Search

This study aims to understand if and how AI-based beauty filters impact the attractiveness halo effect with changes in age, gender and ethnicity

Algorithmic fairness is of utmost societal importance, yet the current trend in large-scale machine learning models requires training with massive datasets that are typically biased. In this context, pre-processing methods that focus on modeling and correcting bias in the data emerge as valuable approaches. In this paper, we propose FairShap, a novel pre-processing (re-weighting) method for fair algorithmic decision-making through data valuation by means of Shapley Values. Our approach is model agnostic and easily interpretable, as it measures the contribution of each training data point to a predefined fairness metric. We empirically validate FairShap on several state-of-the-art datasets of different nature, with a variety of training scenarios and models and show how it outperforms other methods, yielding fairer models with higher or similar levels of accuracy. We also illustrate FairShap's interpretability by means of histograms and latent space visualizations. We believe that this work represents a promising direction in interpretable and model-agnostic approaches to algorithmic fairness that yield competitive accuracy even when only biased datasets are available., Comment: 12 pages, 6 figures, 1 appendix

Human perception, memory and decision-making are impacted by tens of cognitive biases and heuristics that influence our actions and decisions. Despite the pervasiveness of such biases, they are generally not leveraged by today's Artificial Intelligence (AI) systems that model human behavior and interact with humans. In this theoretical paper, we claim that the future of human-machine collaboration will entail the development of AI systems that model, understand and possibly replicate human cognitive biases. We propose the need for a research agenda on the interplay between human cognitive biases and Artificial Intelligence. We categorize existing cognitive biases from the perspective of AI systems, identify three broad areas of interest and outline research directions for the design of AI systems that have a better understanding of our own biases., Comment: 14 pages, 1 figure; Accepted for presentation at the AAAI Fall Symposium 2022 on Thinking Fast and Slow and Other Cognitive Theories in AI. Corrected typos

This short paper proposes a preliminary and yet insightful investigation of racial biases in beauty filters techniques currently used on social media. The obtained results are a call to action for researchers in Computer Vision: such biases risk being replicated and exaggerated in the Metaverse and, as a consequence, they deserve more attention from the community., Comment: To be published at the CV4Metaverse workshop at ECCV

Graph Neural Networks (GNNs) have been shown to achieve competitive results to tackle graph-related tasks, such as node and graph classification, link prediction and node and graph clustering in a variety of domains. Most GNNs use a message passing framework and hence are called MPNNs. Despite their promising results, MPNNs have been reported to suffer from over-smoothing, over-squashing and under-reaching. Graph rewiring and graph pooling have been proposed in the literature as solutions to address these limitations. However, most state-of-the-art graph rewiring methods fail to preserve the global topology of the graph, are neither differentiable nor inductive, and require the tuning of hyper-parameters. In this paper, we propose DiffWire, a novel framework for graph rewiring in MPNNs that is principled, fully differentiable and parameter-free by leveraging the Lovász bound. The proposed approach provides a unified theory for graph rewiring by proposing two new, complementary layers in MPNNs: CT-Layer, a layer that learns the commute times and uses them as a relevance function for edge re-weighting; and GAP-Layer, a layer to optimize the spectral gap, depending on the nature of the network and the task at hand. We empirically validate the value of each of these layers separately with benchmark datasets for graph classification. We also perform preliminary studies on the use of CT-Layer for homophilic and heterophilic node classification tasks. DiffWire brings together the learnability of commute times to related definitions of curvature, opening the door to creating more expressive MPNNs., 27 pages, 24 figures and 6 tables. Accepted at Learning on Graphs Conference 2022. A. Arnaiz-Rodriguez et al., DiffWire: Inductive Graph Rewiring via the Lov\'asz Bound. Proceedings of the First Learning on Graphs Conference (LoG 2022), PMLR 198, Virtual Event, December, 2022

Augmented Reality or AR filters on selfies have become very popular on social media platforms for a variety of applications, including marketing, entertainment and aesthetics. Given the wide adoption of AR face filters and the importance of faces in our social structures and relations, there is increased interest by the scientific community to analyze the impact of such filters from a psychological, artistic and sociological perspective. However, there are few quantitative analyses in this area mainly due to a lack of publicly available datasets of facial images with applied AR filters. The proprietary, close nature of most social media platforms does not allow users, scientists and practitioners to access the code and the details of the available AR face filters. Scraping faces from these platforms to collect data is ethically unacceptable and should, therefore, be avoided in research. In this paper, we present OpenFilter, a flexible framework to apply AR filters available in social media platforms on existing large collections of human faces. Moreover, we share FairBeauty and B-LFW, two beautified versions of the publicly available FairFace and LFW datasets and we outline insights derived from the analysis of these beautified datasets.

