arXiv subjects

Sofia Yfantidou

Publications and source records attributed to Sofia Yfantidou.

13 recordsLinked to original sources

arXiv2026

(Un)fair devices: Moving beyond AI accuracy in personal sensing

Personal devices are omnipresent in our lives, seamlessly monitoring our activities, from smart rings tracking sleep patterns to smartwatches keeping an eye on missed heartbeats. The rich data streams from such devices fuel advanced Artificial Intelligence (AI) applications. Instead of solely relying on direct sensor measurements, these applications are increasingly leveraging Machine Learning (ML) model estimates to derive insights. But are these estimates biased or not? This literature review delivers compelling evidence about the impact of hidden biases that creep into ML models deployed on personal devices. We discuss critical bias issues drawn from prior work such as racial bias in pulse oximeters, weight bias in optical heart rate sensors, and sex bias in audio-based diagnostics. In response to these challenges, we advocate for a shift from prioritizing performance-oriented evaluations of personal devices to adopting assessments grounded in a human-centered approach. To facilitate this transition, we provide guidelines for the design, development, evaluation, and use of unbiased AI in personal devices, recognizing their potential impact on improving our health, lifestyle, and productivity -- more than any other technology.

cs.CY↗

arXiv2025

Mind the XAI Gap: A Human-Centered LLM Framework for Democratizing Explainable AI

Artificial Intelligence (AI) is rapidly embedded in critical decision-making systems, however their foundational ``black-box'' models require eXplainable AI (XAI) solutions to enhance transparency, which are mostly oriented to experts, making no sense to non-experts. Alarming evidence about AI's unprecedented human values risks brings forward the imperative need for transparent human-centered XAI solutions. In this work, we introduce a domain-, model-, explanation-agnostic, generalizable and reproducible framework that ensures both transparency and human-centered explanations tailored to the needs of both experts and non-experts. The framework leverages Large Language Models (LLMs) and employs in-context learning to convey domain- and explainability-relevant contextual knowledge into LLMs. Through its structured prompt and system setting, our framework encapsulates in one response explanations understandable by non-experts and technical information to experts, all grounded in domain and explainability principles. To demonstrate the effectiveness of our framework, we establish a ground-truth contextual ``thesaurus'' through a rigorous benchmarking with over 40 data, model, and XAI combinations for an explainable clustering analysis of a well-being scenario. Through a comprehensive quality and human-friendliness evaluation of our framework's explanations, we prove high content quality through strong correlations with ground-truth explanations (Spearman rank correlation=0.92) and improved interpretability and human-friendliness to non-experts through a user study (N=56). Our overall evaluation confirms trust in LLMs as HCXAI enablers, as our framework bridges the above Gaps by delivering (i) high-quality technical explanations aligned with foundational XAI methods and (ii) clear, efficient, and interpretable human-centered explanations for non-experts.

cs.LG↗

arXiv2025

Ties of Trust: a bowtie model to uncover trustor-trustee relationships in LLMs

The rapid and unprecedented dominance of Artificial Intelligence (AI), particularly through Large Language Models (LLMs), has raised critical trust challenges in high-stakes domains like politics. Biased LLMs' decisions and misinformation undermine democratic processes, and existing trust models fail to address the intricacies of trust in LLMs. Currently, oversimplified, one-directional approaches have largely overlooked the many relationships between trustor (user) contextual factors (e.g. ideology, perceptions) and trustee (LLMs) systemic elements (e.g. scientists, tool's features). In this work, we introduce a bowtie model for holistically conceptualizing and formulating trust in LLMs, with a core component comprehensively exploring trust by tying its two sides, namely the trustor and the trustee, as well as their intricate relationships. We uncover these relationships within the proposed bowtie model and beyond to its sociotechnical ecosystem, through a mixed-methods explanatory study, that exploits a political discourse analysis tool (integrating ChatGPT), by exploring and responding to the next critical questions: 1) How do trustor's contextual factors influence trust-related actions? 2) How do these factors influence and interact with trustee systemic elements? 3) How does trust itself vary across trustee systemic elements? Our bowtie-based explanatory analysis reveals that past experiences and familiarity significantly shape trustor's trust-related actions; not all trustor contextual factors equally influence trustee systemic elements; and trustee's human-in-the-loop features enhance trust, while lack of transparency decreases it. Finally, this solid evidence is exploited to deliver recommendations, insights and pathways towards building robust trusting ecosystems in LLM-based solutions.

cs.CY↗

arXiv2025

AgoraSpeech: A multi-annotated comprehensive dataset of political discourse through the lens of humans and AI

Political discourse datasets are important for gaining political insights, analyzing communication strategies or social science phenomena. Although numerous political discourse corpora exist, comprehensive, high-quality, annotated datasets are scarce. This is largely due to the substantial manual effort, multidisciplinarity, and expertise required for the nuanced annotation of rhetorical strategies and ideological contexts. In this paper, we present AgoraSpeech, a meticulously curated, high-quality dataset of 171 political speeches from six parties during the Greek national elections in 2023. The dataset includes annotations (per paragraph) for six natural language processing (NLP) tasks: text classification, topic identification, sentiment analysis, named entity recognition, polarization and populism detection. A two-step annotation was employed, starting with ChatGPT-generated annotations and followed by exhaustive human-in-the-loop validation. The dataset was initially used in a case study to provide insights during the pre-election period. However, it has general applicability by serving as a rich source of information for political and social scientists, journalists, or data scientists, while it can be used for benchmarking and fine-tuning NLP and large language models (LLMs).

