1. LLM in the loop: a framework for contextualizing counterfactual segment perturbations in point cloudsVeljka Kočić, Niko Lukač, Dzemail Rozajac, Stefan Schweng, Christoph Gollob, Arne Nothdurft, Karl Stampfer, Javier Del Ser, Andreas Holzinger, 2025, izvirni znanstveni članek Opis: Point Cloud Data analysis has seen a major leap forward with the introduction of PointNet algorithms, revolutionizing how we process 3D environments. Yet, despite these advancements, key challenges remain, particularly in optimizing segment perturbations to influence model outcomes in a controlled and meaningful way. Traditional methods struggle to generate realistic and contextually appropriate perturbations, limiting their effectiveness in critical applications like autonomous systems and urban planning. This paper takes a bold step by integrating Large Language Models into the counterfactual reasoning process, unlocking a new level of automation and intelligence in segment perturbation. Our approach begins with semantic segmentation, after which LLMs intelligently select optimal replacement segments based on features such as class label, color, area, and height. By leveraging the reasoning capabilities of LLMs, we generate perturbations that are not only computationally efficient but also semantically meaningful. The proposed framework undergoes rigorous evaluation, combining human inspection of LLM-generated suggestions with quantitative analysis of semantic classification model performance across different LLM variants. By bridging the gap between geometric transformations and high-level semantic reasoning, this research redefines how we approach perturbation generation in Point Cloud Data analysis. The results pave the way for more interpretable, adaptable, and intelligent AI-driven solutions, bringing us closer to realworld applications where explainability and robustness are paramount.
Ključne besede: explainable AI, point cloud data, counterfactual reasoning, LiDAR, 3D point cloud data, interpretability, human-centered AI, large language models, K-nearest neighbors
Objavljeno v DKUM: 19.05.2025; Ogledov: 0; Prenosov: 2
 Celotno besedilo (7,24 MB) |
2. Enhancing trust in automated 3D point cloud data interpretation through explainable counterfactualsAndreas Holzinger, Niko Lukač, Dzemail Rozajac, Emil Johnston, Veljka Kočić, Bernhard Hoerl, Christoph Gollob, Arne Nothdurft, Karl Stampfer, Stefan Schweng, Javier Del Ser, 2025, izvirni znanstveni članek Opis: This paper introduces a novel framework for augmenting explainability in the interpretation of point cloud data by fusing expert knowledge with counterfactual reasoning. Given the complexity and voluminous nature of point cloud datasets, derived predominantly from LiDAR and 3D scanning technologies, achieving interpretability remains a significant challenge, particularly in smart cities, smart agriculture, and smart forestry. This research posits that integrating expert knowledge with counterfactual explanations – speculative scenarios illustrating how altering input data points could lead to different outcomes – can significantly reduce the opacity of deep learning models processing point cloud data. The proposed optimization-driven framework utilizes expert-informed ad-hoc perturbation techniques to generate meaningful counterfactual scenarios when employing state-of-the-art deep learning architectures. The optimization process minimizes a multi-criteria objective comprising counterfactual metrics such as similarity, validity, and sparsity, which are specifically tailored for point cloud datasets. These metrics provide a quantitative lens for evaluating the interpretability of the counterfactuals. Furthermore, the proposed framework allows for the definition of explicit interpretable counterfactual perturbations at its core, thereby involving the audience of the model in the counterfactual generation pipeline and ultimately, improving their overall trust in the process. Results demonstrate a notable improvement in both the interpretability of the model’s decisions and the actionable insights delivered to end-users. Additionally, the study explores the role of counterfactual reasoning, coupled with expert input, in enhancing trustworthiness and enabling human-in-the-loop decision-making processes. By bridging the gap between complex data interpretations and user comprehension, this research advances the field of explainable AI, contributing to the development of transparent, accountable, and human-centered artificial intelligence systems.
Ključne besede: explainable AI, point cloud data, counterfactual reasoning, information fusion, interpretability, human-centered AI
Objavljeno v DKUM: 06.03.2025; Ogledov: 0; Prenosov: 4
 Celotno besedilo (186,97 KB) |
3. Clinical reasoning needs to be explicitly addressed in health professions curricula : recommendations from a European consortiumIoannis Parodis, Lina Andersson, Steven J. Durning, Inga Hege, Jure Knez, Andrzej Kononowicz, Marie Lidskog, Tadej Petreski, Magdalena Szopa, Samuel Edelbring, 2021, izvirni znanstveni članek Opis: Clinical reasoning entails the application of knowledge and skills to collect and integrate information, typically with the goal of arriving at a diagnosis and management plan based on the patient’s unique circumstances and preferences. Evidence-informed, structured, and explicit teaching and assessment of clinical reasoning in educational programs of medical and other health professions remain unmet needs. We herein summarize recommendations for clinical reasoning learning objectives (LOs), as derived from a consensus approach among European and US researchers and health professions educators. A four-step consensus approach was followed: (1) identification of a convenience sample of the most relevant and applied national LO catalogues for health professions educational programs (N = 9) from European and US countries, (2) extraction of LOs related to clinical reasoning and translation into English, (3) mapping of LOs into predefined categories developed within the Erasmus+ Developing, implementing, and disseminating an adaptive clinical reasoning curriculum for healthcare students and educators (DID-ACT) consortium, and (4) synthesis of analysis findings into recommendations for how LOs related to clinical reasoning could be presented and incorporated in LO catalogues, upon consensus. Three distinct recommendations were formulated: (1) make clinical reasoning explicit, (2) emphasize interprofessional and collaboration aspects of clinical reasoning, and (3) include aspects of teaching and assessment of clinical reasoning. In addition, the consortium understood that implementation of bilingual catalogues with English as a common language might contribute to lower heterogeneity regarding amount, structure, and level of granularity of clinical reasoning LOs across countries. These recommendations will hopefully motivate and guide initiatives towards the implementation of LOs related to clinical reasoning in existing and future LO catalogues.