Statistics on migration flows are often derived from census data, which suffer from intrinsic limitations, including costs and infrequent sampling. When censuses are used, there is typically a time gap - up to a few years - between the data collection process and the computation and publication of relevant statistics. This gap is a significant drawback for the analysis of a phenomenon that is continuously and rapidly changing. Alternative data sources, such as surveys and field observations, also suffer from reliability, costs, and scale limitations. The ubiquity of mobile phones enables an accurate and efficient collection of up-to-date data related to migration. Indeed, passively collected data by the mobile network infrastructure via aggregated, pseudonymized Call Detail Records (CDRs) is of great value to understand human migrations. Through the analysis of mobile phone data, we can shed light on the mobility patterns of migrants, detect spontaneous settlements and understand the daily habits, levels of integration, and human connections of such vulnerable social groups. This Chapter discusses the importance of leveraging mobile phone data as an alternative data source to gather precious and previously unavailable insights on various aspects of migration. Also, we highlight pending challenges that would need to be addressed before we can effectively benefit from the availability of mobile phone data to help make better decisions that would ultimately improve millions of people's lives., Comment: To appear as a book chapter in "Data Science for Migration and Mobility Studies" edited by Dr. Emre Eren Korkmaz, Dr. Albert Ali Salah

In complex industrial settings, it is common practice to monitor the operation of machines in order to detect undesired states, adjust maintenance schedules, optimize system performance or collect usage statistics of individual machines. In this work, we focus on estimating the power output of a Combined Heat and Power (CHP) machine of a medium-sized company facility by analyzing the total facility power consumption. We formulate the problem as a time-series classification problem, where the class label represents the CHP power output. As the facility is fully instrumented and sensor measurements from the CHP are available, we generate the training labels in an automated fashion from the CHP sensor readings. However, sensor failures result in mislabeled training data samples which are hard to detect and remove from the dataset. Therefore, we propose a novel multi-task deep learning approach that jointly trains a classifier and an autoencoder with a shared embedding representation. The proposed approach targets to gradually correct the mislabelled data samples during training in a self-supervised fashion, without any prior assumption on the amount of label noise. We benchmark our approach on several time-series classification datasets and find it to be comparable and sometimes better than state-of-the-art methods. On the real-world use-case of predicting the CHP power output, we thoroughly evaluate the architectural design choices and show that the final architecture considerably increases the robustness of the learning process and consistently beats other recent state-of-the-art algorithms in the presence of unstructured as well as structured label noise.

The coronavirus 2019–2020 pandemic (COVID-19) poses unprecedented challenges for governments and societies around the world ( 1 ). Nonpharmaceutical interventions have proven to be critical for delaying and containing the COVID-19 pandemic ( 2 – 6 ). These include testing and tracing, bans on large gatherings, nonessential business and school and university closures, international and domestic mobility restrictions and physical isolation, and total lockdowns of regions and countries. Decision-making and evaluation or such interventions during all stages of the pandemic life cycle require specific, reliable, and timely data not only about infections but also about human behavior, especially mobility and physical copresence. We argue that mobile phone data, when used properly and carefully, represents a critical arsenal of tools for supporting public health actions across early-, middle-, and late-stage phases of the COVID-19 pandemic. Seminal work on human mobility has shown that aggregate and (pseudo-)anonymized mobile phone data can assist the modeling of the geographical spread of epidemics ( 7 – 11 ). Thus, researchers and governments have started to collaborate with private companies, most notably mobile network operators and location intelligence companies, to estimate the effectiveness of control measures in a number of countries, including Austria, Belgium, Chile, China, Germany, France, Italy, Spain, United Kingdom, and the United States ( 12 – 21 ). There is, however, little coordination or information exchange between these national or even regional initiatives ( 22 ). Although ad hoc mechanisms leveraging mobile phone data can be effectively (but not easily) developed at the local or national level, regional or even global collaborations seem to be much more difficult given the number of actors, the range of interests and priorities, the variety of legislations concerned, and the need to protect civil liberties. The global scale and spread of the COVID-19 pandemic highlight the need for a more harmonized or coordinated approach. In the …

This paper describes how mobile phone data can guide government and public health authorities in determining the best course of action to control the COVID-19 pandemic and in assessing the effectiveness of control measures such as physical distancing. It identifies key gaps and reasons why this kind of data is only scarcely used, although their value in similar epidemics has proven in a number of use cases. It presents ways to overcome these gaps and key recommendations for urgent action, most notably the establishment of mixed expert groups on national and regional level, and the inclusion and support of governments and public authorities early on. It is authored by a group of experienced data scientists, epidemiologists, demographers and representatives of mobile network operators who jointly put their work at the service of the global effort to combat the COVID-19 pandemic.

