9607 research outputs found
Sort by
The Unseen Hand: How User Background Shapes AI Text-to-Image Generation
Get PDFText-To-Image (TTI) generators are becoming widely used and are often promoted as "democratizing" the making of images independently of the skill set of the user. But relatively little is known about whether people with different educations, skill sets and literacies use these tools differently, and how their backgrounds influence the quality of the results. In this article we investigate the impact of Visual Literacy (VL), AI Literacy (AIL) and Prompt Engineering Literacy (PEL) on prompt use. More precisely we are examining how they eachimpact prompt usage patterns, linguistic and semantic composition, and the structure and morphology of prompts. Additionally, we employed Process Analysis to investigate how participants with different literacyapproach the creative design process with TTI. Our results show that individuals scoring high on Visual Literacy (VL) tend to employ a larger vocabulary and more nuanced language, greater prompt variety and make more references to art genres, styles, and movements. In contrast, participants scoring high AI Literacy (AIL) demonstrated a deeper understanding of AI interaction, influencing their prompting strategies, and tended to adhere closely to the exact wording of the brief, treating it as precise specifications. The study's findings have the potential of significantly impacting educations, both by actively teaching students to experiment with prompt variations and deepen expressive visual vocabulary as well as designing curricula that incorporate structured exercises on iterative prompt refinement, explicitly teaching how to adjust prompts based on AIoutput to achieve desired visual outcomes
Explaining Model-Based Systems Engineering – Towards a Semiotic Perspective
No full textWhile the benefits of Model-Based System Engineering (MBSE) are recognized, its adoption is hampered by seemingly pragmatic problems, such as a harsh learning curve and lack of mature tooling and integration. While these perceived drawbacks are often described, they are only rarely explained – it is unclear what exactly about MBSE is challenging to use and learn, and how to determine whether new approaches mitigate this challenge. It this work, we propose a theory for explanation for three observations: (1) MBSE is more appreciated by non-system engineers, than by those that create the models, (2) system models, particularly in the early stage, are used for communication and rarely for automation, and (3) graphical notation is seen as both the major advantage of MBSE and a big drawback when learning it. We use a cognitive framework based on semiotics, conceptual spaces and naturalness and provide a first explanation of these observations: A graphical model can be interpreted either in terms of the domain it models, or in terms of the language it is expressed in. These two interpretations are in parallel when a user interprets a model and can interfere, which leads to problems in understanding
Fifth Workshop on Recommender Systems for Human Resources (RecSys in HR 2025)
No full textProceedings of the 5th Workshop on Recommender Systems for Human Resources (RecSys-in-HR 2025) co-located with the 19th ACM Conference on Recommender Systems (RecSys 2025
CXL Memory Performance for In-Memory Data Processing
No full textThe Compute Express Link (CXL) standard enables new forms of memory management and access across devices and servers. Based on PCIe, it enables cache-coherent access to remote memory. This widens the design space for database systems by expanding the available memory beyond memory local to the CPU. Efficiently utilizing CXL-attached memory requires conscious decisions by data systems about data placement and management. In this paper, we provide an in-depth analysis of database operation performance with data interleaved across multiple CXL memory devices. We experimentally evaluate the memory access performance for basic access patterns, the performance impact of placing data across multiple CXL memory devices for in-memory column scans and in-memory B+tree operations, and the performance impact of placing data in CXL memory for an in-memory database system when running the analytical TPC-H workload. Our experiments show that access to CXL-attached memory does not have to penalize performance over local access, but careful workload-aware data management is required. Our TPC-H evaluation shows that placing table columns based on access frequencies allows storing over 80% of the table data in CXL memory with a performance of 85% of a local-memory-only solution
DP-Morph: Improving the Privacy-Utility-Performance Trade-off for Differentially Private OCT Segmentation
No full textOptical Coherence Tomography (OCT) images show a cross-section of the retina and are used for early detection of retinal diseases and glaucoma through analysis of the retinal layers. Advances in deep learning have enabled state-of-the-art OCT segmentation models to support this analysis. However, using medical data to train these models raises concerns about patient privacy. For example, membership inference attacks allow an adversary to determine whether a particular data point was included in the training data. Differentially Private Stochastic Gradient Descent (DPSGD) improves the privacy of deep learning models by ensuring that these models do not disclose sensitive information about individual data points. However, implementing DPSGD may cause decreased model accuracy and/or increased computational demands. In this paper, we evaluate the privacy, utility, and computational performance of five OCT segmentation models trained using DPSGD on graphics processing units (GPUs). To improve utility, we then propose DP-Morph, a novel privacy-preserving modification of DPSGD based on morphology. We show that DP-Morph improves segmentation performance, for example, increasing the Dice coefficient of LFUNet from 0.50 to 0.70 for a privacy budget of 200