Ključne besede: clinical reasoning, curriculum development, curriculum mapping, health professions education, medical education
Objavljeno v DKUM: 08.10.2024; Ogledov: 0; Prenosov: 4
Celotno besedilo (323,61 KB)
Gradivo ima več datotek! Več... |
4. |
5. Metacognitive accuracy and learning to learn : a developmental perspectiveKarin Bakračevič, 2012, izvirni znanstveni članek Opis: Metacognition belongs to higher-order mental processes and enables us to control, plan and accordingly regulate our own learning and problem solving process. In the present study we researched developmental changes in different reasoning domains and in metacognitive accuracy, which is considered as part of successful metacognitive monitoring/regulation, and as an essential element of self-regulated learning and learning to learn competence. The study involved 282 participants from four different age groups: 13-15-, 23-25-, 33-35- and 43-45- year olds. These participants solved tasks addressed to spatial, verbal-propositional and social reasoning, and evaluated their own performance on these tasks. To specify possible differences in metacognitive accuracy, the metacognitive accuracy index was computed. Results showed that metacognitive evaluations were accurate in spatial domain, less accurate in verbal-propositional and quite inaccurate in the social domain. The accuracy of self-evaluation increased with age and males were more accurate in their self-evaluations than females. Improvement of metacognitive accuracy with age is in tune with findings that metacognition becomes more effective with development and that people with age become more reflective and self-aware.
Ključne besede: reasoning, metacognition, metacognitive accuracy, self-regulated learning
Objavljeno v DKUM: 15.12.2017; Ogledov: 1635; Prenosov: 125
Celotno besedilo (488,20 KB)
Gradivo ima več datotek! Več... |
6. |
7. Reasoning and self-awareness from adolescence to middle age : organization and development as a function of educationAndreas Demetriou, Karin Bakračevič, 2009, izvirni znanstveni članek Opis: This study involved four age groups: 13-15-, 23-25-, 33-35-, and 43-45-yr-olds. All adult groups involved persons with university education andpersons with low education. Participants (1) solved tasks addressed to spatial, propositional, and social reasoning, (2) evaluated their own performance and the difficulty of the tasks, and (3) answered an inventory probing their self-concept for these reasoning domains and for self-awareness and self-regulation. Structural modeling revealed that performance, self-evaluation, and self-representation are systematically interrelated. Performance in spatial and propositional reasoning stabilized in early adulthood, whereas in social reasoning and self-evaluation, performance improved throughout the age span studied. Educated persons performed better and rated themselves accordingly across all domains. The implications of these findings for the general theory of intelligence and cognitive developmentafter adolescence are discussed. The functional shift model is proposed to account for changes in the relative power of different abilities with increasing age.
Ključne besede: education, developmental psychology, cognitive processes, self-awareness, reasoning
Objavljeno v DKUM: 07.06.2012; Ogledov: 2251; Prenosov: 104
Povezava na celotno besedilo |
8. Multimedia learning material in pedagogical methodology and problem solving strategiesTomaž Bratina, 2012, samostojni znanstveni sestavek ali poglavje v monografski publikaciji Opis: Problem solving is a reasoning process oriented to reach the final state which usually means the solution. To reach the final state certain mental steps are required. More precisely the process of problem solving can be described as the systematic sequence of cognitive steps toward the solution or conclusion. The sequence itself is the strategy. Regarding the kind of the problems or its nature diverse problem solving strategies can be applied. In the present literature the problem solving strategies are described as reasoning strategies. The application of multimedia learning materials and good practices are confirming their benefits. The students using multimedia learning materials achieved better results than students learning from the textbook and/or attended regular lectures. They are also more successful in the application of the reasoning strategies. In the teaching of pedagogical methodology the application of reasoning strategies is important in achieving higher level of understanding and successful implementation of statistical outcomes in the education. The article will present the results of research in assessing the influence of the multimedia learning materials to the application of reasoning strategies.
Ključne besede: e-learning materials, problem solving, reasoning strategies, pedagogical methodology, statistics
Objavljeno v DKUM: 07.06.2012; Ogledov: 2125; Prenosov: 39
Povezava na celotno besedilo |
9. |