Physical inactivity is an increasingly serious global health problem, which implies a strong need for effective and engaging interventions. Smartphone technology offers new possibilities to address physical activity promotion. For app-based interventions to have an impact, both the effectiveness and user appreciation of the app are important. In this paper, we explore the user appreciation of the Active2Gether intervention, which offers personalized coaching to increase physical activity levels in daily life. The results are compared to the evaluation of a simplified version of the Active2Gether app (in which no coaching messages are sent) and the Fitbit app. Overall, the results reveal that users of a physical activity app appreciate a coaching feature to be included (on top of self-monitoring functionalities), but are also critical of how it is implemented (in terms of the number and content of the messages). The results also show that it is important to find a balance in the number of messages sent: too many messages seem to be perceived as annoying, but on the other hand, such system-initiated user interaction seems to reduce dropout.

This research examines how wearable technology and supporting applications can help office workers maintain good posture and guide them to carry out shoulder exercises at their workplace. Specifically, we describe a smart garment designed to monitor upper body posture that provides vibrotactile notifications at different joint areas in order to remind users to correct their posture. We present the design and evaluation of a related smartphone application that supports shoulder training exercises to treat and prevent shoulder pain. The usability of the system for shoulder training was evaluated positively in a laboratory test (N=17). The effectiveness of the system for posture monitoring was assessed with a field deployment (N=25) in which students working with laptops used the posture monitoring system for a whole day. The results demonstrate the system can help the participants facilitate improving their posture in sedentary work.

This paper describes a real world deployment of a context-aware mobile app recommender system (RS) called Frappe. Utilizing a hybrid-approach, we conducted a large-scale app market deployment with 1000 Android users combined with a small-scale local user study involving 33 users. The resulting usage logs and subjective feedback enabled us to gather key insights into (1) context-dependent app usage and (2) the perceptions and experiences of end-users while interacting with context-aware mobile app recommendations. While Frappe performs very well based on usage-centric evaluation metrics insights from the small-scale study reveal some negative user experiences. Our results point to a number of actionable lessons learned specifically related to designing, deploying and evaluating mobile context-aware RS in-the-wild with real users.

Este estudio se realizó gracias a la activación del "SERVICIO DE MUESTREO Y ANÁLISIS PARA LA RED DE CONTROL Y VIGILANCIA DE LA CALIDAD DE LAS AGUAS DE BAÑO DE LA REGIÓN DE MURCIA". Se evalúa la situación de las aguas costeras de la Región de Murcia mediante el análisis de parámetros físicos, químicos y biológicos del agua de mar. El muestreo se realizó en trece playas o zonas costeras del litoral, generalmente asociadas a zonas de vertidos, durante los meses de verano del 2009. Con el objetivo de poder establecer los criterios sanitarios que deben cumplir las aguas de baño para garantizar su calidad. En el caso del Mar Menor, sus aguas se pueden ver afectadas especialmente por el fenómeno de eutrofización debido al aporte de nutrientes que suministran las principales ramblas que en él desembocan. A su vez se estudió la posible correlación entre las diversas variables medidas para poder realizar una comparativa entre el Mar Menor y el Mediterráneo. Asociación de Jóvenes Investigadores de Cartagena, (AJICT). Universidad Politécnica de Cartagena. Escuela Técnica Superior de Ingeniería Industrial UPCT, (ETSII). Escuela Técnica Superior de Ingeniería Agronómica, (ETSIA), Escuela Técnica Superior de Ingeniería de Telecomunicación (ETSIT). Escuela de Ingeniería de Caminos, Canales, y Puertos y de Ingeniería de Minas, (EICM). Fundación Séneca, Agencia Regional de Ciencia y Tecnología. Parque Tecnológico de Fuente Álamo. Grupo Aquiline

In this study, we argue that modality planning in multimodal presentation systems needs to consider the modality characteristics at not only the presentational level but also the cognitive level, especially in a situation where the information load is high and the user task is time-critical. As a first step towards automatic cognitive-aware modality planning, we integrated the effect of different modalities on cognitive load and performance, using a high-load information presentation scenario. Mainly based on modality-related psychology theories, we selected five modality conditions (text, image, text+image, text+speech, and text+sound) and made hypotheses about their effects on cognitive load. Modality effects were evaluated by two cognitive load measurements and two performance measurements. Results confirmed most of the predicted modality effects, and showed that these effects become significant when the information load and the task demand are high. The findings of this study suggest that it is highly necessary to encode modality-related principles of human cognition into the modality planning procedure for systems that support high-load human-computer interaction.

