In the last decade, Federated Learning (FL) has gained relevance in training collaborative models without sharing sensitive data. Since its birth, Centralized FL (CFL) has been the most common approach in the literature, where a unique entity creates global models. However, using a centralized approach has the disadvantages of bottleneck at the server node, single point of failure, and trust needs. Decentralized Federated Learning (DFL) arose to solve these aspects by embracing the principles of data sharing minimization and decentralized model aggregation without relying on centralized architectures. However, despite the work done in DFL, the literature has not (i) studied the main fundamentals differentiating DFL and CFL; (ii) reviewed application scenarios and solutions using DFL; and (iii) analyzed DFL frameworks to create and evaluate new solutions. To this end, this article identifies and analyzes the main fundamentals of DFL in terms of federation architectures, topologies, communication mechanisms, security approaches, and key performance indicators. Additionally, the paper at hand explores existing mechanisms to optimize critical DFL fundamentals. Then, this work analyzes and compares the most used DFL application scenarios and solutions according to the fundamentals previously defined. After that, the most relevant features of the current DFL frameworks are reviewed and compared. Finally, the evolution of existing DFL solutions is analyzed to provide a list of trends, lessons learned, and open challenges.
Federated learning (FL) allows participants to collaboratively train machine and deep learning models while protecting data privacy. However, the FL paradigm still presents drawbacks affecting its trustworthiness since malicious participants could launch adversarial attacks against the training process. Related work has studied the robustness of horizontal FL scenarios under different attacks. However, there is a lack of work evaluating the robustness of decentralized vertical FL and comparing it with horizontal FL architectures affected by adversarial attacks. Thus, this work proposes three decentralized FL architectures, one for horizontal and two for vertical scenarios, namely HoriChain, VertiChain, and VertiComb. These architectures present different neural networks and training protocols suitable for horizontal and vertical scenarios. Then, a decentralized, privacy-preserving, and federated use case with non-IID data to classify handwritten digits is deployed to evaluate the performance of the three architectures. Finally, a set of experiments computes and compares the robustness of the proposed architectures when they are affected by different data poisoning based on image watermarks and gradient poisoning adversarial attacks. The experiments show that even though particular configurations of both attacks can destroy the classification performance of the architectures, HoriChain is the most robust one.
Drowsiness is a major concern for drivers and one of the leading causes of traffic accidents. Advances in Cognitive Neuroscience and Computer Science have enabled the detection of drivers' drowsiness by using Brain-Computer Interfaces (BCIs) and Machine Learning (ML). Nevertheless, several challenges remain open and should be faced. First, a comprehensive enough evaluation of drowsiness detection performance using a heterogeneous set of ML algorithms is missing in the literature. Last, it is needed to study the detection performance of scalable ML models suitable for groups of subjects and compare it with the individual models proposed in the literature. To improve these limitations, this work presents an intelligent framework that employs BCIs and features based on electroencephalography (EEG) for detecting drowsiness in driving scenarios. The SEED-VIG dataset is used to feed different ML regressors and three-class classifiers and then evaluate, analyze, and compare the best-performing models for individual subjects and groups of them. More in detail, regarding individual models, Random Forest (RF) obtained a 78% f1-score, improving the 58% obtained by models used in the literature such as Support Vector Machine (SVM). Concerning scalable models, RF reached a 79% f1-score, demonstrating the effectiveness of these approaches. The lessons learned can be summarized as follows: i) not only SVM but also other models not sufficiently explored in the literature are relevant for drowsiness detection, and ii) scalable approaches suitable for groups of subjects are effective to detect drowsiness, even when new subjects that are not included in the models training are evaluated.
