysis of the eye-tracking results indicates that task solvers with more transitions between DN and DCN benefit less from the visualization. We discuss the generalizability of the results and practical implications.

physics.ed-ph↗

arXiv2025

Human in the Latent Loop (HILL): Interactively Guiding Model Training Through Human Intuition

Latent space representations are critical for understanding and improving the behavior of machine learning models, yet they often remain obscure and intricate. Understanding and exploring the latent space has the potential to contribute valuable human intuition and expertise about respective domains. In this work, we present HILL, an interactive framework allowing users to incorporate human intuition into the model training by interactively reshaping latent space representations. The modifications are infused into the model training loop via a novel approach inspired by knowledge distillation, treating the user's modifications as a teacher to guide the model in reshaping its intrinsic latent representation. The process allows the model to converge more effectively and overcome inefficiencies, as well as provide beneficial insights to the user. We evaluated HILL in a user study tasking participants to train an optimal model, closely observing the employed strategies. The results demonstrated that human-guided latent space modifications enhance model performance while maintaining generalization, yet also revealing the risks of including user biases. Our work introduces a novel human-AI interaction paradigm that infuses human intuition into model training and critically examines the impact of human intervention on training strategies and potential biases.

cs.LG↗

arXiv2025

TxP: Reciprocal Generation of Ground Pressure Dynamics and Activity Descriptions for Improving Human Activity Recognition

Sensor-based human activity recognition (HAR) has predominantly focused on Inertial Measurement Units and vision data, often overlooking the capabilities unique to pressure sensors, which capture subtle body dynamics and shifts in the center of mass. Despite their potential for postural and balance-based activities, pressure sensors remain underutilized in the HAR domain due to limited datasets. To bridge this gap, we propose to exploit generative foundation models with pressure-specific HAR techniques. Specifically, we present a bidirectional Text$\times$Pressure model that uses generative foundation models to interpret pressure data as natural language. TxP accomplishes two tasks: (1) Text2Pressure, converting activity text descriptions into pressure sequences, and (2) Pressure2Text, generating activity descriptions and classifications from dynamic pressure maps. Leveraging pre-trained models like CLIP and LLaMA 2 13B Chat, TxP is trained on our synthetic PressLang dataset, containing over 81,100 text-pressure pairs. Validated on real-world data for activities such as yoga and daily tasks, TxP provides novel approaches to data augmentation and classification grounded in atomic actions. This consequently improved HAR performance by up to 12.4\% in macro F1 score compared to the state-of-the-art, advancing pressure-based HAR with broader applications and deeper insights into human movement.

cs.AI↗

arXiv2025

Talk2X -- An Open-Source Toolkit Facilitating Deployment of LLM-Powered Chatbots on the Web

Integrated into websites, LLM-powered chatbots offer alternative means of navigation and information retrieval, leading to a shift in how users access information on the web. Yet, predominantly closed-sourced solutions limit proliferation among web hosts and suffer from a lack of transparency with regard to implementation details and energy efficiency. In this work, we propose our openly available agent Talk2X leveraging an adapted retrieval-augmented generation approach (RAG) combined with an automatically generated vector database, benefiting energy efficiency. Talk2X's architecture is generalizable to arbitrary websites offering developers a ready to use tool for integration. Using a mixed-methods approach, we evaluated Talk2X's usability by tasking users to acquire specific assets from an open science repository. Talk2X significantly improved task completion time, correctness, and user experience supporting users in quickly pinpointing specific information as compared to standard user-website interaction. Our findings contribute technical advancements to an ongoing paradigm shift of how we access information on the web.

cs.AI↗

arXiv2025

Towards Sustainable Web Agents: A Plea for Transparency and Dedicated Metrics for Energy Consumption

Improvements in the area of large language models have shifted towards the construction of models capable of using external tools and interpreting their outputs. These so-called web agents have the ability to interact autonomously with the internet. This allows them to become powerful daily assistants handling time-consuming, repetitive tasks while supporting users in their daily activities. While web agent research is thriving, the sustainability aspect of this research direction remains largely unexplored. We provide an initial exploration of the energy and CO2 cost associated with web agents. Our results show how different philosophies in web agent creation can severely impact the associated expended energy. We highlight lacking transparency regarding the disclosure of model parameters and processes used for some web agents as a limiting factor when estimating energy consumption. As such, our work advocates a change in thinking when evaluating web agents, warranting dedicated metrics for energy consumption and sustainability.

cs.AI↗

arXiv2024

GenAI Assisting Medical Training

Medical procedures such as venipuncture and cannulation are essential for nurses and require precise skills. Learning this skill, in turn, is a challenge for educators due to the number of teachers per class and the complexity of the task. The study aims to help students with skill acquisition and alleviate the educator's workload by integrating generative AI methods to provide real-time feedback on medical procedures such as venipuncture and cannulation.

