Fabien Lotte

Thibault de Surrel, Fabien Lotte, Sylvain Chevallier et al.

Circular and non-flat data distributions are prevalent across diverse domains of data science, yet their specific geometric structures often remain underutilized in machine learning frameworks. A principled approach to accounting for the underlying geometry of such data is pivotal, particularly when extending statistical models, like the pervasive Gaussian distribution. In this work, we tackle those issue by focusing on the manifold of symmetric positive definite (SPD) matrices, a key focus in information geometry. We introduce a non-isotropic wrapped Gaussian by leveraging the exponential map, we derive theoretical properties of this distribution and propose a maximum likelihood framework for parameter estimation. Furthermore, we reinterpret established classifiers on SPD through a probabilistic lens and introduce new classifiers based on the wrapped Gaussian model. Experiments on synthetic and real-world datasets demonstrate the robustness and flexibility of this geometry-aware distribution, underscoring its potential to advance manifold-based data analysis. This work lays the groundwork for extending classical machine learning and statistical methods to more complex and structured data.

Côme Annicchiarico, Fabien Lotte, Jérémie Mattout

Neurofeedback training (NFT) aims to teach self-regulation of brain activity through real-time feedback, but suffers from highly variable outcomes and poorly understood mechanisms, hampering its validation. To address these issues, we propose a formal computational model of the NFT closed loop. Using Active Inference, a Bayesian framework modelling perception, action, and learning, we simulate agents interacting with an NFT environment. This enables us to test the impact of design choices (e.g., feedback quality, biomarker validity) and subject factors (e.g., prior beliefs) on training. Simulations show that training effectiveness is sensitive to feedback noise or bias, and to prior beliefs (highlighting the importance of guiding instructions), but also reveal that perfect feedback is insufficient to guarantee high performance. This approach provides a tool for assessing and predicting NFT variability, interpret empirical data, and potentially develop personalized training protocols.

Thibault de Surrel, Florian Yger, Fabien Lotte et al.

The goal of this paper is to show how different machine learning tools on the Riemannian manifold $\mathcal{P}_d$ of Symmetric Positive Definite (SPD) matrices can be united under a probabilistic framework. For this, we will need several Gaussian distributions defined on $\mathcal{P}_d$. We will show how popular classifiers on $\mathcal{P}_d$ can be reinterpreted as Bayes Classifiers using these Gaussian distributions. These distributions will also be used for outlier detection and dimension reduction. By showing that those distributions are pervasive in the tools used on $\mathcal{P}_d$, we allow for other machine learning tools to be extended to $\mathcal{P}_d$.

Jelena Mladenović, Jeremy Frey, Smeety Pramij et al.

Objective. Neural self-regulation is necessary for achieving control over brain-computer interfaces (BCIs). This can be an arduous learning process especially for motor imagery BCI. Various training methods were proposed to assist users in accomplishing BCI control and increase performance. Notably the use of biased feedback, i.e. non-realistic representation of performance. Benefits of biased feedback on performance and learning vary between users (e.g. depending on their initial level of BCI control) and remain speculative. To disentangle the speculations, we investigate what personality type, initial state and calibration performance (CP) could benefit from a biased feedback. Methods. We conduct an experiment (n=30 for 2 sessions). The feedback provided to each group (n=10) is either positively, negatively or not biased. Results. Statistical analyses suggest that interactions between bias and: 1) workload, 2) anxiety, and 3) self-control significantly affect online performance. For instance, low initial workload paired with negative bias is associated to higher peak performances (86%) than without any bias (69%). High anxiety relates negatively to performance no matter the bias (60%), while low anxiety matches best with negative bias (76%). For low CP, learning rate (LR) increases with negative bias only short term (LR=2%) as during the second session it severely drops (LR=-1%). Conclusion. We unveil many interactions between said human factors and bias. Additionally, we use prediction models to confirm and reveal even more interactions. Significance. This paper is a first step towards identifying optimal biased feedback for a personality type, state, and CP in order to maximize BCI performance and learning.

Léa Pillette, Camille Jeunet, Roger N'Kambou et al.

Mental Imagery based Brain-Computer Interfaces (MI-BCI) enable their users to control an interface, e.g., a prosthesis, by performing mental imagery tasks only, such as imagining a right arm movement while their brain activity is measured and processed by the system. Designing and using a BCI requires users to learn how to produce different and stable patterns of brain activity for each of the mental imagery tasks. However, current training protocols do not enable every user to acquire the skills required to use BCIs. These training protocols are most likely one of the main reasons why BCIs remain not reliable enough for wider applications outside research laboratories. Learning companions have been shown to improve training in different disciplines, but they have barely been explored for BCIs so far. This article aims at investigating the potential benefits learning companions could bring to BCI training by improving the feedback, i.e., the information provided to the user, which is primordial to the learning process and yet have proven both theoretically and practically inadequate in BCI. This paper first presents the potentials of BCI and the limitations of current training approaches. Then, it reviews both the BCI and learning companion literature regarding three main characteristics of feedback: its appearance, its social and emotional components and its cognitive component. From these considerations, this paper draws some guidelines, identify open challenges and suggests potential solutions to design and use learning companions for BCIs.