Traffic accidents are the leading cause of death among young people, a problem that today costs an enormous number of victims. Several technologies have been proposed to prevent accidents, being Brain-Computer Interfaces (BCIs) one of the most promising. In this context, BCIs have been used to detect emotional states, concentration issues, or stressful situations, which could play a fundamental role in the road since they are directly related to the drivers' decisions. However, there is no extensive literature applying BCIs to detect subjects' emotions in driving scenarios. In such a context, there are some challenges to be solved, such as (i) the impact of performing a driving task on the emotion detection and (ii) which emotions are more detectable in driving scenarios. To improve these challenges, this work proposes a framework focused on detecting emotions using electroencephalography with machine learning and deep learning algorithms. In addition, a use case has been designed where two scenarios are presented. The first scenario consists in listening to sounds as the primary task to perform, while in the second scenario listening to sound becomes a secondary task, being the primary task using a driving simulator. In this way, it is intended to demonstrate whether BCIs are useful in this driving scenario. The results improve those existing in the literature , achieving 99% accuracy for the detection of two emotions (non-stimuli and angry), 93% for three emotions (non-stimuli, angry and neutral) and 75% for four emotions (non-stimuli, angry, neutral and joy).
Brain-Computer Interfaces are devices that enable two-way communication between an individual's brain and external devices, allowing the acquisition of neural activity and neurostimulation. Considering the first one, electroencephalographic signals are widely used for the acquisition of subjects' information. Therefore, a manipulation of the data acquired by a vulnerable BCI framework may cause a malfunction of the deployed applications. In this regard, this paper defines four noise-based cyberattacks attempting to generate fake P300 waves in two different phases of a BCI framework. A set of experiments show that the greater the attacker's knowledge regarding the P300 waves, processes, and data of the BCI framework, the higher the attack impact. In this sense, the attacker with less knowledge impacts 1% in the acquisition phase and 4% in the processing phase, while the attacker with the most knowledge impacts 22% and 74%, respectively.
Brain-Computer Interfaces (BCIs) are bidirectional devices that have allowed people to control computers or external devices through their brain activity. The P300 Speller is one of the most widely used BCI applications, where subjects can transmit textual information mentally with satisfactory performance. However, the P300 Speller still has room for improvement in practical use, such as selecting the best balance between accuracy and speed. Based on a lack of literature in this direction, this study evaluates two distinct approaches to the P300 Speller. The first is based on rows and columns following the traditional implementation, while the second is based on regions, employing subsets of characters during spelling. In both approaches, the effects of two different stimulus presentation parameters (the number of repetitions per stimulus and the interval between them) on the accuracy and performance efficiency of the P300 Speller are studied. The results show that both approaches obtain similar values in terms of detection performance, obtaining around 75% F1-score for predicting a character with four series of 12 blinks per character. In addition, the region-based approach presents a more robust scheme for false predictions, maintaining a similar spelling duration. The theoretical study performed indicates that spelling a character requires around one minute.
As recently reported by the World Health Organization (WHO), the high use of intelligent devices such as smartphones, multimedia systems, or billboards causes an increase in distraction and, consequently, fatal accidents while driving. The use of EEG-based Brain-Computer Interfaces (BCIs) has been proposed as a promising way to detect distractions. However, existing solutions are not well suited for driving scenarios. They do not consider complementary data sources, such as contextual data, nor guarantee realistic scenarios with real-time communications between components. This work proposes an automatic framework for detecting distractions using BCIs and a realistic driving simulator. The framework employs different supervised Machine Learning (ML)-based models on classifying the different types of distractions using Electroencephalography (EEG) and contextual driving data collected by car sensors, such as line crossings or objects detection. This framework has been evaluated using a driving scenario without distractions and a similar one where visual and cognitive distractions are generated for ten subjects. The proposed framework achieved 83.9% F1-score with a binary model and 73% with a multiclass model using EEG, improving 7% in binary classification and 8% in multi-class classification by incorporating contextual driving into the training dataset. Finally, the results were confirmed by a neurophysiological study, which revealed significantly higher voltage in selective attention and multitasking.
In recent years, the growth of brain-computer interfaces (BCIs) has been remarkable in specific application fields, such as the medical sector or the entertainment industry. Most of these fields use evoked potentials, like P300, to obtain neural data able to handle prostheses or achieve greater immersion experience in videogames. The natural use of BCI involves the management of sensitive users' information as behaviors, emotions, or thoughts. In this context, new security breaches in BCI are offering cybercriminals the possibility of collecting sensitive data and affecting subjects' physical integrity, which are critical issues. For all these reasons, the fact of applying efficient cybersecurity mechanisms has become a main challenge. To improve this challenge, this chapter proposes a framework able to detect cyberattacks affecting one of the most typical scenarios of BCI, the generation of P300 through visual stimuli. A pool of experiments demonstrates the performance of the proposed framework.
