Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 35
Encontrada(s) 1 página(s)

Dipòsit Digital de Documents de la UAB
Publicación Artículo científico (JournalArticle)

Antibiotic-loaded bone cement is associated with a reduction of the risk of revision of total knee arthroplasty, Analysis of the Catalan Arthroplasty Register

  • Gil González, Sergi|||0000-0001-5939-0769
  • Velasco-Regúlez, Borja|||0000-0003-4718-3388
  • Cerquides Bueno, Jesus|||0000-0002-3752-644X
  • Hinarejos Gómez, Pedro|||0000-0001-7567-1317
  • Monllau García, Joan Carles|||0000-0002-2418-4874
  • Pelfort, Xavier|||0000-0001-9876-9894
The purpose of this study was to analyse the impact on peri-prosthetic joint infection (PJI) rate and prosthetic survival using antibiotic-loaded bone cement (ALBC) versus plain cement during total knee arthroplasty (TKA). A retrospective cohort study was conducted. The main data source was the Catalan Arthroplasty Register (RACat). TKAs with surgery date between 1 January 2011 and 31 December 2020 were analysed and followed up until 31 December 2023. The main variable of interest was the type of cement (ALBC vs. plain cement), and several endpoints (septic revision, aseptic revision, and all-cause revision) were considered. The analysed outcomes were revision rates, survival rates and risk factors' hazard ratios (HR). A total of 22,781 TKAs were analysed, 13,125 (57.6%) of them with plain cement and 9656 (42.4%) with ALBC. The septic revision rate was lower in the ALBC group after 3 months of follow-up (0.52% vs. 0.78%, p value = 0.04). Prosthetic survival with respect to the aseptic revision endpoint was also higher for the ALBC group during the whole follow-up period (~158 months). Regarding risk factors for infection, ALBC showed a protective effect, HR: 0.53 (0.44, 0.63), while sex (being male) and the analysed comorbidities increased the risk. ALBC is associated with a reduction in both the septic revision and the aseptic revision rate after TKA, and thus with higher prosthetic survival. Level III, Therapeutic, retrospective.
Dipòsit Digital de Documents de la UAB
Publicación Artículo científico (JournalArticle)

The first causal inference analysis of the Catalan Arthroplasty Register shows a positive effect of antibiotic-loaded bone cement on knee prosthesis survival

  • Velasco-Regúlez, Borja|||0000-0003-4718-3388
  • Gil González, Sergi|||0000-0001-5939-0769
  • Cerquides Bueno, Jesus|||0000-0002-3752-644X
The survival of a knee prosthesis is one of the most important indicators of the success or failure of a knee arthroplasty. An intervention that could increase prosthetic survival is the use of antibiotic-loaded bone cement (ALBC) during primary surgery, but the evidence for this is not conclusive. The question of whether such an intervention increases prosthetic survival is a causal one, and yet it has never been addressed with causal methods in the observational studies literature. This constitutes a serious limitation, as there is growing evidence that the best-suited framework for addressing causal questions with observational data is causal inference. In the present study, causal inference methods were employed to answer the research question of whether ALBC increases prosthetic survival. In particular, directed acyclic graphs were used for identification and causal survival forests were used to estimate the effect of interest. The rationale behind these methods is provided in the main text, and technical details are provided in Supporting Information: File S3. Data from the Catalan Arthroplasty Register were analysed. ALBC had an effect of increasing the overall prosthetic survival by 8% after 120 months of follow-up. The intervention had a positive effect across all the subgroups of the population defined by confounding variables, but the effect was greater in men, young patients, patients with rheumatoid arthritis or obesity, and patients who smoked or abused alcohol. The chosen causal assumptions had an impact on the obtained results, empirically showing the importance of using a causal framework. ALBC increased knee prosthesis survival among patients in the Catalan public healthcare system. Causal inference methods are the most appropriate for answering causal questions about the effect of ALBC on prosthetic survival when the analysed data are observational. Level III.
e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publicación Artículo científico (JournalArticle)

Modeling opinion misperception and the emergence of silence in online social system

  • Vilone, Daniele
  • Polizzi di Sorrentino, Eugenia
In the last decades an increasing deal of research has investigated the phenomenon of opinion misperception in human communities and, more recently, in social media. Opinion misperception is the wrong evaluation by community's members of the real distribution of opinions or beliefs about a given topic. In this work we explore the mechanisms giving rise to opinion misperception in social media groups, which are larger than physical ones and have peculiar topological features. By means of numerical simulations, we suggest that the structure of connections of such communities plays indeed a role in distorting the perception of the agents about others' beliefs, but it is essentially an indirect effect. Moreover, we show that the main ingredient that generates misperception is a spiral of silence induced by few, well connected and charismatic agents, which rapidly drives the majority of individuals to stay silent without disclosing their true belief, leading minoritarian opinions to appear more widespread throughout the community., Daniele Vilone was partially supported by project SERICS (PE00000014) under the MUR National Recovery and Resilience Plan funded by the European Union NextGenerationEU. Eugenia Polizzi was partially supported by the EU H2020 ICT48 project "Humane AI Net" under contract n 952026. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publicación Artículo científico (JournalArticle)

Bad polarization in structurally cohesive communities

  • Michelini, Matteo
  • Polizzi, Eugenia
  • Vilone, Daniele
Communities frequently experience belief polarization, even in the face of overwhelming scientific evidence supporting one side of the debate. Current explanations for this phenomenon, which we define as bad polarization, attribute its emergence to the influence of social incentives on belief formation. However, these explanations presuppose the existence of a fragmented community in which opposing groups develop different beliefs. Here, we provide a model of belief formation in which social incentives bring about bad polarization even in structurally cohesive communities. We assume agents to distribute a finite amount of social support among their like-minded neighbors and to sample evidence partially in order to form the belief that grants them the highest possible support. Accordingly, we show that bad polarization emerges more frequently when communities are highly connected, and that bad believers, individuals holding unsupported beliefs, are consistently a minority. Bad polarization is driven by the competition for social support: bad believers form a minority because this allows them to gain a higher amount of support than if they adhered to the majority view., Open Access funding enabled and organized by Projekt DEAL. Daniele Vilone was partially supported by project SERICS (PE00000014) under the MUR National Recovery and Resilience Plan funded by the European Union NextGenerationEU. Eugenia Polizzi was partially supported by the EU H2020 ICT48 project ‘Humane AI Net’ under contract n 952026.
Repositorio Digital de la UPF
Publicación Artículo científico (JournalArticle)

Antibiotic-loaded bone cement is associated with a reduction of the risk of revision of total knee arthroplasty: Analysis of the Catalan Arthroplasty Register

  • Gil-Gonzalez, Sergi
  • Velasco-Regúlez, Borja
  • Cerquides, Jesus
  • Hinarejos Gómez, Pedro
  • Monllau García, Juan Carlos
  • Pelfort López, Javier
Purpose: The purpose of this study was to analyse the impact on peri-prosthetic joint infection (PJI) rate and prosthetic survival using antibiotic-loaded bone cement (ALBC) versus plain cement during total knee arthroplasty (TKA). Methods: A retrospective cohort study was conducted. The main data source was the Catalan Arthroplasty Register (RACat). TKAs with surgery date between 1 January 2011 and 31 December 2020 were analysed and followed up until 31 December 2023. The main variable of interest was the type of cement (ALBC vs. plain cement), and several endpoints (septic revision, aseptic revision, and all-cause revision) were considered. The analysed outcomes were revision rates, survival rates and risk factors' hazard ratios (HR). Results: A total of 22,781 TKAs were analysed, 13,125 (57.6%) of them with plain cement and 9656 (42.4%) with ALBC. The septic revision rate was lower in the ALBC group after 3 months of follow-up (0.52% vs. 0.78%, p value = 0.04). Prosthetic survival with respect to the aseptic revision endpoint was also higher for the ALBC group during the whole follow-up period (~158 months). Regarding risk factors for infection, ALBC showed a protective effect, HR: 0.53 (0.44, 0.63), while sex (being male) and the analysed comorbidities increased the risk. Conclusions: ALBC is associated with a reduction in both the septic revision and the aseptic revision rate after TKA, and thus with higher prosthetic survival. Level of evidence: Level III, Therapeutic, retrospective., This work was supported by the Ministerio de Ciencia e Innovación (projects PID2022-136787NB-I00 and PID2019-104156GB-I00), the European Commission H2020 program (grants 101060693, 101057693, 952026) and the Agència de Gestió d'Ajuts Universitaris i de Recerca (Award Number: 2020-18). None of the funding sources had a direct role in this work.
Repositorio Digital de la UPF
Publicación Artículo científico (JournalArticle)