cs.AI↗

arXiv2024

LLM-Generated Tips Rival Expert-Created Tips in Helping Students Answer Quantum-Computing Questions

Individual teaching is among the most successful ways to impart knowledge. Yet, this method is not always feasible due to large numbers of students per educator. Quantum computing serves as a prime example facing this issue, due to the hype surrounding it. Alleviating high workloads for teachers, often accompanied with individual teaching, is crucial for continuous high quality education. Therefore, leveraging Large Language Models (LLMs) such as GPT-4 to generate educational content can be valuable. We conducted two complementary studies exploring the feasibility of using GPT-4 to automatically generate tips for students. In the first one students (N=46) solved four multiple-choice quantum computing questions with either the help of expert-created or LLM-generated tips. To correct for possible biases towards LLMs, we introduced two additional conditions, making some participants believe that they were given expert-created tips, when they were given LLM-generated tips and vice versa. Our second study (N=23) aimed to directly compare the LLM-generated and expert-created tips, evaluating their quality, correctness and helpfulness, with both experienced educators and students participating. Participants in our second study found that the LLM-generated tips were significantly more helpful and pointed better towards relevant concepts than the expert-created tips, while being more prone to be giving away the answer. While participants in the first study performed significantly better in answering the quantum computing questions when given tips labeled as LLM-generated, even if they were created by an expert. This phenomenon could be a placebo effect induced by the participants' biases for LLM-generated content. Ultimately, we find that LLM-generated tips are good enough to be used instead of expert tips in the context of quantum computing basics.

cs.HC↗

arXiv2024

Visualizing Quantum States: A Pilot Study on Problem Solving in Quantum Information Science Education

In the rapidly evolving interdisciplinary field of quantum information science and technology, a major obstacle is the need to understand advanced mathematics to solve complex problems. Current findings in educational research suggest that incorporating visualizations into problem-solving settings can have beneficial effects on students' performance and cognitive load compared to relying solely on symbolic problem-solving content. Visualizations like the (dimensional) circle notation enable us to represent not only single-qubit but also more complex multi-qubit states, entanglement, and quantum algorithms. In this pilot study, we aim to take an initial step toward identifying the contexts in which students benefit from the presentation of visualizations of single- and multi-qubit systems in addition to mathematical formalism. For this purpose, we propose a set of test items and a comprehensive methodology to assess students' performance and cognitive load when solving problems. This is a pilot investigation with a large breadth of questions intended to generate hypotheses and guide larger-scale, more focused studies in the future. Specifically, we compare two approaches: using the mathematical-symbolic Dirac Notation alone and using it in combination with the (dimensional) circle notation. In surveys in one-, two- and three-qubit systems, we gather qualitative data from five, five and two think-aloud interviews, identifying problems that students encounter and their problem-solving strategies. In addition, we analyze quantitative data (performance and cognitive load) from 23, 27 and 17 participants in surveys on one-, two- and three-qubit systems recruited mainly from our quantum computing lectures. We find that most of the test items are appropriate for a heterogeneous target group, as they can differentiate between participants in terms of performance and time taken...

quant-ph↗

arXiv2024

Text me the data: Generating Ground Pressure Sequence from Textual Descriptions for HAR

In human activity recognition (HAR), the availability of substantial ground truth is necessary for training efficient models. However, acquiring ground pressure data through physical sensors itself can be cost-prohibitive, time-consuming. To address this critical need, we introduce Text-to-Pressure (T2P), a framework designed to generate extensive ground pressure sequences from textual descriptions of human activities using deep learning techniques. We show that the combination of vector quantization of sensor data along with simple text conditioned auto regressive strategy allows us to obtain high-quality generated pressure sequences from textual descriptions with the help of discrete latent correlation between text and pressure maps. We achieved comparable performance on the consistency between text and generated motion with an R squared value of 0.722, Masked R squared value of 0.892, and FID score of 1.83. Additionally, we trained a HAR model with the the synthesized data and evaluated it on pressure dynamics collected by a real pressure sensor which is on par with a model trained on only real data. Combining both real and synthesized training data increases the overall macro F1 score by 5.9 percent.

cs.LG↗

arXiv2024

Lessons Learned from Designing an Open-Source Automated Feedback System for STEM Education

