Different mental states result in different synchronizations or desynchronizations between multiple brain regions, and subsequently, electroencephalogram (EEG) connectivity analysis gains increasing attention in brain computer interfaces (BCIs). Conventional connectivity analysis is usually conducted at the scalp-level and in an unsupervised manner. However, due to the volume conduction effect, EEG data suffer from low signal-to-noise ratio and poor spatial resolution. Thus, it is hard to effectively identify the task-related connectivity pattern at the scalp-level using unsupervised method. There exist extensive discriminative spatial filtering methods for different BCI paradigms. However, in conventional spatial filter optimization methods, signal correlations or connectivities are not taken into consideration in the objective functions. To address the issue, in this work, we propose a discriminative connectivity pattern-learning method. In the proposed framework, EEG correlations are used as the features, with which Fisher's ratio objective function is adopted to optimize spatial filters. The proposed method is evaluated with a binary motor imagery EEG dataset. Experimental results show that more connectivity information are maintained with the proposed method, and classification accuracies yielded by the proposed method are comparable to conventional discriminative spatial filtering method.

Brain decoding has contributed to the development of cognitive neuroscience and the production of brain-machine interfaces/brain-computer interfaces (BCI/BMI). For brain decoding, electroencephalography (EEG), which allows the observation of the electrophysiological activities of neurons, is widely used to observe brain activity. In particular, an EEG read from electrodes installed on a scalp, which has some advantages when it comes to cost, size, and ease of measurement, is a promising recording method for producing noninvasive BMIs against magnetoencephalograms, functional magnetic resonance imaging, and so on. However, an EEG signal has low spatial resolution and is highly affected by noise. Moreover, the recoding of an EEG is time-consuming and tires BMI users. Therefore, signal processing techniques that can robustly extract brain activity patterns from EEG signals with a low signal-to-noise ratio and a small sample size are necessary. One approach to achieve such techniques is to incorporate additional information retrieved separately from an EEG in signal processing. This chapter describes some techniques that can accomplish this. The signal structures include physical structures, such as the location of the electrodes, and functional structures, such as synchronizing brain regions. The discussion on the importance of the signal structures in a source analysis of EEG signals that can improve the BMI performance will be discussed in this chapter. In contrast to a source analysis, which finds the brain patterns in a source domain, regularization incorporating the signal structures in a sensor domain will be discussed. Moreover, we will show the signal processing in a graph spectral domain that is a vector space derived from the signal structure.

Machine learning has become a core component in brain-computer interfaces (BCIs). Unfortunately, the use of machine learning typically requires the collection of subject specific labelled data. This process is time-consuming and not productive from a user's point of view, as during calibration the user has to follow given instructions and cannot make own decisions. Only after calibration, the user is able to use the BCI freely. In this chapter, we describe how the supervised calibration process can be circumvented by unsupervised learning in which the decoder is trained while the user is utilising the system. We discuss three variations. First, expectation-maximisation (EM)-based training, which works wells empirically but can sometimes be unstable. Learning from label proportions (LLP)-based training, which is guaranteed to converge to the optimal solution, but learns more slowly. Third, a hybrid approach combining the stability of LLP with the speed of learning of EM in a highly efficient and effective approach that can readily replace supervised decoders for event-related potential BCI.

Electroencephalography (EEG)-based brain-computer interfaces (BCIs) have proven promising for a wide range of applications, from communication and control for motor impaired users to gaming targeted at the general public, real-time mental state monitoring and stroke rehabilitation, to name a few. Despite this promising potential, BCIs are still scarcely used outside laboratories for practical applications. The main reason preventing EEG-based BCIs from being widely used is arguably their poor usability, which is notably due to their low robustness and reliability. To operate a BCI, the user has to encode commands in his/her EEG signals, typically using mental imagery tasks, such as imagining hand movement or mental calculation. The execution of these tasks leads to specific EEG patterns, which the machine has to decode by using signal processing and machine learning. So far, to address the reliability issue of BCI, most research efforts have been focused on command decoding only. However, if the user is unable to encode commands in her EEG patterns, no signal-processing algorithm would be able to decode them. Therefore, we argue in this chapter that BCI design is not only a decoding challenge (i.e., translating EEG signals into control commands) but also a human-computer interaction challenge, which aims at ensuring the user can control the BCI. Interestingly enough, there are a number of open challenges to take the user into account, for which signal-processing and machine-learning methods could provide solutions. These challenges notably concerns (1) the modeling of the user and (2) understanding and improving how and what the user is learning. More precisely, the BCI community should first work on user modeling, i.e., modeling and updating the user's states and skills overtime from his/herEEG signals, behavior, BCI performances and possibly other sensors. The community should also identify new performance metrics-beyond classification accuracy-that could better describe users' skills at BCI control. Second, the BCI community has to understand how and what the user learns to control the BCI. This includes thoroughly identifying the features to be extracted and the classifier to be used to ensure the user's understanding of the feedback resulting from them, as well as how to present this feedback. Being able to update machine-learning parameters in a specific manner and a precise moment to favor learning without confusing the user with the ever-changeable feedback is another challenge. Finally, it is necessary to gain a clearer understanding of the reasons why mental commands are sometimes correctly decoded and sometimes not; what makes people sometimes fail at BCI control, in order to be able to guide them to do better. Overall, this chapter identifies a number of open and important challenges for the BCI community, at the user level, to which experts in machine learning and signal processing could contribute.