Human-AI coevolution

  • Pedreschi, Dino
  • Pappalardo, Luca
  • Ferragina, Emanuele
  • Baeza Yates, Ricardo
  • Barabási, Albert-László
  • Dignum, Frank
  • Dignum, Virginia
  • Eliassi-Rad, Tina
  • Giannotti, Fosca
  • Kertész, János
  • Knott, Alistair
  • Ioannidis, Yannis
  • Lukowicz, Paul
  • Passarella, Andrea
  • Pentland, Alex
  • Shawe-Taylor, John
  • Vespignani, Alessandro
Human-AI coevolution, defined as a process in which humans and AI algorithms continuously influence each other, increasingly characterises our society, but is understudied in artificial intelligence and complexity science literature. Recommender systems and assistants play a prominent role in human-AI coevolution, as they permeate many facets of daily life and influence human choices through online platforms. The interaction between users and AI results in a potentially endless feedback loop, wherein users' choices generate data to train AI models, which, in turn, shape subsequent user preferences. This human-AI feedback loop has peculiar characteristics compared to traditional human-machine interaction and gives rise to complex and often “unintended” systemic outcomes. This paper introduces human-AI coevolution as the cornerstone for a new field of study at the intersection between AI and complexity science focused on the theoretical, empirical, and mathematical investigation of the human-AI feedback loop. In doing so, we: (i) outline the pros and cons of existing methodologies and highlight shortcomings and potential ways for capturing feedback loop mechanisms; (ii) propose a reflection at the intersection between complexity science, AI and society; (iii) provide real-world examples for different human-AI ecosystems; and (iv) illustrate challenges to the creation of such a field of study, conceptualising them at increasing levels of abstraction, i.e., scientific, legal and socio-political., This work has been partially supported by PNRR - M4C2 - Investimento 1.3, Partenariato Esteso (grant No. PE00000013) - “FAIR - Future Artificial Intelligence Research” - Spoke 1 “Human-centered AI”, funded by the European Commission under the Next Generation EU programme; EU H2020 projects HumaneAI-net G.A. 952026 and SoBigData++ G.A. 871042; European Research Council ERC-2018-ADG 834756 “XAI: Science and technology for the eXplanation of AI decision making”; CHIST-ERA grant CHIST-ERA-19-XAI-010, by MUR (grant No. not available), FWF (grant No. I 5205), EPSRC (grant No. EP/V055712/1), NCN (grant No. 2020/02/Y/ST6/00064), ETAg (grant No. SLTAT21096), BNSF (grant No. KΠ-06-AOO2/5); PNRR (Piano Nazionale di Ripresa e Resilienza) in the context of the research program 20224CZ5X4_PE6_PRIN 2022 “URBAI – Urban Artificial Intelligence” (CUP B53D23012770006), funded by European Union – Next Generation EU; Emanuele Ferragina has been partially supported by a public grant overseen by the French National Research Agency (ANR) as part of the ‘Investissements d'Avenir’ program LIEPP (ANR-11-LABX-0091, ANR-11-IDEX-0005-02) and the Université de Paris IdEx (ANR-18-IDEX-0001). We thank Vincenzo Vivarini for the inspiration about the importance of the feedback loop in complex social systems. There is always a human agent beyond any tactical system. We also thank Daniele Fadda for the support on making the figures. Finally, we thank all members of KDD-Lab for insightful discussions about human-AI coevolution.
Repositorio Digital de la UPF
Publicación Artículo científico (JournalArticle)

Generative AI models should include detection mechanisms as a condition for public release

  • Knott, Alistair
  • Pedreschi, Dino
  • Chatila, Raja
  • Chakraborti, Tapabrata
  • Leavy, Susan
  • Baeza Yates, Ricardo
  • Eyers, David
  • Trotman, Andrew
  • Teal, Paul D.
  • Biecek, Przemyslaw
  • Russell, Stuart
  • Bengio, Yoshua
The new wave of ‘foundation models’—general-purpose generative AI models, for production of text (e.g., ChatGPT) or images (e.g., MidJourney)—represent a dramatic advance in the state of the art for AI. But their use also introduces a range of new risks, which has prompted an ongoing conversation about possible regulatory mechanisms. Here we propose a specific principle that should be incorporated into legislation: that any organization developing a foundation model intended for public use must demonstrate a reliable detection mechanism for the content it generates, as a condition of its public release. The detection mechanism should be made publicly available in a tool that allows users to query, for an arbitrary item of content, whether the item was generated (wholly or partly) by the model. In this paper, we argue that this requirement is technically feasible and would play an important role in reducing certain risks from new AI models in many domains. We also outline a number of options for the tool’s design, and summarize a number of points where further input from policymakers and researchers would be required., We are grateful to the Global Partnership on AI (GPAI) and the Montreal International Center of Expertise in Artificial Intelligence (CEIMIA) for supporting this project. Yoshua Bengio thanks CIFAR and NSERC; Stuart Russell thanks Open Philanthropy Foundation for support of the Center for Human-Compatible AI at UC Berkeley. Dino Pedreschi has been supported by the European Commission under the NextGeneration Programme PE0013 - ‘FAIR - Future Artificial Intelligence Research’ Spoke 1, ‘Human-Centered AI’, and under the EU H2020 ICT-48 Network of Excellence n.952026 ‘Human-AI net’. We would like to thank colleagues at the EU (Lucilla Sioli’s group at DG-CNECT) and the OECD (Sebastian Hallensleben’s group) for useful discussions, and Marcin Betkier and Rebecca Downes for comments on an earlier draft of this article. We also thank the two anonymous reviewers for their useful comments. The views expressed here, and any remaining errors, are of course our own.
Dipòsit Digital de la UB
Publicación Artículo científico (JournalArticle)

A Conceptual Probabilistic Framework for Annotation Aggregation of Citizen Science Data

  • Cerquides Bueno, Jesús
  • Mülâyim, Mehmet Oguz
  • Hernández-González, Jerónimo
  • Shankar, Amudha Ravi
  • Fernández Márquez, Jose Luis
Over the last decade, hundreds of thousands of volunteers have contributed to science by collecting or analyzing data. This public participation in science, also known as citizen science, has contributed to significant discoveries and led to publications in major scientific journals. However, little attention has been paid to data quality issues. In this work we argue that being able to determine the accuracy of data obtained by crowdsourcing is a fundamental question and we point out that, for many real-life scenarios, mathematical tools and processes for the evaluation of data quality are missing. We propose a probabilistic methodology for the evaluation of the accuracy of labeling data obtained by crowdsourcing in citizen science. The methodology builds on an abstract probabilistic graphical model formalism, which is shown to generalize some already existing label aggregation models. We show how to make practical use of the methodology through a comparison of data obtained from different citizen science communities analyzing the earthquake that took place in Albania in 2019.
Dipòsit Digital de la UB
Publicación Artículo científico (JournalArticle)