Aline Roc, Léa Pillette, B. N'Kaoua et al.

Mental Imagery based Brain-Computer Interfaces (MI-BCI) are a mean to control digital technologies by performing MI tasks alone. Throughout MI-BCI use, human supervision (e.g., experimenter or caregiver) plays a central role. While providing emotional and social feedback, people present BCIs to users and ensure smooth users' progress with BCI use. Though, very little is known about the influence experimenters might have on the results obtained. Such influence is to be expected as social and emotional feedback were shown to influence MI-BCI performances. Furthermore, literature from different fields showed an experimenter effect, and specifically of their gender, on experimental outcome. We assessed the impact of the interaction between experi-menter and participant gender on MI-BCI performances and progress throughout a session. Our results revealed an interaction between participants gender, experimenter gender and progress over runs. It seems to suggest that women experimenters may positively influence partici-pants' progress compared to men experimenters.

Giuseppe Amato, Malte Behrmann, Frédéric Bimbot et al.

Thanks to the Big Data revolution and increasing computing capacities, Artificial Intelligence (AI) has made an impressive revival over the past few years and is now omnipresent in both research and industry. The creative sectors have always been early adopters of AI technologies and this continues to be the case. As a matter of fact, recent technological developments keep pushing the boundaries of intelligent systems in creative applications: the critically acclaimed movie "Sunspring", released in 2016, was entirely written by AI technology, and the first-ever Music Album, called "Hello World", produced using AI has been released this year. Simultaneously, the exploratory nature of the creative process is raising important technical challenges for AI such as the ability for AI-powered techniques to be accurate under limited data resources, as opposed to the conventional "Big Data" approach, or the ability to process, analyse and match data from multiple modalities (text, sound, images, etc.) at the same time. The purpose of this white paper is to understand future technological advances in AI and their growing impact on creative industries. This paper addresses the following questions: Where does AI operate in creative Industries? What is its operative role? How will AI transform creative industries in the next ten years? This white paper aims to provide a realistic perspective of the scope of AI actions in creative industries, proposes a vision of how this technology could contribute to research and development works in such context, and identifies research and development challenges.

Jelena Mladenović, Jérémy Frey, Emmanuel Maby et al.

Adaptive Brain-Computer interfaces (BCIs) have shown to improve performance, however a general and flexible framework to implement adaptive features is still lacking. We appeal to a generic Bayesian approach, called Active Inference (AI), to infer user's intentions or states and act in a way that optimizes performance. In realistic P300-speller simulations, AI outperforms traditional algorithms with an increase in bit rate between 18% and 59%, while offering a possibility of unifying various adaptive implementations within one generic framework.

Jelena Mladenović, Jérémie Mattout, Fabien Lotte

There are numerous possibilities and motivations for an adaptive BCI, which may not be easy to clarify and organize for a newcomer to the field. To our knowledge, there has not been any work done in classifying the literature on adaptive BCI in a comprehensive and structured way. We propose a conceptual framework, a taxonomy of adaptive BCI methods which encompasses most important approaches to fit them in such a way that a reader can clearly visualize which elements are being adapted and for what reason. In the interest of having a clear review of existing adaptive BCIs, this framework considers adaptation approaches for both the user and the machine, i.e., using instructional design observations as well as the usual machine learning techniques. This framework not only provides a coherent review of such extensive literature but also enables the reader to perceive gaps and flaws in the current BCI systems, which would hopefully bring novel solutions for an overall improvement.

Jelena Mladenović, Jérémy Frey, Manon Bonnet-Save et al.

Major issues in Brain Computer Interfaces (BCIs) include low usability and poor user performance. This paper tackles them by ensuring the users to be in a state of immersion, control and motivation, called state of flow. Indeed, in various disciplines, being in the state of flow was shown to improve performances and learning. Hence, we intended to draw BCI users in a flow state to improve both their subjective experience and their performances. In a Motor Imagery BCI game, we manipulated flow in two ways: 1) by adapting the task difficulty and 2) by using background music. Results showed that the difficulty adaptation induced a higher flow state, however music had no effect. There was a positive correlation between subjective flow scores and offline performance, although the flow factors had no effect (adaptation) or negative effect (music) on online performance. Overall, favouring the flow state seems a promising approach for enhancing users' satisfaction, although its complexity requires more thorough investigations.

Jérémy Frey, Renaud Gervais, Thibault Lainé et al.

Teegi is an anthropomorphic and tangible avatar exposing a users' brain activity in real time. It is connected to a device sensing the brain by means of electroencephalog-raphy (EEG). Teegi moves its hands and feet and closes its eyes along with the person being monitored. It also displays on its scalp the associated EEG signals, thanks to a semi-spherical display made of LEDs. Attendees can interact directly with Teegi -- e.g. move its limbs -- to discover by themselves the underlying brain processes. Teegi can be used for scientific outreach to introduce neurotechnologies in general and brain-computer interfaces (BCI) in particular.

Jérémy Frey, Maxime Daniel, Julien Castet et al.