Electroencephalography (EEG)-based brain-computer interfaces (BCIs) are developed to provide access channels for alternative communication and control systems to people with severe speech and physical impairments. Designs that exploit evoked response potentials (ERPs) in EEG constitute the majority of research efforts dedicated to noninvasive BCIs. Visual, auditory, and tactile stimulation paradigms are used to actively probe the user's brain to collect EEG evidence towards inferring intent in the context of the particular application. As assistive technology devices, however, existing EEG-based BCIs lack sufficient speed and accuracy to safely and reliably restore function at acceptable levels. This is mainly because the recorded EEG signals are not only noisy with a low signal-to-noise ratio, but are also nonstationary, due to physiological or environmental artifacts, sensor failure, and user fatigue. In this chapter, we address how reliable intent inference engines with reasonable speed and accuracy can be developed using parametric modeling. Examples of real-world data in the framework of the ERP-based BCI paradigm are provided to exemplify our detection and classification methods.

In this chapter, we introduce an architecture for rapid serial visual presentation (RSVP)-based brain-computer interface (BCI) systems that use electroencephalography (EEG). Hereafter, we will refer to the coupling of the RSVP protocol with EEG to support a target-search BCI as RSVP-EEG. Our focus in this chapter is on a review of feature extraction and classification algorithms applied in RSVP-EEG development. We briefly present the commonly deployed algorithms and describe their properties based on the literature. We conclude with a discussion on the future trajectory of this exciting branch of BCI research.

Brain-computer interface (BCI) is relatively a new approach to communication between man and machine, which translates brain activity into commands for communication and control. As BCI is capable of detecting human intentions, it is a promising communication tool for paralyzed patients for communicating with external world. Many of the current BCI systems employ electroencephalogram (EEG) which is the most widely used noninvasive brain activity recording technique. EEG signal carries potential features to identify and decode human intentions and mental tasks. Recently, many researchers have started exploiting the possibilities of BCI in entertainment and cognitive skill enhancement. BCI-based games have been identified as a unique entertainment mechanism nowadays, “controlling a 2-D, 3-D or virtual computer game solely by player's brain waves.” BCI games work based on a neurofeedback paradigm which allows an individual to self-regulate his brain signal in response to the real-time visual or auditory feedback of his brain waves/features. This neurofeedback in a gaming environment motivates and trains the players to control his brain features toward the desired stage (self-regulation). This chapter explores the state-of-the-art BCI technology in neurofeedback games, employing EEG signal. It also provides a survey of the existing EEG-based neurofeedback games and evaluates their success rates, challenging factors and influence on players. In neurofeedback games, a number of features extracted from EEG accompanied with sustained attention, selective attention, visuospatial attention, motor imagery, eye movements, etc. have been employed as distinct control signals. We will briefly review and compare various signal processing methodologies and machine-learning techniques employed in those studies to extract and decode the brain features. Besides the structure and algorithms used in neurofeedback games, the therapeutic effects of neurofeedback training and its capabilities for the enhancement of cognitive skills will also be briefly discussed in this chapter. Neurofeedback training helps to rewire brain's underlying neural circuits and to improve brain functions. Therefore, it is considered as an effective tool for boosting cognitive skills of both healthy and the disabled. Specifically, neurofeedback has been considered as an efficient treatment modality for individuals with attention-deficit hyperactive disorder (ADHD). ADHD is characterized by three behavioral symptoms: inattention, hyperactivity and impulsivity. Along with the conventional intervention strategies such as medication, behavioral treatments, etc., neurofeedback in BCI games has also been emerging as a promising modality for treating the attention deficit. We will also discuss portable and economical EEG recording devices currently employed in BCI-based brain training modules/games. Finally, the chapter will be concluded with a brief overview of the neurofeedback developments in the context of BCI-based games until now, their potential impact on the healthy as well as on people with neurological disorders, challenges in transferring the successful protocols from laboratories into the market and hurdles in real-time BCI system design and development.