A Robust Solution to Variational Importance Sampling of Minimum Variance

  • Hernández-González, Jerónimo
  • Cerquides Bueno, Jesús
Importance sampling is a Monte Carlo method where samples are obtained from an alternative proposal distribution. This can be used to focus the sampling process in the relevant parts of space, thus reducing the variance. Selecting the proposal that leads to the minimum variance can be formulated as an optimization problem and solved, for instance, by the use of a variational approach. Variational inference selects, from a given family, the distribution which minimizes the divergence to the distribution of interest. The Rényi projection of order 2 leads to the importance sampling estimator of minimum variance, but its computation is very costly. In this study with discrete distributions that factorize over probabilistic graphical models, we propose and evaluate an approximate projection method onto fully factored distributions. As a result of our evaluation it becomes apparent that a proposal distribution mixing the information projection with the approximate Rényi projection of order 2 could be interesting from a practical perspective.
Depósito Digital e-UCJC
Publicación Artículo científico (JournalArticle)

POTDAI: A Tool to Evaluate the Perceived Operational Trust Degree in Artificial Intelligence Systems

  • Martín Moncunill, David
  • García-Laredo, Eduardo
  • Nieves, Juan Carlos
Está demostrado que la confianza subjetiva de un usuario en un sistema basado en Inteligencia Artificial (IA) es crucial en su uso, incluso más decisiva que la eficacia y eficiencia objetivas del sistema. Por ello, se han propuesto diferentes métodos para analizar la confianza en la IA. En nuestra investigación, nos propusimos evaluar cómo el grado de confianza percibido en un sistema de IA podría afectar a la decisión final de un usuario de seguir las recomendaciones de la IA. Para ello, establecimos los criterios de confianza que debería cumplir dicha evaluación siguiendo un enfoque de cocreación con un grupo multidisciplinar de 10 expertos. Tras una revisión sistemática de 3.204 artículos, comprobamos que ninguna de las herramientas cumplía los criterios de inclusión. Así pues, presentamos la herramienta denominada ''Grado de Confianza Operativa Percibida en la IA'' (POTDAI) que se basa en las conclusiones del grupo de expertos y el análisis bibliográfico, con una metodología que añade rigor a la empleada anteriormente para crear herramientas de evaluación similares. Proponemos un cuestionario breve para una aplicación rápida y sencilla, inspirado en la versión original del Modelo de Aceptación de la Tecnología (TAM) con seis ítems tipo Likert. De esta forma, respondemos también a la necesidad señalada por autores como Vorm y Combs de ampliar el TAM para abordar cuestiones relacionadas con la percepción del usuario en sistemas con un componente de IA. Así, el POTDAI puede utilizarse solo o en combinación con el TAM para obtener información adicional sobre su utilidad y facilidad de uso., HumanE AI Network (HumanE-AI-Net) Research Project, Umeå University - Computing Science Department & Umeå University - Police Education Unit,, Comet Global Innovation S.L., Institut de Seguretat Pública de Catalunya -ISPC
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

A Robust Solution to Variational Importance Sampling of Minimum Variance

  • Hernández-González, Jerónimo
  • Cerquides, Jesús
© 2020 by the authors., Importance sampling is a Monte Carlo method where samples are obtained from an alternative proposal distribution. This can be used to focus the sampling process in the relevant parts of space, thus reducing the variance. Selecting the proposal that leads to the minimum variance can be formulated as an optimization problem and solved, for instance, by the use of a variational approach. Variational inference selects, from a given family, the distribution which minimizes the divergence to the distribution of interest. The Rényi projection of order 2 leads to the importance sampling estimator of minimum variance, but its computation is very costly. In this study with discrete distributions that factorize over probabilistic graphical models, we propose and evaluate an approximate projection method onto fully factored distributions. As a result of our evaluation it becomes apparent that a proposal distribution mixing the information projection with the approximate Rényi projection of order 2 could be interesting from a practical perspective., This research was partially funded by the CI-SUSTAIN project (PID2019-104156GB-I00) of the Spanish Ministry of Science and Innovation, the CROWD4SDG (872944) and the Humane-AI-Net (952026) projects, both of the EU Horizon 2020 programme. J.H.-G. carried out a large part of this work while at IIIA-CSIC with a Juan de la Cierva-Incorporación grant from the Spanish Ministry of Science and Innovation., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Qualitative case-based reasoning and learning

  • Homem, Thiago
  • Santos, Paulo E.
  • Reali Costa, A. H
  • Bianchi, Reinaldo Augusto da Costa
  • López de Mántaras, Ramón
The development of autonomous agents that perform tasks with the same dexterity as performed by humans is one of the challenges of artificial intelligence and robotics. This motivates the research on intelligent agents, since the agent must choose the best action in a dynamic environment in order to maximise the final score. In this context, the present paper introduces a novel algorithm for Qualitative Case-Based Reasoning and Learning (QCBRL), which is a case-based reasoning system that uses qualitative spatial representations to retrieve and reuse cases by means of relations between objects in the environment. Combined with reinforcement learning, QCBRL allows the agent to learn new qualitative cases at runtime, without assuming a pre-processing step. In order to avoid cases that do not lead to the maximum performance, QCBRL executes case-base maintenance, excluding these cases and obtaining new (more suitable) ones. Experimental evaluation of QCBRL was conducted in a simulated robot-soccer environment, in a real humanoid-robot environment and on simple tasks in two distinct gridworld domains. Results show that QCBRL outperforms traditional RL methods. As a result of running QCBRL in autonomous soccer matches, the robots performed a higher average number of goals than those obtained when using pure numerical models. In the gridworlds considered, the agent was able to learn optimal and safety policies., Paulo Eduardo Santos acknowledges support from FAPESP-IBM Proc. 2016/18792-9. Ramon Lopez de Mantaras has been partially supported by the Generalitat de Catalunya Grant number 2017 SGR 172, by the CSIC Grant PIE 201750E090, and by the European Project “HumanE-AI-Net” H2020-ICT-48-2020-RIA-952026. Anna Helena Reali Costa acknowledges support from CNPq Proc. 425860/2016-7 and 307027/2017-1. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brazil (CAPES) - Finance Code 001.
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

A Conceptual Probabilistic Framework for Annotation Aggregation of Citizen Science Data

  • Cerquides, Jesús
  • Mulayim, Mehmet Oguz
  • Hernández-González, Jerónimo
  • Ravi Shankar, Amudha
  • Fernández-Márquez, José Luis
Over the last decade, hundreds of thousands of volunteers have contributed to science by collecting or analyzing data. This public participation in science, also known as citizen science, has contributed to significant discoveries and led to publications in major scientific journals. However, little attention has been paid to data quality issues. In this work we argue that being able to determine the accuracy of data obtained by crowdsourcing is a fundamental question and we point out that, for many real-life scenarios, mathematical tools and processes for the evaluation of data quality are missing. We propose a probabilistic methodology for the evaluation of the accuracy of labeling data obtained by crowdsourcing in citizen science. The methodology builds on an abstract probabilistic graphical model formalism, which is shown to generalize some already existing label aggregation models. We show how to make practical use of the methodology through a comparison of data obtained from different citizen science communities analyzing the earthquake that took place in Albania in 2019., This work was partially supported by the projects Crowd4SDG and Humane-AI-net, which have received funding from the European Union’s Horizon 2020 research and innovation program under grant agreements No 872944 and No 952026, respectively. This work was also partially supported by the project CI-SUSTAIN funded by the Spanish Ministry of Science and Innovation (PID2019-104156GB-I00). We acknowledge the support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI). J.H.-G. is a Serra Húnter Fellow., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Otros (Other)