Measuring brain activity with electroencephalography (EEG) is mature enough to assess mental states. Combined with existing methods, such tool can be used to strengthen the understanding of user experience. We contribute a set of methods to estimate continuously the user's mental workload, attention and recognition of interaction errors during different interaction tasks. We validate these measures on a controlled virtual environment and show how they can be used to compare different interaction techniques or devices, by comparing here a keyboard and a touch-based interface. Thanks to such a framework, EEG becomes a promising method to improve the overall usability of complex computer systems.

Renaud Gervais, Jérémy Frey, Alexis Gay et al.

We propose a toolkit for creating Tangible Out-of-Body Experiences: exposing the inner states of users using physiological signals such as heart rate or brain activity. Tobe can take the form of a tangible avatar displaying live physiological readings to reflect on ourselves and others. Such a toolkit could be used by researchers and designers to create a multitude of potential tangible applications, including (but not limited to) educational tools about Science Technologies Engineering and Mathematics (STEM) and cognitive science, medical applications or entertainment and social experiences with one or several users or Tobes involved. Through a co-design approach, we investigated how everyday people picture their physiology and we validated the acceptability of Tobe in a scientific museum. We also give a practical example where two users relax together, with insights on how Tobe helped them to synchronize their signals and share a moment.

Dennis Wobrock, Jérémy Frey, Delphine Graeff et al.

Designing 3D User Interfaces (UI) requires adequate evaluation tools to ensure good usability and user experience. While many evaluation tools are already available and widely used, existing approaches generally cannot provide continuous and objective measures of usa-bility qualities during interaction without interrupting the user. In this paper, we propose to use brain (with ElectroEncephaloGraphy) and physiological (ElectroCardioGraphy, Galvanic Skin Response) signals to continuously assess the mental effort made by the user to perform 3D object manipulation tasks. We first show how this mental effort (a.k.a., mental workload) can be estimated from such signals, and then measure it on 8 participants during an actual 3D object manipulation task with an input device known as the CubTile. Our results suggest that monitoring workload enables us to continuously assess the 3DUI and/or interaction technique ease-of-use. Overall, this suggests that this new measure could become a useful addition to the repertoire of available evaluation tools, enabling a finer grain assessment of the ergonomic qualities of a given 3D user interface.

Jérémy Frey, Aurélien Appriou, Fabien Lotte et al.

With stereoscopic displays, a depth sensation that is too strong could impede visual comfort and result in fatigue or pain. Electroencephalography (EEG) is a technology which records brain activity. We used it to develop a novel brain-computer interface that monitors users' states in order to reduce visual strain. We present the first proof-of-concept system that discriminates comfortable conditions from uncomfortable ones during stereoscopic vision using EEG. It reacts within 1s to depth variations, achieving 63% accuracy on average and 74% when 7 consecutive variations are measured. This study could lead to adaptive systems that automatically suit stereoscopic displays to users and viewing conditions.

Jérémy Frey, Renaud Gervais, Stéphanie Fleck et al.

We introduce Teegi, a Tangible ElectroEncephaloGraphy (EEG) Interface that enables novice users to get to know more about something as complex as brain signals, in an easy, en- gaging and informative way. To this end, we have designed a new system based on a unique combination of spatial aug- mented reality, tangible interaction and real-time neurotech- nologies. With Teegi, a user can visualize and analyze his or her own brain activity in real-time, on a tangible character that can be easily manipulated, and with which it is possible to interact. An exploration study has shown that interacting with Teegi seems to be easy, motivating, reliable and infor- mative. Overall, this suggests that Teegi is a promising and relevant training and mediation tool for the general public.

Jérémy Frey, Léonard Pommereau, Fabien Lotte et al.

The conflict between vergence (eye movement) and accommodation (crystalline lens deformation) occurs in every stereoscopic display. It could cause important stress outside the "zone of comfort", when stereoscopic effect is too strong. This conflict has already been studied using questionnaires, during viewing sessions of several minutes. The present pilot study describes an experimental protocol which compares two different comfort conditions using electroencephalography (EEG) over short viewing sequences. Analyses showed significant differences both in event-related potentials (ERP) and in frequency bands power. An uncomfortable stereoscopy correlates with a weaker negative component and a delayed positive component in ERP. It also induces a power decrease in the alpha band and increases in theta and beta bands. With fast responses to stimuli, EEG is likely to enable the conception of adaptive systems, which could tune the stereoscopic experience according to each viewer.

Jérémy Frey, Christian Mühl, Fabien Lotte et al.

Evaluating human-computer interaction is essential as a broadening population uses machines, sometimes in sensitive contexts. However, traditional evaluation methods may fail to combine real-time measures, an "objective" approach and data contextualization. In this review we look at how adding neuroimaging techniques can respond to such needs. We focus on electroencephalography (EEG), as it could be handled effectively during a dedicated evaluation phase. We identify workload, attention, vigilance, fatigue, error recognition, emotions, engagement, flow and immersion as being recognizable by EEG. We find that workload, attention and emotions assessments would benefit the most from EEG. Moreover, we advocate to study further error recognition through neuroimaging to enhance usability and increase user experience.