Associative learning is a fundamental ability biological systems possess in order to adapt to a nonstationary environment. One of the core aspects of associative learning theoretical frameworks is that surprising events drive learning by signalling the need to update the system’s beliefs about the probability structure governing stimuli associations. Specifically, the central neural system generates internal predictions to anticipate the causes of its perceptual experience and compute a prediction error to update its generative model of the environment, an idea generally known as the predictive coding framework. However, it is not clear whether the brain generates these predictions only for goal-oriented behavior or they are more a general characteristic of the brain function. In this thesis, I explored the role of task relevance in modulating brain activity when exposed to sensory associative learning task. In the first study, participants were asked to perform a perceptual detection task while audio-visual stimuli were presented as distractors. These distractors possessed a probability structure that made some of them more paired than others. Results showed that occipital activity triggered by the conditioned stimulus was elicited just before the arrival of the unconditioned visual stimulus. Moreover, occipital activity after the onset of the unconditioned stimulus followed a pattern of precision-weighted prediction errors. In the second study, two more sessions were added to the task in the previous study in which the probability structure for all stimuli associations was identical and the whole experiment was spanned in six days across two weeks. Results showed a difference in the modulation of the beta band induced by the presentation of the unconditioned stimulus preceded by the predictive and unpredictive conditioned auditory stimuli by comparing the pre and post sessions activity. In the third study, participants were exposed to a similar task with respect to the second study with the modification that there was a condition in which the conditioned-unconditioned stimulus association was task-relevant, thus allowing to directly compare task-relevant and task-irrelevant associations. Results showed that both types of associations had similar patterns in terms of activity and functional connectivity when comparing the brain responses to the onset of the unconditioned visual stimulus. Taken together, these findings demonstrate irrelevant associations rely on the same neural mechanisms of relevant ones. Thus, even if task relevance plays a modulatory role on the strength of the neural effects of associative learning, predictive processes take place in sensory associative learning regardless of task relevance.

<p>Video coding schemes designed based on sequential or predictive coding models are vulnerable to the loss of encoded frames at the decoder end. Motivated by this observation, in this thesis we propose two new coding models: robust sequential coding and robust predictive coding. For the Gauss-Markov source with the mean squared error distortion measure, we characterize certain supporting hyperplanes of the rate region of these two coding problems. The proof is divided into three steps: 1) it is shown that each supporting hyperplane of the rate region of Gaussian robust sequential coding admits a max-min lower bound; 2) the corresponding min-max upper bound is shown to be achievable by a robust predictive coding scheme; 3) a saddle point analysis proves that the max-min lower bound coincides with the min-max upper bound. Furthermore, it is shown that the proposed robust predictive coding scheme can be implemented using a successive quantization system. Theoretical and experimental results indicate that this scheme has a desirable \self-recovery" property. Our investigation also reveals an information-theoretic minimax theorem and the associated extremal inequalities.</p> / Doctor of Philosophy (PhD)

Στα πλαίσια της διπλωματικής εργασίας αναπτύχθηκε μία εφαρμογή, η οποία προσδιορίζει τις τρεις πρώτες συχνότητες συντονισμού της φωνητικής κοιλότητας κατά τη διαδικασία της φώνησης φωνηέντων. Οι τρεις αυτές συχνότητες παρέχουν επαρκή πληροφορία για τον προσδιορισμό του φωνήεντου. Η φώνηση εξομοιώνεται με σήμα εισόδου το οποίο παρουσιάζει κορυφές σε αναμενόμενες περιοχές συχνοτήτων. Ο προσδιορισμός των συχνοτήτων συντονισμού στηρίζεται στη μέθοδο βραχύχρονης ανάλυσης Fourier. Η εφαρμογή αναπτύχθηκε σε περιβάλλον μVision της Keil, σε γλώσσα προγραμματισμού C, για τον μικροελεγκτή STM32F103RB της ST Microelectronics. / In the context of this thesis an application was developed, that is capable of estimating the first three formant frequencies (resonances of the vocal tract) in the event of voicing of vowels. These three frequencies provide us enough information to determine the vowel that is voiced. The human voice is being emulated by an input signal which has peaks in the anticipated frequency regions. The formant frequencies are being estimated based on the short-time Fourier analysis method. The application was developed in Keil μVision programming suite, in C programming language, for the STM32F103RB microcontroller by ST Microelectronics.

Είναι γεγονός ότι το ηλεκτροκαρδιογράφημα είναι ένα πολύ καλά μελετημένο σήμα. Ειδικά τα τελευταία χρόνια, έχει προταθεί ένας μεγάλος αριθμός αλγορίθμων επεξεργασίας, συμπίεσης, αυτόματης διάγνωσης, φιλτραρίσματος, αποθορυβοποίησης και κωδικοποίησης. Σ’ αυτή τη διπλωματική εργασία, προτείνουμε ένα αποδοτικό αλγόριθμο κωδικοποίησης χωρίς απώλειες για δεδομένα από δωδεκακάναλλο ηλεκτροκαρδιογράφημα. Ο κωδικοποιητής υλοποιεί ένα πολυγραμμικό μοντέλο υψηλής απόδοσης, το οποίο είναι «ειδικευμένο στους ασθενείς», ενδοκαναλικής πρόβλεψη και εφαρμόζει το πρότυπο κωδικοποίησης MPEG-4 ALS για διακαναλική πρόβλεψη και κωδικοποίηση. Τα αποτελέσματα του αλγορίθμου συγκρίθηκαν με τεχνικές κωδικοποίησης εντροπίας χωρίς απώλειες και δείχνουν αύξηση της απόδοσης κωδικοποίησης. / The Electrocardiogram (ECG) is one of the most well studied medical signals. A large number of ECG processing algorithms have being proposed over the years covering the areas of ECG noise filtering, automated diagnostic interpretation and coding. In this master thesis, we propose a robust multi-channel ECG encoder architecture, which operates on 12-channel ECG data. The encoder utilizes highly efficient multi-linear patient specific models for inter-channel prediction and the MPEG-4 Audio Lossless Coding (ALS) architecture for intra-channel prediction and coding. The results of the algorithm show improved performance over standard encoding techniques.