Analytical and numerical treatment of continuous ageing in the voter model

  • Baron, Joseph W.
  • Peralta, Antonio F.
  • Galla, Tobias
  • Toral, Raúl
The conventional voter model is modified so that an agent's switching rate depends on the `age' of the agent, that is, the time since the agent last switched opinion. In contrast to previous work, age is continuous in the present model. We show how the resulting individual-based system with non-Markovian dynamics and concentration-dependent rates can be handled both computationally and analytically. Lewis' thinning algorithm can be modified in order to provide an efficient simulation method. Analytically, we demonstrate how the asymptotic approach to an absorbing state (consensus) can be deduced. We discuss three special cases of the age dependent switching rate: one in which the concentration of voters can be approximated by a fractional differential equation, another for which the approach to consensus is exponential in time, and a third case in which the system reaches a frozen state instead of consensus. Finally, we include the effects of spontaneous change of opinion, i.e., we study a noisy voter model with continuous ageing. We demonstrate that this can give rise to a continuous transition between coexistence and consensus phases. We also show how the stationary probability distribution can be approximated, despite the fact that the system cannot be described by a conventional master equation., We acknowledge funding from the Spanish Ministry of Science, Innovation and Universities, the Agency AEI and FEDER (EU) under the grant PACSS (RTI2018-093732-B-C22), and the Maria de Maeztu program for Units of Excellence in R&D (MDM-2017-0711) funded by MCIN/AEI/10.13039/501100011033. A.F.P. acknowledges support from projects AFOSR (Grant No. FA8655-20-1-7020) and EU H2020 Humane AI-net (Grant No. 952026)., No
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Analytical and numerical treatment of continuous ageing in the voter model

  • Baron, Joseph W.
  • Peralta, Antonio F.
  • Galla, Tobias
  • Toral, Raúl
The conventional voter model is modified so that an agent’s switching rate depends on the ‘age’ of the agent—that is, the time since the agent last switched opinion. In contrast to previous work, age is continuous in the present model. We show how the resulting individual-based system with non-Markovian dynamics and concentration-dependent rates can be handled both computationally and analytically. The thinning algorithm of Lewis and Shedler can be modified in order to provide an efficient simulation method. Analytically, we demonstrate how the asymptotic approach to an absorbing state (consensus) can be deduced. We discuss three special cases of the age-dependent switching rate: one in which the concentration of voters can be approximated by a fractional differential equation, another for which the approach to consensus is exponential in time, and a third case in which the system reaches a frozen state instead of consensus. Finally, we include the effects of a spontaneous change of opinion, i.e., we study a noisy voter model with continuous ageing. We demonstrate that this can give rise to a continuous transition between coexistence and consensus phases. We also show how the stationary probability distribution can be approximated, despite the fact that the system cannot be described by a conventional master equation., The APC was funded by the Spanish Ministry of Science, Innovation and Universities, the Agency AEI and FEDER (EU) under the grant PACSS (RTI2018-093732-B-C22), and the Maria de Maeztu program for Units of Excellence in R&D (MDM-2017-0711) funded by MCIN/AEI/10.13039/501100011033. A.F.P. acknowledges support from projects AFOSR (Grant No. FA8655-20-1-7020) and EU H2020 Humane AI-Net (Grant No. 952026)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Enhancing Medical Image Segmentation: Ground Truth Optimization through Evaluating Uncertainty in Expert Annotations

  • Athanasiou, Georgios
  • Arcos Rosell, Josep Lluís
  • Cerquides, Jesús
The surge of supervised learning methods for segmentation lately has underscored the critical role of label quality in predicting performance. This issue is prevalent in the domain of medical imaging, where high annotation costs and inter-observer variability pose significant challenges. Acquiring labels commonly involves multiple experts providing their interpretations of the “true” segmentation labels, each influenced by their individual biases. The blind acceptance of these noisy labels as the ground truth restricts the potential effectiveness of segmentation algorithms. Here, we apply coupled convolutional neural network approaches to a small-sized real-world dataset of bovine cumulus oocyte complexes. This is the first time these methods have been applied to a real-world annotation medical dataset, since they were previously tested only on artificially generated labels of medical and non-medical datasets. This dataset is crucial for healthy embryo development. Its application revealed an important challenge: the inability to effectively learn distinct confusion matrices for each expert due to large areas of agreement. In response, we propose a novel method that focuses on areas of high uncertainty. This approach allows us to understand the individual characteristics better, extract their behavior, and use this insight to create a more sophisticated ground truth using maximum likelihood. These findings contribute to the ongoing discussion of leveraging machine learning algorithms for medical image segmentation, particularly in scenarios involving multiple human annotators., G.A. is supported by the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 860960 and by project CI-SUSTAIN (PID2019-104156GB-I00), funded by the Spanish Ministry of Science and Innovation. G.A. is a Ph.D. student at the doctoral program for computer science at the Universitat Autonoma de Barcelona. J.C. is supported by the project Humane-AI-net, which has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No. 952026., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Otros (Other)

Developing and Validating Tools for the Automated Analysis and Enhancement of Online Discussions

  • Komander, Bjoern
  • Cerquides, Jesús
  • Piera, Jaume
Second International Conference on Hybrid Human-Artificial Intelligence (HHAI 2023).-- 3 pages, Online discussions play a vital role in today’s society and recent advances in Natural Language Processing (NLP) demonstrate promising potential for analyzing and facilitating them. This paper proposes a comprehensive framework that involves developing NLP tools based on the latest advances in Large Language Models (LLMs), creating and automating measures to assess the quality of discussions, integrating these tools into online discussion platforms, and designing and implementing adequate experiments to evaluate their usefulness. We emphasize the importance of both user feedback and well-designed experiments to assess how the accelerated development of LLMs can contribute to society. By collaborating with various stakeholders in citizen science and deliberative democracy platforms, we seek to integrate our framework in a meaningful manner. Overall, our framework offers a roadmap for leveraging LLMs to enhance online discussions in a responsible and effective way, This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement Nº 101034328, HumanE-AI-Net project N. 952026, and from the Horizon Europe programme project GUARDEN 101060693, With the institutional support of the ‘Severo Ochoa Centre of Excellence’ accreditation (CEX2019-000928-S), Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

The role of complexity for digital twins of cities

  • Caldarelli, G.
  • Arcaute, Elsa
  • Barthelemy, Marc
  • Batty, M.
  • Gershenson, C.
  • Helbing, D.
  • Mancuso, S.
  • Moreno, Yamir
  • Ramasco, José J.
  • Rozenblat, C.
  • Sánchez, Ángel
  • Fernández-Villacañas, J. L.
We argue that theories and methods drawn from complexity science are urgently needed to guide the development and use of digital twins for cities. The theoretical framework from complexity science takes into account both the short-term and the long-term dynamics of cities and their interactions. This is the foundation for a new approach that treats cities not as large machines or logistic systems but as mutually interwoven self-organizing phenomena, which evolve, to an extent, like living systems., G.C. acknowledges support from EU Project ‘HumanE-AI-Net’, no. 952026 and from project SERICS (PE00000014) under the NRRP MUR program funded by the EU - NGEU. C.G. acknowledges support from UNAM-PAPIIT projects (IN107919, IV100120, IN105122) and from the PASPA program from UNAM-DGAPA. D.H. acknowledges support through the project ‘CoCi: Co-Evolving City Life’, which has received funding from the European Research Council (ERC) under the European Union Horizon 2020 research and innovation program under grant agreement no. 833168. Y.M. acknowledges support from the Government of Aragon through grant E36-20R (FENOL), and from MCIN/AEI/10.13039/501100011033 through grant PID2020-115800GB-I00. E.A. and M. Batty acknowledge support from the Alan Turing Institute under QUANT2-Contract-CID-3815811 and from the UK Regions Digital Research Facility (UKRDRF) EP/M023583/1 through EPSRC. J.J.R. acknowledges funding from MCIN/AEI/10.13039/501100011033 and Fondo Europeo de Desarrollo Regional (FEDER, UE) under Project APASOS (PID2021-122256NB-C22), and the María de Maeztu Program for units of Excellence in R&D CEX2021-001164-M by MCIN/AEI/10.13039/501100011033. A.S. acknowledges support from project BASIC (PGC2018-098186-B-I00) funded by MCIN/AEI/10.13039/501100011033 and by ‘ERDF A way of making Europe’., With funding from the Spanish government through the "Severo Ochoa Centre of Excellence" accreditation (CEX2021-001164-M)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Otros (Other)