As distance learning becomes increasingly important and artificial intelligence tools continue to advance, automated systems for individual learning have attracted significant attention. However, the scarcity of open-source online tools that are capable of providing personalized feedback has restricted the widespread implementation of research-based feedback systems. In this work, we present RATsApp, an open-source automated feedback system (AFS) that incorporates research-based features such as formative feedback. The system focuses on core STEM competencies such as mathematical competence, representational competence, and data literacy. It also allows lecturers to monitor students' progress. We conducted a survey based on the technology acceptance model (TAM2) among a set of students (N=64). Our findings confirm the applicability of the TAM2 framework, revealing that factors such as the relevance of the studies, output quality, and ease of use significantly influence the perceived usefulness. We also found a linear relation between the perceived usefulness and the intention to use, which in turn is a significant predictor of the frequency of use. Moreover, the formative feedback feature of RATsApp received positive feedback, indicating its potential as an educational tool. Furthermore, as an open-source platform, RATsApp encourages public contributions to its ongoing development, fostering a collaborative approach to improve educational tools.

cs.CY↗

arXiv2023

Challenges and Opportunities of Moderating Usage of Large Language Models in Education

The increased presence of large language models (LLMs) in educational settings has ignited debates concerning negative repercussions, including overreliance and inadequate task reflection. Our work advocates moderated usage of such models, designed in a way that supports students and encourages critical thinking. We developed two moderated interaction methods with ChatGPT: hint-based assistance and presenting multiple answer choices. In a study with students (N=40) answering physics questions, we compared the effects of our moderated models against two baseline settings: unmoderated ChatGPT access and internet searches. We analyzed the interaction strategies and found that the moderated versions exhibited less unreflected usage (e.g., copy \& paste) compared to the unmoderated condition. However, neither ChatGPT-supported condition could match the ratio of reflected usage present in internet searches. Our research highlights the potential benefits of moderating language models, showing a research direction toward designing effective AI-supported educational strategies.

cs.HC↗

arXiv2023

Unreflected Acceptance -- Investigating the Negative Consequences of ChatGPT-Assisted Problem Solving in Physics Education

Large language models (LLMs) have recently gained popularity. However, the impact of their general availability through ChatGPT on sensitive areas of everyday life, such as education, remains unclear. Nevertheless, the societal impact on established educational methods is already being experienced by both students and educators. Our work focuses on higher physics education and examines problem solving strategies. In a study, students with a background in physics were assigned to solve physics exercises, with one group having access to an internet search engine (N=12) and the other group being allowed to use ChatGPT (N=27). We evaluated their performance, strategies, and interaction with the provided tools. Our results showed that nearly half of the solutions provided with the support of ChatGPT were mistakenly assumed to be correct by the students, indicating that they overly trusted ChatGPT even in their field of expertise. Likewise, in 42% of cases, students used copy & paste to query ChatGPT -- an approach only used in 4% of search engine queries -- highlighting the stark differences in interaction behavior between the groups and indicating limited reflection when using ChatGPT. In our work, we demonstrated a need to (1) guide students on how to interact with LLMs and (2) create awareness of potential shortcomings for users.

physics.ed-ph↗

arXiv2023

A Synthetic Benchmarking Pipeline to Compare Camera Calibration Algorithms

Accurate camera calibration is crucial for various computer vision applications. However, measuring calibration accuracy in the real world is challenging due to the lack of datasets with ground truth to evaluate them. In this paper, we present SynthCal, a synthetic camera calibration benchmarking pipeline that generates images of calibration patterns to measure and enable accurate quantification of calibration algorithm performance in camera parameter estimation. We present a SynthCal generated calibration dataset with four common patterns, two camera types, and two environments with varying view, distortion, lighting, and noise levels for both monocular and multi-camera systems. The dataset evaluates both single and multi-view calibration algorithms by measuring re-projection and root-mean-square errors for identical patterns and camera settings. Additionally, we analyze the significance of different patterns using different calibration configurations. The experimental results demonstrate the effectiveness of SynthCal in evaluating various calibration algorithms and patterns.

cs.CV↗

arXiv2023

Visualizing Entanglement in multi-Qubit Systems

In the field of quantum information science and technology, the representation and visualization of quantum states and related processes are essential for both research and education. In this context, a focus especially lies on ensembles of few qubits. There exist many powerful representations for single-qubit and multi-qubit systems, such as the famous Bloch sphere and generalizations. Here, we utilize the dimensional circle notation as a representation of such ensembles, adapting the so-called circle notation of qubits and the idea of representing the n-particle system in an n-dimensional space. We show that the mathematical conditions for separability lead to symmetry conditions of the quantum state visualized, offering a new perspective on entanglement in few-qubit systems and therefore on various quantum algorithms. In this way, dimensional notations promise significant potential for conveying nontrivial quantum entanglement properties and processes in few-qubit systems to a broader audience, and could enhance understanding of these concepts as a bridge between intuitive quantum insight and formal mathematical descriptions.

quant-ph↗