Federated learning (FL) has been proposed as a privacy-preserving approach in distributed machine learning. A federated learning architecture consists of a central server and a number of clients that have access to private, potentially sensitive data. Clients are able to keep their data in their local machines and only share their locally trained model's parameters with a central server that manages the collaborative learning process. FL has delivered promising results in real-life scenarios, such as healthcare, energy, and finance. However, when the number of participating clients is large, the overhead of managing the clients slows down the learning. Thus, client selection has been introduced as a strategy to limit the number of communicating parties at every step of the process. Since the early na\"{i}ve random selection of clients, several client selection methods have been proposed in the literature. Unfortunately, given that this is an emergent field, there is a lack of a taxonomy of client selection methods, making it hard to compare approaches. In this paper, we propose a taxonomy of client selection in Federated Learning that enables us to shed light on current progress in the field and identify potential areas of future research in this promising area of machine learning., Comment: 17 pages, 3 figures, 1 appendix, accepted to TMLR

Basado es el discurso pronunciado por la autora en su ingreso a la Real Academia de Ingeniería, se centra en la Inteligencia Artificial, haciendo un breve recorrido por su historia, describiendo su impacto actual y planteando los retos que presenta desde diferentes perspectivas. Está escrito en primera persona, a fin de llevar de la mano a los lectores en el acercamiento a un tema complejo pero que afecta crecientemente la vida de todos Basado es el discurso pronunciado por la autora en su ingreso a la Real Academia de Ingeniería, se centra en la Inteligencia Artificial, haciendo un breve recorrido por su historia, describiendo su impacto actual y planteando los retos que presenta desde diferentes perspectivas. Está escrito en primera persona, a fin de llevar de la mano a los lectores en el acercamiento a un tema complejo pero que afecta crecientemente la vida de todos

Learning and reasoning over graphs is increasingly done by means of probabilistic models, e.g. exponential random graph models, graph embedding models, and graph neural networks. When graphs are modeling relations between people, however, they will inevitably reflect biases, prejudices, and other forms of inequity and inequality. An important challenge is thus to design accurate graph modeling approaches while guaranteeing fairness according to the specific notion of fairness that the problem requires. Yet, past work on the topic remains scarce, is limited to debiasing specific graph modeling methods, and often aims to ensure fairness in an indirect manner. We propose a generic approach applicable to most probabilistic graph modeling approaches. Specifically, we first define the class of fair graph models corresponding to a chosen set of fairness criteria. Given this, we propose a fairness regularizer defined as the KL-divergence between the graph model and its I-projection onto the set of fair models. We demonstrate that using this fairness regularizer in combination with existing graph modeling approaches efficiently trades-off fairness with accuracy, whereas the state-of-the-art models can only make this trade-off for the fairness criterion that they were specifically designed for.

One potential method of improving the efficiency of human-computer interaction is to display information subliminally. Such information cannot be recalled consciously, but has some impact on the perceiver. However, it is not yet clear whether people can extract meaning from subliminal presentation of information in mobile contexts. We therefore explored subliminal semantic priming on smartphones. This builds on mixed evidence for subliminal priming across HCI in general, and mixed evidence for the effect of subliminal affective priming on smartphones. Our semi-controlled experiment (n=103) investigated subliminal processing of numerical information on smartphones. We found evidence that concealed transfer of information is possible to a very limited extent, but little evidence of a semantic effect. Overall, the impact is effectively negligible for practical applications. We discuss the implications of our results for real-world deployments and outline future research themes as HCI moves beyond mobile.

In this paper the basic information about methodology of assessment human body stiffness during vertical jump analysis is presented. Ten subjects (professional football player) perform ten periodic jumps and vertical ground reaction force is measured using a force plate with one axial load cell force sensor. The stiffness calculation is based on the analogy of the periodic jumping and oscillation movement of the system which consists of spring and body appropriate mass. The frequency of oscillation is obtained using Fourier transform.