Expert Finding for Cititzen Science

  • Komander, Bjoern
  • Cerquides, Jesús
  • Piera, Jaume
  • Chan, Jeffrey
  • Alavi, Azadeh
25th International Conference of the Catalan Association for Artificial Intelligence, 25-27 October 2023, Barcelona.-- 11 pages, 2 figures, 2 tables, Citizen Science brings together scientists and public participants to collaborate on a wide range of applications and fields. With this approach, Citizen Science advances scientific research and communication while accounting for various stakeholders. Across many Citizen Science projects, digital discussion platforms play an essential role for self-governance and self-organisation. In order to increase the quality of the discussions held on these platforms, we propose a model that recommends users to new discussions in which they are likely to contribute meaningful content. Our model learns relevant user representations based on the quality of past interactions between users and discussion threads, as well as the text content of questions, using a ranking loss function, an approximation of the NDCG metric, and matrix factorization. We demonstrate that our approach is able to predict potential experts on unseen discussion threads and outperforms several baselines. Compared to state-of-the-art expert finding techniques, the architecture of our model is significantly less complex, while focusing on a mostly overlooked ranking loss function, This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement Nº 101034328, HumanE-AI-Net project N. 952026, and from the Horizon Europe programme project GUARDEN 101060693. [...] With the institutional support of the ‘Severo Ochoa Centre of Excellence’ accreditation (CEX2019-000928-S), Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Crowdsourced geolocation: detailed exploration of mathematical and computational modeling approaches

  • Ballester, Rocco
  • Labeyrie, Yanis
  • Mulayim, Mehmet Oguz
  • Fernández-Márquez, José Luis
  • Cerquides, Jesús
In emergency situations, social media platforms produce a vast amount of real-time data that holds immense value, particularly in the first 72 h following a disaster event. Despite previous efforts, efficiently determining the geographical location of images related to a new disaster remains an unresolved operational challenge. Currently, the state-of-the-art approach for dealing with these first response mapping is first filtering and then submitting the images to be geolocated to a volunteer crowd, assigning the images randomly to the volunteers. In this work, we extend our previous paper (Ballester et al., 2023) to explore the potential of artificial intelligence (AI) in aiding emergency responders and disaster relief organizations in geolocating social media images from a zone recently hit by a disaster. Our contributions include building two different models in which we try to (i) be able to learn volunteers’ error profiles and (ii) intelligently assign tasks to those volunteers who exhibit higher proficiency. Moreover, we present methods that outperform random allocation of tasks, analyze the effect on the models’ performance when varying numerous parameters, and show that for a given set of tasks and volunteers, we are able to process them with a significantly lower annotation budget, that is, we are able to make fewer volunteer solicitations without losing any quality on the final consensus., This work has received funding from the European Union’s Horizon 2020 research and innovation program HumanE-AI-Net, Crowd4SDG, and K-HealthInAIR projects (N. 952026, N. 872944, and N. 101057 693, respectively) and from the grant PID2022-136787NB-I00 funded by MCIN/AEI/10.13039/501100011033., No
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Otros (Other)

Implementing 3-SAT Gadgets for Quantum Annealers with Random Instances

  • Rodríguez-Farrés, Pol
  • Ballester, Rocco
  • Ansótegui, Carlos
  • Levy, Jordi
  • Cerquides, Jesús
The 7-volume set LNCS 14832 – 14838 constitutes the proceedings of the 24th International Conference on Computational Science, ICCS 2024, which took place in Malaga, Spain, during July 2–4, 2024.- Proceedings, Part VI., The Maximum Boolean Satisfiability Problem (also known as the Max-SAT problem) is the problem of determining the maximum number of disjunctive clauses that can be satisfied (i.e., made true) by an assignment of truth values to the formula’s variables. This is a generalization of the well-known Boolean Satisfiability Problem (also known as the SAT problem), the first problem that was proven to be NP-complete. With the proliferation of quantum computing, a current approach to tackle this optimization problem is Quantum Annealing (QA). In this work, we compare several gadgets that translate 3-SAT problems into Quadratic Unconstrained Binary Optimization (QUBO) problems to be able to solve them in a quantum annealer. We show the performance superiority of the not-yet-considered gadgets in comparison to state-of-the-art approaches when solving random instances in D-Wave’s quantum annealer., This work was founded partially by the Government of Catalonia Exp. Num. 2023 DI 00071, by Enzyme Advising Group under the development of an industrial PhD on Quantum Machine Learning, by CSIC’s JAE Intro JAEICU_23_00782 and by grant PID2022-136787NB-I00 funded by MCIN/AEI/10.13039/501100011033. Jesus Cerquides is funded by European Union KHealthInAir, GUARDEN, and Humane-AI projects with No. 101057693, 101060693, and 952026., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Nudging punishment against sharing of fake news

  • Meiske, Biljana
  • Álvarez Benjumea, Amalia
  • Andrighetto, Giulia
  • Polizzi, Eugenia
Corrective comments posted in response to misinformation shared on social media are not only effective in reducing belief in misinformation among those observing the interaction but also serve as a publicly observable social punishment against fake news sharing. We suggest that, by visibly displaying punishing behavior, corrective comments have the potential to function as a norm nudge, updating observers’ perception of the norms regulating punishment, the so-called “meta-norms”. We conducted a preregistered online experiment in which participants joined discussions resembling an online forum and could comment on posts shared by previous users. We manipulated participants’ possibility to observe corrective comments replying to a post sharing fake news and measured the effect of this variation on their propensity to leave a corrective comment. We show that participants exposed to posts corrected by other users are significantly more likely to reply with a corrective comment, even after controlling for participants’ perceived accuracy of the shared post. Participants exposed to the corrections provided by previous users perceive replying with corrections as more socially appropriate. Our results suggest that social corrections work as a (meta) norm nudge, increasing the probability of punishing norm violations by increasing the social appropriateness of correcting. Our findings suggest that interventions targeting “would-be” enforcers could complement existing policies specifically directed at norm violators., This work was funded by the European Commission funded project “Humane AI Network” (Grant 952026) and by the Research Grant Open call 2023 (Vetenskapsrådet) ‘Social Norms for Cooperation Under Collective Risk’, Sweden (grant no. 2023-01306), Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Jekyll institute or Mrs Hyde? gender identification with machine learning