Brain-computer interfaces (BCIs) started being used in clinical scenarios, reaching nowadays new fields such as entertainment or learning. Using BCIs, neuronal activity can be monitored for various purposes, with the study of the central nervous system response to certain stimuli being one of them, being the case of evoked potentials. However, due to the sensitivity of these data, the transmissions must be protected, with blockchain being an interesting approach to ensure the integrity of the data. This work focuses on the visual sense, and its relationship with the P300 evoked potential, where several open challenges related to the privacy of subjects' information and thoughts appear when using BCI. The first and most important challenge is whether it would be possible to extract sensitive information from evoked potentials. This aspect becomes even more challenging and dangerous if the stimuli are generated when the subject is not aware or conscious that they have occurred. There is an important gap in this regard in the literature, with only one work existing dealing with subliminal stimuli and BCI and having an unclear methodology and experiment setup. As a contribution of this paper, a series of experiments, five in total, have been created to study the impact of visual stimuli on the brain tangibly. These experiments have been applied to a heterogeneous group of ten subjects. The experiments show familiar visual stimuli and gradually reduce the sampling time of known images, from supraliminal to subliminal. The study showed that supraliminal visual stimuli produced P300 potentials about 50% of the time on average across all subjects. Reducing the sample time between images degraded the attack, while the impact of subliminal stimuli was not confirmed. Additionally, younger subjects generally presented a shorter response latency. This work corroborates that subjects' sensitive data can be extracted using visual stimuli and P300.
Most of the current Brain-Computer Interfaces (BCIs) application scenarios use electroencephalographic signals (EEG) containing the subject's information. It means that if EEG were maliciously manipulated, the proper functioning of BCI frameworks could be at risk. Unfortunately, it happens in frameworks sensitive to noise-based cyberattacks, and more efforts are needed to measure the impact of these attacks. This work presents and analyzes the impact of four noise-based cyberattacks attempting to generate fake P300 waves in two different phases of a BCI framework. A set of experiments show that the greater the attacker's knowledge regarding the P300 waves, processes, and data of the BCI framework, the higher the attack impact. In this sense, the attacker with less knowledge impacts 1% in the acquisition phase and 4% in the processing phase, while the attacker with the most knowledge impacts 22% and 74%, respectively.
The CyberBrain Project was one of the 6 winning proposals of the 2020 Call for Research Projects organized by Bitbrain. This project aims to identify and exploit BCI cybersecurity issues within an advanced driver assistance scenario, defining three use cases where Bitbrain devices can be applied to this environment. Moreover, CyberBrain will design and implement a framework to detect cyberattacks affecting the BCI lifecycle of Bitbrain products.
Since the first reported case in Wuhan in late 2019, COVID-19 has rapidly spread worldwide, dramatically impacting the lives of millions of citizens. To deal with the severe crisis resulting from the pandemic, worldwide institutions have been forced to make decisions that profoundly affect the socio-economic realm. In this sense, researchers from diverse knowledge areas are investigating the behavior of the disease in a rush against time. In both cases, the lack of reliable data has been an obstacle to carry out such tasks with accuracy. To tackle this challenge, COnVIDa (https://convida.inf.um.es) has been designed and developed as a user-friendly tool that easily gathers rigorous multidisciplinary data related to the COVID-19 pandemic from different data sources. In particular, the pandemic expansion is analyzed with variables of health nature, but also social ones, mobility, etc. Besides, COnVIDa permits to smoothly join such data, compare and download them for further analysis. Due to the open-science nature of the project, COnVIDa is easily extensible to any other region of the planet. In this way, COnVIDa becomes a data facilitator for decision-making processes, as well as a catalyst for new scientific researches related to this pandemic.