cs.CL↗

arXiv2024

Using Self-supervised Learning Can Improve Model Fairness

Self-supervised learning (SSL) has become the de facto training paradigm of large models, where pre-training is followed by supervised fine-tuning using domain-specific data and labels. Despite demonstrating comparable performance with supervised methods, comprehensive efforts to assess SSL's impact on machine learning fairness (i.e., performing equally on different demographic breakdowns) are lacking. Hypothesizing that SSL models would learn more generic, hence less biased representations, this study explores the impact of pre-training and fine-tuning strategies on fairness. We introduce a fairness assessment framework for SSL, comprising five stages: defining dataset requirements, pre-training, fine-tuning with gradual unfreezing, assessing representation similarity conditioned on demographics, and establishing domain-specific evaluation processes. We evaluate our method's generalizability on three real-world human-centric datasets (i.e., MIMIC, MESA, and GLOBEM) by systematically comparing hundreds of SSL and fine-tuned models on various dimensions spanning from the intermediate representations to appropriate evaluation metrics. Our findings demonstrate that SSL can significantly improve model fairness, while maintaining performance on par with supervised methods-exhibiting up to a 30% increase in fairness with minimal loss in performance through self-supervision. We posit that such differences can be attributed to representation dissimilarities found between the best- and the worst-performing demographics across models-up to x13 greater for protected attributes with larger performance discrepancies between segments.

cs.LG↗

arXiv2024

Evaluating Fairness in Self-supervised and Supervised Models for Sequential Data

Self-supervised learning (SSL) has become the de facto training paradigm of large models where pre-training is followed by supervised fine-tuning using domain-specific data and labels. Hypothesizing that SSL models would learn more generic, hence less biased, representations, this study explores the impact of pre-training and fine-tuning strategies on fairness (i.e., performing equally on different demographic breakdowns). Motivated by human-centric applications on real-world timeseries data, we interpret inductive biases on the model, layer, and metric levels by systematically comparing SSL models to their supervised counterparts. Our findings demonstrate that SSL has the capacity to achieve performance on par with supervised methods while significantly enhancing fairness--exhibiting up to a 27% increase in fairness with a mere 1% loss in performance through self-supervision. Ultimately, this work underscores SSL's potential in human-centric computing, particularly high-stakes, data-scarce application domains like healthcare.

cs.LG↗

arXiv2023

Negotiation Strategies in Ubiquitous Human-Computer Interaction: A Novel Storyboards Scale & Field Study

In today's connected society, self-tracking technologies (STTs), such as wearables and mobile fitness apps, empower humans to improve their health and well-being through ubiquitous physical activity monitoring, with several personal and societal benefits. Despite the advances in such technologies' hardware, low user engagement and decreased effectiveness limitations demand more informed and theoretically-founded Human-Computer Interaction designs. To address these challenges, we build upon the previously unexplored Leisure Constraints Negotiation Model and the Transtheoretical Model to systematically define and assess the effectiveness of STTs' features that acknowledge users' contextual constraints and establish human-negotiated STTs narratives. Specifically, we introduce and validate a human-centric scale, StoryWear, which exploits and explores eleven dimensions of negotiation strategies that humans utilize to overcome constraints regarding exercise participation, captured through an inclusive storyboards format. Based on our preliminary studies, StoryWear shows high reliability, rendering it suitable for future work in ubiquitous computing. Our results indicate that negotiation strategies vary in perceived effectiveness and have higher appeal for existing STTs' users, with self-motivation, commitment, and understanding of the negative impact of non-exercise placed at the top. Finally, we give actionable guidelines for real-world implementation and a commentary on the future of personalized training.

cs.HC↗

arXiv2023

FairComp: Workshop on Fairness and Robustness in Machine Learning for Ubiquitous Computing

How can we ensure that Ubiquitous Computing (UbiComp) research outcomes are both ethical and fair? While fairness in machine learning (ML) has gained traction in recent years, fairness in UbiComp remains unexplored. This workshop aims to discuss fairness in UbiComp research and its social, technical, and legal implications. From a social perspective, we will examine the relationship between fairness and UbiComp research and identify pathways to ensure that ubiquitous technologies do not cause harm or infringe on individual rights. From a technical perspective, we will initiate a discussion on data practices to develop bias mitigation approaches tailored to UbiComp research. From a legal perspective, we will examine how new policies shape our community's work and future research. We aim to foster a vibrant community centered around the topic of responsible UbiComp, while also charting a clear path for future research endeavours in this field.