  • Gombert, Arnault
  • Sánchez-López, Borja
  • Cerquides, Jesús
Social media platforms offer an invaluable wealth of data to understand what is taking place in our society. However, social media data hides demographic biases related to characteristics such as gender or age. Therefore, considering social media data as representative of the population can lead to fallacious interpretations. For instance, in France in 2021, women represent 51.6% of the population1, whereas on Twitter they represent only 33.5% of French users 2. With such a significant difference between social network user demographics and the actual population, detecting the gender or age before delving into a deeper analysis of social phenomena becomes a priority. In this paper, we tackle the gender detection problem on Twitter. We introduce miniAM2, which is an assemblage model of an enriched distillation with weak-supervised learning. Our contributions are threefold: (i) a novel multilingual model that outperforms existing models in both accuracy and speed, allowing for real-time gender detection and organization status on Twitter based on their name, screen_name, and description, making it lighter and faster than state-of-the-art; (ii) an innovative assemblage multi-language strategy that enriches a distillation process with weak-supervised learning using minimal annotated data, and (iii) a unique method to adapt the model to similar languages without requiring annotated data in the target language, which provides significant advancements in handling resource-poor languages in gender detection tasks. We provide our model on demand so social scientists can use it for their analysis., This work is partially supported by the projects Crowd4SDG, GUARDEN, and Humane-AI-net, which have received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreements No. 872944, No. 101060693 and No. 952026, respectively., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Antibiotic-loaded bone cement is associated with a reduction of the risk of revision of total knee arthroplasty: Analysis of the Catalan Arthroplasty Register

  • Gil-González, Sergi
  • Velasco-Regúlez, Borja
  • Cerquides, Jesús
  • Hinarejos, Pedro
  • Monllau, Joan Carles
  • Pelfort, Xavier
Purpose The purpose of this study was to analyse the impact on peri-prosthetic joint infection (PJI) rate and prosthetic survival using antibiotic-loaded bone cement (ALBC) versus plain cement during total knee arthroplasty (TKA)., Methods A retrospective cohort study was conducted. The main data source was the Catalan Arthroplasty Register (RACat). TKAs with surgery date between 1 January 2011 and 31 December 2020 were analysed and followed up until 31 December 2023. The main variable of interest was the type of cement (ALBC vs. plain cement), and several endpoints (septic revision, aseptic revision, and all-cause revision) were considered. The analysed outcomes were revision rates, survival rates and risk factors' hazard ratios (HR)., Results A total of 22,781 TKAs were analysed, 13,125 (57.6%) of them with plain cement and 9656 (42.4%) with ALBC. The septic revision rate was lower in the ALBC group after 3 months of follow-up (0.52% vs. 0.78%, p value = 0.04). Prosthetic survival with respect to the aseptic revision endpoint was also higher for the ALBC group during the whole follow-up period (~158 months). Regarding risk factors for infection, ALBC showed a protective effect, HR: 0.53 (0.44, 0.63), while sex (being male) and the analysed comorbidities increased the risk., Conclusions ALBC is associated with a reduction in both the septic revision and the aseptic revision rate after TKA, and thus with higher prosthetic survival., Level of Evidence Level III, Therapeutic, retrospective., This work was supported by the Ministerio de Ciencia e Innovación (projects PID2022-136787NB-I00 and PID2019-104156GB-I00), the European Commission H2020 program (grants 101060693, 101057693, 952026) and the Agència de Gestió d'Ajuts Universitaris i de Recerca (Award Number: 2020-18). None of the funding sources had a direct role in this work., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Quantum federated learning: a comprehensive literature review of foundations, challenges, and future directions

  • Ballester, Rocco
  • Cerquides, Jesús
  • Artiles, Luis
Federated learning (FL) is a recent technique that emerged to handle the vast amount of training data needed in machine learning algorithms while fulfilling data owners’ privacy challenges in such scenarios. Simultaneously, the field of quantum computing (QC), using quantum properties such as entanglement and superposition to perform computation, has experienced exponential growth, theoretically proving to be more efficient in specific machine learning tasks and creating the discipline known as quantum machine learning (QML). Thus, an emerging body of knowledge has started studying the combination of these two research agendas, giving rise to the field of quantum federated learning (QFL). In this review, we systematically classify the existing literature through a novel taxonomy, identify current trends and challenges, and highlight research gaps and future directions to support the continued development of this emerging field., This work was founded partially by the Government of Catalonia Exp. Num. 2023 DI 00071 and by Strategic Platform under the development of an industrial PhD in Quantum Machine Learning. Jesus Cerquides is funded by European Union KHealthInAir, GUARDEN, and Humane-AI projects with No. 101057693, 101060693, and 952026. This work was supported by grant PID2022-136787NB-I00 funded by MCIN/AEI/10.13039/501100011033., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

The first causal inference analysis of the Catalan Arthroplasty Register shows a positive effect of antibiotic-loaded bone cement on knee prosthesis survival

  • Velasco-Regulez, Borja
  • Gil-Gonzalez, Sergi
  • Cerquides, Jesús
[Data availability statemente] The data employed for this study were individual patient pseudo-anonymized data and were obtained under the Data Analysis Programme for Research and Innovation in Healthcare (PADRIS), managed by AQuAS, the Agency for Health Quality and Assessment of Catalonia. This data cannot be made available publicly., [Purpose] The survival of a knee prosthesis is one of the most important indicators of the success or failure of a knee arthroplasty. An intervention that could increase prosthetic survival is the use of antibiotic-loaded bone cement (ALBC) during primary surgery, but the evidence for this is not conclusive. The question of whether such an intervention increases prosthetic survival is a causal one, and yet it has never been addressed with causal methods in the observational studies literature. This constitutes a serious limitation, as there is growing evidence that the best-suited framework for addressing causal questions with observational data is causal inference., [Methods] In the present study, causal inference methods were employed to answer the research question of whether ALBC increases prosthetic survival. In particular, directed acyclic graphs were used for identification and causal survival forests were used to estimate the effect of interest. The rationale behind these methods is provided in the main text, and technical details are provided in Supporting Information: File S3. Data from the Catalan Arthroplasty Register were analysed., [Results] ALBC had an effect of increasing the overall prosthetic survival by 8% after 120 months of follow-up. The intervention had a positive effect across all the subgroups of the population defined by confounding variables, but the effect was greater in men, young patients, patients with rheumatoid arthritis or obesity, and patients who smoked or abused alcohol. The chosen causal assumptions had an impact on the obtained results, empirically showing the importance of using a causal framework., [Conclusions] ALBC increased knee prosthesis survival among patients in the Catalan public healthcare system. Causal inference methods are the most appropriate for answering causal questions about the effect of ALBC on prosthetic survival when the analysed data are observational., [Level of Evidence] Level III., The authors thank our contributors, Olga Martínez Cruz (Catalan Arthroplasty Register [RACat], Agència de Qualitat i Avaluació Sanitàries de Catalunya [AQuAS], Barcelona, Spain), Francesc Pallisó Folch (Advisory Committee, Catalan Arthroplasty Register [RACat]; Hospital Universitari Santa Maria, Gestió de Serveis Sanitaris, Lleida, Spain) and Mireia Espallargues Carreras (Catalan Arthroplasty Register [RACat], Agència de Qualitat i Avaluació Sanitàries de Catalunya [AQuAS]; Network for Research on Chronicity, Primary Care and Health Promotion [RICAPPS], Spain) for their role in the data extraction of the RACat and their valuable comments for the manuscript. This work was supported by the Ministerio de Ciencia e Innovación (projects PID2022-136787NB-I00 and PID2019-104156GB-I00), the European Commission H2020 programme (grants 101060693, 101057693 and 952026) and the Agència de Gestió d'Ajuts Universitaris i de Recerca (Award Number: 2020-18). None of the funding sources had a direct role in this work., Peer reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Intentional policy graphs: A pipeline for explaining agent behavior through intentions