Fast Deterministic Chromatic Number under the Asymptotic Rank Conjecture
No full textIn this paper we further explore the recently discovered connection by Björklund and Kaski [STOC 2024] and Pratt [STOC 2024] between the asymptotic rank conjecture of Strassen [Progr. Math. 1994] and the three-way partitioning problem. We show that under the asymptotic rank conjecture, the chromatic number of an n-vertex graph can be computed deterministically in O (1.99982n ) time, thus giving a conditional answer to a question of Zamir [ICALP 2021], and questioning the optimality of the 2n poly(n ) time algorithm for chromatic number by Björklund, Husfeldt, and Koivisto [SICOMP 2009].Viewed in the other direction, if chromatic number indeed requires deterministic algorithms to run in close to 2n time, we obtain a sequence of explicit tensors of superlinear rank, falsifying the asymptotic rank conjecture.Our technique is a combination of earlier algorithms for detecting k-colorings for small k and enumerating k-colorable subgraphs, with an extension and derandomisation of Pratt’s tensor-based algorithm for balanced three-way partitioning to the unbalanced case
AI-Driven Neighborhood Selection in Large Neighborhood Search for Representative Container Vessel Stowage Planning
No full textThe representative container vessel stowage planning problem (CSPP) is a large-scale combinatorial optimization challenge with numerous constraints and decision variables. Due to its complexity, exact solution methods often struggle to find feasible or optimal solutions within a reasonable timeframe, necessitating the use of heuristic approaches. A widely used metaheuristic is the large neighborhood search (LNS) framework, which has demonstrated effectiveness across various combinatorial optimization problems with a manageable number of neighborhoods. However, a full-featured CSPP requires a large set of neighborhoods, rendering conventional selection heuristics ineffective in identifying promising neighborhoods during the search. This reduces the performance of LNS, highlighting the need for more intelligent selection strategies.In this planned work, we leverage an AI-driven neighborhood selection heuristic within the LNS framework to solve representative instances of the CSPP. The search will be formulated as a Markov decision process, where state features capture solution characteristics, neighborhood selection serves as the action, stochastic transitions update solutions based on neighborhood operators, and rewards are based on objective value and feasibility satisfaction. Our AI-assisted LNS framework will be evaluated on real-life instances. Its performance will be compared to a baseline vanilla LNS to assess improvements in solution quality and computational costs
Tangles: Unpacking Extended Collision Experiences with Soma Trajectories
Get PDFWe reappraise the idea of colliding with robots, moving from a position that tries to avoid or mitigate collisions to one that considers them an important facet of human interaction. We report on a soma design workshop that explored how our bodies could collide with telepresence robots, mobility aids and a quadruped robot. Based on our findings, we employed soma trajectories to analyse collisions as extended experiences that negotiate key transitions of consent, preparation, launch, contact, ripple, sting, untangle, debris and reflect. We then employed these ideas to analyse two collision experiences, an accidental collision between a person and a drone and the deliberate design of a robot to play with cats, revealing how real-world collisions involve the complex and ongoing entanglement of soma trajectories. We discuss how viewing collisions as entangled trajectories, or ‘tangles’, can be used analytically, as a design approach, and as a lens to broach ethical complexity
The Roguelike as Poetic form
No full textThis article proposes an analysis of roguelikes as video game poetic form
Body Politics: Unpacking Tensions and Future Perspectives for Body-Centric Design Research in HCI
Get PDFHuman bodies are deeply political as they carry historical and social meanings, including race, gender, sexuality, ethnicity, class, and abilities. The expanding body-centric research in HCI can be traced in the plurality of methods, theories and domains that take bodies as a central point of departure, when designing or studying interaction with technologies. This one-day workshop will bring together researchers and practitioners within the CHI community to discuss, map, and unpack emerging tensions and challenges on the topic of body politics for HCI. Interested participants are invited to submit examples from their own research, which, in the workshop, will be used as a point of departure to critically reflect on and expand body-centric methods, theories and domains through the lens of body politics. Workshop outcomes will include charting future directions for body-centric research to address challenges and opportunities of acknowledging that bodies are always political in design research