cs.CY↗

arXiv2023

The State of Algorithmic Fairness in Mobile Human-Computer Interaction

This paper explores the intersection of Artificial Intelligence and Machine Learning (AI/ML) fairness and mobile human-computer interaction (MobileHCI). Through a comprehensive analysis of MobileHCI proceedings published between 2017 and 2022, we first aim to understand the current state of algorithmic fairness in the community. By manually analyzing 90 papers, we found that only a small portion (5%) thereof adheres to modern fairness reporting, such as analyses conditioned on demographic breakdowns. At the same time, the overwhelming majority draws its findings from highly-educated, employed, and Western populations. We situate these findings within recent efforts to capture the current state of algorithmic fairness in mobile and wearable computing, and envision that our results will serve as an open invitation to the design and development of fairer ubiquitous technologies.

cs.HC↗

arXiv2023

Uncovering Bias in Personal Informatics

Personal informatics (PI) systems, powered by smartphones and wearables, enable people to lead healthier lifestyles by providing meaningful and actionable insights that break down barriers between users and their health information. Today, such systems are used by billions of users for monitoring not only physical activity and sleep but also vital signs and women's and heart health, among others. Despite their widespread usage, the processing of sensitive PI data may suffer from biases, which may entail practical and ethical implications. In this work, we present the first comprehensive empirical and analytical study of bias in PI systems, including biases in raw data and in the entire machine learning life cycle. We use the most detailed framework to date for exploring the different sources of bias and find that biases exist both in the data generation and the model learning and implementation streams. According to our results, the most affected minority groups are users with health issues, such as diabetes, joint issues, and hypertension, and female users, whose data biases are propagated or even amplified by learning models, while intersectional biases can also be observed.

cs.CY↗

arXiv2023

UBIWEAR: An end-to-end, data-driven framework for intelligent physical activity prediction to empower mHealth interventions

It is indisputable that physical activity is vital for an individual's health and wellness. However, a global prevalence of physical inactivity has induced significant personal and socioeconomic implications. In recent years, a significant amount of work has showcased the capabilities of self-tracking technology to create positive health behavior change. This work is motivated by the potential of personalized and adaptive goal-setting techniques in encouraging physical activity via self-tracking. To this end, we propose UBIWEAR, an end-to-end framework for intelligent physical activity prediction, with the ultimate goal to empower data-driven goal-setting interventions. To achieve this, we experiment with numerous machine learning and deep learning paradigms as a robust benchmark for physical activity prediction tasks. To train our models, we utilize, "MyHeart Counts", an open, large-scale dataset collected in-the-wild from thousands of users. We also propose a prescriptive framework for self-tracking aggregated data preprocessing, to facilitate data wrangling of real-world, noisy data. Our best model achieves a MAE of 1087 steps, 65% lower than the state of the art in terms of absolute error, proving the feasibility of the physical activity prediction task, and paving the way for future research.

cs.AI↗

arXiv2022

12 Years of Self-tracking for Promoting Physical Activity from a User Diversity Perspective: Taking Stock and Thinking Ahead

Despite the indisputable personal and societal benefits of regular physical activity, a large portion of the population does not follow the recommended guidelines, harming their health and wellness. The World Health Organization has called upon governments, practitioners, and researchers to accelerate action to address the global prevalence of physical inactivity. To this end, an emerging wave of research in ubiquitous computing has been exploring the potential of interactive self-tracking technology in encouraging positive health behavior change. Numerous findings indicate the benefits of personalization and inclusive design regarding increasing the motivational appeal and overall effectiveness of behavior change systems, with the ultimate goal of empowering and facilitating people to achieve their goals. However, most interventions still adopt a "one-size-fits-all" approach to their design, assuming equal effectiveness for all system features in spite of individual and collective user differences. To this end, we analyze a corpus of 12 years of research in self-tracking technology for health behavior change, focusing on physical activity, to identify those design elements that have proven most effective in inciting desirable behavior across diverse population segments. We then provide actionable recommendations for designing and evaluating behavior change self-tracking technology based on age, gender, occupation, fitness, and health condition. Finally, we engage in a critical commentary on the diversity of the domain and discuss ethical concerns surrounding tailored interventions and directions for moving forward.

cs.HC↗

arXiv2022

14 Years of Self-Tracking Technology for mHealth -- Literature Review: Lessons Learnt and the PAST SELF Framework

In today's connected society, many people rely on mHealth and self-tracking (ST) technology to help them adopt healthier habits with a focus on breaking their sedentary lifestyle and staying fit. However, there is scarce evidence of such technological interventions' effectiveness, and there are no standardized methods to evaluate their impact on people's physical activity (PA) and health. This work aims to help ST practitioners and researchers by empowering them with systematic guidelines and a framework for designing and evaluating technological interventions to facilitate health behavior change (HBC) and user engagement (UE), focusing on increasing PA and decreasing sedentariness. To this end, we conduct a literature review of 129 papers between 2008 and 2022, which identifies the core ST HCI design methods and their efficacy, as well as the most comprehensive list to date of UE evaluation metrics for ST. Based on the review's findings, we propose PAST SELF, a framework to guide the design and evaluation of ST technology that has potential applications in industrial and scientific settings. Finally, to facilitate researchers and practitioners, we complement this paper with an open corpus and an online, adaptive exploration tool for the PAST SELF data.

cs.HC↗