  • Giménez Ábalos, Víctor
  • Álvarez Napagao, Sergio|||0000-0001-9946-9703
  • Tormos Llorente, Adrián
  • Montese, Sara
  • Cortés García, Claudio Ulises|||0000-0003-0192-3096
  • Vázquez Salceda, Javier|||0000-0003-1732-9446
Agents increasingly operate in complex environments, where coherent behavior often emerges from opaque decision-making processes. While such systems can be highly effective, this lack of transparency limits trust, auditing, and meaningful human understanding. We introduce intentional policy graphs, a post hoc, model- agnostic framework that explains agent behavior in terms of intentions: probabilistic commitments to desired outcomes inferred from partial observations. By extending policy graphs with a formal notion of intention, we move beyond action-level descriptions toward telic explanations of why agents pursue particular trajectories. The framework provides a complete construction pipeline, design principles, and quantitative metrics that explicitly characterize the trade-off between interpretability and reliability. Intentions support structured answers to what, how, and why questions, enabling both local and global explanations of behavior. We demonstrate the approach in a cooperative multi-agent game and on real-world human driving data, highlighting its generality and explanatory power without access to internal reasoning models., This work has been partially supported by the AIXPERT (grant agreement ID: 101214389), AI4CCAM (grant agreement ID: 101076911), and HumanE-AI- Net (grant agreement ID: 952026) European projects and fellowships to V.G.-A., A.T., and S.M. within the ‘‘Generacio´n D’’ initiative, Red.es, MTDFP, for talent attraction (C005/24-ED CV1). This work was partially funded by the European Union NextGenerationEU funds, through PRTR., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Ladder of intentions: unifying agent architectures for explainability and transferability

  • Giménez Ábalos, Víctor
  • Tormos Llorente, Adrián
  • Edström, Filip
  • Álvarez Napagao, Sergio|||0000-0001-9946-9703
  • Vázquez Salceda, Javier|||0000-0003-1732-9446
  • Brännström, Mattias
  • Lindqvist, John
Within the field of Autonomous Agents, the predominant paradigm is that agents perceive, reflect, reason, and act on an environment, employing some specific decision mechanism to pick actions. Nonetheless, the process that originates the decisions may differ depending on the agent, as this paradigm is agnostic about its concrete action selection inference. However, the need for being able to explain these decisions is constantly increasing, and the heterogeneity of the internal processes of agents has resulted in different ad hoc techniques for each architecture, for providing explanations with disparate validation mechanisms, hindering efforts at comparing mechanisms. To tackle this, in this contribution, we propose a unifying architecture framework based on causality, beliefs, and intentions. This framework allows for the examination of heterogeneous agents (from BDI and RL to LLM-based agents) without modification. This approach clearly decouples declarative and procedural knowledge, as well as designer-given versus learnt representations. It categorises what kind of questions can be answered by each agent reasoning component and allows a more seamless workflow for transferring knowledge between diverse agent architectures., This work has been partially supported by the HUMANE (Grant agreement ID: 952026) and V. Gimenez-Abalos’ fellowship within the “Generación D” initiative, Red.es, MTDFP, for talent atraction (C005/24-ED CV1). Funded by the European Union NextGenerationEU funds, through PRTR., Peer Reviewed
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Human-AI coevolution

  • Pedreschi, Dino
  • Pappalardo, Luca
  • Ferragina, Emanuele
  • Baeza Yates, Ricardo
  • Barabási, Albert-László
  • Dignum, Frank
  • Dignum, Virginia
  • Eliassi-Rad, Tina
  • Giannotti, Fosca
  • Kertész, János
  • Knott, Alistair
  • Ioannidis, Yannis
  • Lukowicz, Paul
  • Passarella, Andrea
  • Pentland, Alex
  • Shawe-Taylor, John
  • Vespignani, Alessandro
Human-AI coevolution, defined as a process in which humans and AI algorithms continuously influence each other, increasingly characterises our society, but is understudied in artificial intelligence and complexity science literature. Recommender systems and assistants play a prominent role in human-AI coevolution, as they permeate many facets of daily life and influence human choices through online platforms. The interaction between users and AI results in a potentially endless feedback loop, wherein users' choices generate data to train AI models, which, in turn, shape subsequent user preferences. This human-AI feedback loop has peculiar characteristics compared to traditional human-machine interaction and gives rise to complex and often “unintended” systemic outcomes. This paper introduces human-AI coevolution as the cornerstone for a new field of study at the intersection between AI and complexity science focused on the theoretical, empirical, and mathematical investigation of the human-AI feedback loop. In doing so, we: (i) outline the pros and cons of existing methodologies and highlight shortcomings and potential ways for capturing feedback loop mechanisms; (ii) propose a reflection at the intersection between complexity science, AI and society; (iii) provide real-world examples for different human-AI ecosystems; and (iv) illustrate challenges to the creation of such a field of study, conceptualising them at increasing levels of abstraction, i.e., scientific, legal and socio-political., This work has been partially supported by PNRR - M4C2 - Investimento 1.3, Partenariato Esteso (grant No. PE00000013) - “FAIR - Future Artificial Intelligence Research” - Spoke 1 “Human-centered AI”, funded by the European Commission under the Next Generation EU programme; EU H2020 projects HumaneAI-net G.A. 952026 and SoBigData++ G.A. 871042; European Research Council ERC-2018-ADG 834756 “XAI: Science and technology for the eXplanation of AI decision making”; CHIST-ERA grant CHIST-ERA-19-XAI-010, by MUR (grant No. not available), FWF (grant No. I 5205), EPSRC (grant No. EP/V055712/1), NCN (grant No. 2020/02/Y/ST6/00064), ETAg (grant No. SLTAT21096), BNSF (grant No. KΠ-06-AOO2/5); PNRR (Piano Nazionale di Ripresa e Resilienza) in the context of the research program 20224CZ5X4_PE6_PRIN 2022 “URBAI – Urban Artificial Intelligence” (CUP B53D23012770006), funded by European Union – Next Generation EU; Emanuele Ferragina has been partially supported by a public grant overseen by the French National Research Agency (ANR) as part of the ‘Investissements d'Avenir’ program LIEPP (ANR-11-LABX-0091, ANR-11-IDEX-0005-02) and the Université de Paris IdEx (ANR-18-IDEX-0001). We thank Vincenzo Vivarini for the inspiration about the importance of the feedback loop in complex social systems. There is always a human agent beyond any tactical system. We also thank Daniele Fadda for the support on making the figures. Finally, we thank all members of KDD-Lab for insightful discussions about human-AI coevolution.
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Generative AI models should include detection mechanisms as a condition for public release

  • Knott, Alistair
  • Pedreschi, Dino
  • Chatila, Raja
  • Chakraborti, Tapabrata
  • Leavy, Susan
  • Baeza Yates, Ricardo
  • Eyers, David
  • Trotman, Andrew
  • Teal, Paul D.
  • Biecek, Przemyslaw
  • Russell, Stuart
  • Bengio, Yoshua
The new wave of ‘foundation models’—general-purpose generative AI models, for production of text (e.g., ChatGPT) or images (e.g., MidJourney)—represent a dramatic advance in the state of the art for AI. But their use also introduces a range of new risks, which has prompted an ongoing conversation about possible regulatory mechanisms. Here we propose a specific principle that should be incorporated into legislation: that any organization developing a foundation model intended for public use must demonstrate a reliable detection mechanism for the content it generates, as a condition of its public release. The detection mechanism should be made publicly available in a tool that allows users to query, for an arbitrary item of content, whether the item was generated (wholly or partly) by the model. In this paper, we argue that this requirement is technically feasible and would play an important role in reducing certain risks from new AI models in many domains. We also outline a number of options for the tool’s design, and summarize a number of points where further input from policymakers and researchers would be required., We are grateful to the Global Partnership on AI (GPAI) and the Montreal International Center of Expertise in Artificial Intelligence (CEIMIA) for supporting this project. Yoshua Bengio thanks CIFAR and NSERC; Stuart Russell thanks Open Philanthropy Foundation for support of the Center for Human-Compatible AI at UC Berkeley. Dino Pedreschi has been supported by the European Commission under the NextGeneration Programme PE0013 - ‘FAIR - Future Artificial Intelligence Research’ Spoke 1, ‘Human-Centered AI’, and under the EU H2020 ICT-48 Network of Excellence n.952026 ‘Human-AI net’. We would like to thank colleagues at the EU (Lucilla Sioli’s group at DG-CNECT) and the OECD (Sebastian Hallensleben’s group) for useful discussions, and Marcin Betkier and Rebecca Downes for comments on an earlier draft of this article. We also thank the two anonymous reviewers for their useful comments. The views expressed here, and any remaining errors, are of course our own.
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Intentional policy graphs: A pipeline for explaining agent behavior through intentions

  • Giménez Ábalos, Víctor
  • Álvarez Napagao, Sergio
  • Tormos Llorente, Adrián
  • Montese, Sara
  • Cortés García, Claudio Ulises
  • Vázquez Salceda, Javier
Agents increasingly operate in complex environments, where coherent behavior often emerges from opaque decision-making processes. While such systems can be highly effective, this lack of transparency limits trust, auditing, and meaningful human understanding. We introduce intentional policy graphs, a post hoc, model- agnostic framework that explains agent behavior in terms of intentions: probabilistic commitments to desired outcomes inferred from partial observations. By extending policy graphs with a formal notion of intention, we move beyond action-level descriptions toward telic explanations of why agents pursue particular trajectories. The framework provides a complete construction pipeline, design principles, and quantitative metrics that explicitly characterize the trade-off between interpretability and reliability. Intentions support structured answers to what, how, and why questions, enabling both local and global explanations of behavior. We demonstrate the approach in a cooperative multi-agent game and on real-world human driving data, highlighting its generality and explanatory power without access to internal reasoning models., This work has been partially supported by the AIXPERT (grant agreement ID: 101214389), AI4CCAM (grant agreement ID: 101076911), and HumanE-AI- Net (grant agreement ID: 952026) European projects and fellowships to V.G.-A., A.T., and S.M. within the ‘‘Generacio´n D’’ initiative, Red.es, MTDFP, for talent attraction (C005/24-ED CV1). This work was partially funded by the European Union NextGenerationEU funds, through PRTR., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

A Conceptual Probabilistic Framework for Annotation Aggregation of Citizen Science Data

  • Cerquides Bueno, Jesús
  • Mülâyim, Mehmet Oguz
  • Hernández-González, Jerónimo
  • Shankar, Amudha Ravi
  • Fernández Márquez, Jose Luis
Over the last decade, hundreds of thousands of volunteers have contributed to science by collecting or analyzing data. This public participation in science, also known as citizen science, has contributed to significant discoveries and led to publications in major scientific journals. However, little attention has been paid to data quality issues. In this work we argue that being able to determine the accuracy of data obtained by crowdsourcing is a fundamental question and we point out that, for many real-life scenarios, mathematical tools and processes for the evaluation of data quality are missing. We propose a probabilistic methodology for the evaluation of the accuracy of labeling data obtained by crowdsourcing in citizen science. The methodology builds on an abstract probabilistic graphical model formalism, which is shown to generalize some already existing label aggregation models. We show how to make practical use of the methodology through a comparison of data obtained from different citizen science communities analyzing the earthquake that took place in Albania in 2019.
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Ladder of intentions: unifying agent architectures for explainability and transferability

  • Giménez Ábalos, Víctor
  • Tormos Llorente, Adrián
  • Edström, Filip
  • Álvarez Napagao, Sergio
  • Vázquez Salceda, Javier
  • Brännström, Mattias
  • Lindqvist, John
Within the field of Autonomous Agents, the predominant paradigm is that agents perceive, reflect, reason, and act on an environment, employing some specific decision mechanism to pick actions. Nonetheless, the process that originates the decisions may differ depending on the agent, as this paradigm is agnostic about its concrete action selection inference. However, the need for being able to explain these decisions is constantly increasing, and the heterogeneity of the internal processes of agents has resulted in different ad hoc techniques for each architecture, for providing explanations with disparate validation mechanisms, hindering efforts at comparing mechanisms. To tackle this, in this contribution, we propose a unifying architecture framework based on causality, beliefs, and intentions. This framework allows for the examination of heterogeneous agents (from BDI and RL to LLM-based agents) without modification. This approach clearly decouples declarative and procedural knowledge, as well as designer-given versus learnt representations. It categorises what kind of questions can be answered by each agent reasoning component and allows a more seamless workflow for transferring knowledge between diverse agent architectures., This work has been partially supported by the HUMANE (Grant agreement ID: 952026) and V. Gimenez-Abalos’ fellowship within the “Generación D” initiative, Red.es, MTDFP, for talent atraction (C005/24-ED CV1). Funded by the European Union NextGenerationEU funds, through PRTR., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Antibiotic-loaded bone cement is associated with a reduction of the risk of revision of total knee arthroplasty: Analysis of the Catalan Arthroplasty Register

  • Gil-Gonzalez, Sergi
  • Velasco-Regúlez, Borja
  • Cerquides, Jesus
  • Hinarejos Gómez, Pedro Angel
  • Monllau García, Juan Carlos
  • Pelfort López, Javier
Purpose: The purpose of this study was to analyse the impact on peri-prosthetic joint infection (PJI) rate and prosthetic survival using antibiotic-loaded bone cement (ALBC) versus plain cement during total knee arthroplasty (TKA). Methods: A retrospective cohort study was conducted. The main data source was the Catalan Arthroplasty Register (RACat). TKAs with surgery date between 1 January 2011 and 31 December 2020 were analysed and followed up until 31 December 2023. The main variable of interest was the type of cement (ALBC vs. plain cement), and several endpoints (septic revision, aseptic revision, and all-cause revision) were considered. The analysed outcomes were revision rates, survival rates and risk factors' hazard ratios (HR). Results: A total of 22,781 TKAs were analysed, 13,125 (57.6%) of them with plain cement and 9656 (42.4%) with ALBC. The septic revision rate was lower in the ALBC group after 3 months of follow-up (0.52% vs. 0.78%, p value = 0.04). Prosthetic survival with respect to the aseptic revision endpoint was also higher for the ALBC group during the whole follow-up period (~158 months). Regarding risk factors for infection, ALBC showed a protective effect, HR: 0.53 (0.44, 0.63), while sex (being male) and the analysed comorbidities increased the risk. Conclusions: ALBC is associated with a reduction in both the septic revision and the aseptic revision rate after TKA, and thus with higher prosthetic survival. Level of evidence: Level III, Therapeutic, retrospective., This work was supported by the Ministerio de Ciencia e Innovación (projects PID2022-136787NB-I00 and PID2019-104156GB-I00), the European Commission H2020 program (grants 101060693, 101057693, 952026) and the Agència de Gestió d'Ajuts Universitaris i de Recerca (Award Number: 2020-18). None of the funding sources had a direct role in this work.
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

A Robust Solution to Variational Importance Sampling of Minimum Variance

  • Hernández-González, Jerónimo
  • Cerquides Bueno, Jesús
Importance sampling is a Monte Carlo method where samples are obtained from an alternative proposal distribution. This can be used to focus the sampling process in the relevant parts of space, thus reducing the variance. Selecting the proposal that leads to the minimum variance can be formulated as an optimization problem and solved, for instance, by the use of a variational approach. Variational inference selects, from a given family, the distribution which minimizes the divergence to the distribution of interest. The Rényi projection of order 2 leads to the importance sampling estimator of minimum variance, but its computation is very costly. In this study with discrete distributions that factorize over probabilistic graphical models, we propose and evaluate an approximate projection method onto fully factored distributions. As a result of our evaluation it becomes apparent that a proposal distribution mixing the information projection with the approximate Rényi projection of order 2 could be interesting from a practical perspective.