Táto práca sa zaoberá rozpoznávaním výstrelov a pridruženými problémami. Ako prvé je celá vec predstavená a rozdelená na menšie kroky. Ďalej je poskytnutý prehľad zvukových databáz, významné publikácie, akcie a súčasný stav veci spoločne s prehľadom možných aplikácií detekcie výstrelov. Druhá časť pozostáva z porovnávania príznakov pomocou rôznych metrík spoločne s porovnaním ich výkonu pri rozpoznávaní. Nasleduje porovnanie algoritmov rozpoznávania a sú uvedené nové príznaky použiteľné pri rozpoznávaní. Práca vrcholí návrhom dvojstupňového systému na rozpoznávanie výstrelov, monitorujúceho okolie v reálnom čase. V závere sú zhrnuté dosiahnuté výsledky a načrtnutý ďalší postup.

Le comportement moteur est un phénomène où les différentes composantes d’un système biologique sont organisées de façon à assurer la coordination d’un mouvement intentionnel. Selon les théories computationnelles, le comportement est défini comme un problème moteur dont la solution peut être trouvée par des systèmes divisés de manière hiérarchique. Les composantes traitent et communiquent entre elles de l’information représentant les aspects pertinents du problème moteur (positions, trajectoires, vitesses, forces, etc.) lesquels sont censés être organisés à leur tour selon une hiérarchie d’abstraction et de complexité ascendante. Le défi est de faire face à quatre problèmes centraux du comportement : a) Le nombre élevé de degrés de liberté et d’interactions ; b) La redondance des degrés de liberté ; c) L’anticipation des effets du mouvement ; d) L’incertitude dans l’information. Les théories computationnelles classiques proposent des schémas explicatifs composés d’un agencement de différents modèles internes (prospectifs et inverses). Plus récemment, l’approche bayésienne propose un schéma hiérarchique plus homogène lequel est censé faire face aussi à l’incertitude de l’information. Cette recherche démontre que les théories computationnelles, y compris l’approche bayésienne, sont paralysées par un dilemme insurmontable : soit elles peuvent passer à l’échelle de manière computationnellement faisable - les calculs peuvent être réalisés en un temps raisonnable - mais dans ce cas elles ne peuvent pas reproduire la versatilité caractéristique du comportement des êtres vivants ; soit elles aspirent à reproduire la versatilité biologique mais alors elles sont infaisables. / Motor behavior is a phenomenon where the components making up a biological system are organized so as to ensure the coordination of a purposeful movement. According to computational theories, behavior is defined as a motor problem the solution of which can be found by systems divided hierarchically. The components process and communicate information representing the relevant variables of the motor problem (positions, trajectories, velocities, forces, etc.) which are, in turn, assumed to be organized as a hierarchy of increasing abstraction and complexity. The challenge is to tackle the four core problems of behavior: a) The high number of degrees of freedom and their interactions; b) The redundancy of degrees of freedom; c) The anticipation of the effects of movement; d) The uncertainty in information. Classical computational theories advance explanatory schemas made of structured sets of internal models (forward and inverse). More recently, the Bayesian approach advances a more homogeneous hierarchical schema which is supposed to account for uncertainty in information. This research shows that computational theories, including the Bayesian approach, are crippled by an unsolvable dilemma: The first horn is that if the models can scale up while staying computationally tractable, i.e. the computations can be carried out in a reasonable amount of time, then they fail to reproduce the versatility which characterizes the behavior of living beings. The second horn is that if the models aspire to reproduce biological versatility, then they are intractable.

In the “rubber hand illusion,” the participant sees a displaced fake hand being touched congruently with her unseen real hand. This seems to invoke inference of an “illusory” common cause for visual, tactile, and proprioceptive sensations; as evident from a perceived embodiment of the fake hand and the perception of one’s unseen hand location closer toward the position of the fake hand—the so-named “proprioceptive drift.” Curiously, participants may sometimes move their hand in the direction of the fake hand (Asai, 2015). While this could easily be explained as participants actively trying to align the real and fake hands to experience a stronger illusion, they are not aware of these movements (cf. Abdulkarim and Ehrsson, 2018). So there may be better explanation for this observation than that participants were “cheating.” In their recent article, Lanillos et al. (2021) show that the unintentional execution of arm movement forces during a virtual reality based version of the rubber hand illusion—which the authors call “active drift”—can be reproduced by a computational model based on the active inference framework.

Expectations substantially influence perception, but the neural mechanisms underlying this influence are not fully understood. A prominent view is that sensory neurons encode prediction error with respect to expectations on upcoming sensory input. Although the encoding of prediction error has been previously demonstrated in the human auditory cortex (AC), previous studies often induced expectations using stimulus repetition, potentially confounding prediction error with neural habituation. These studies also measured AC as a single population, failing to consider possible predictive specializations of different AC fields. Moreover, the few studies that considered prediction error to stimuli other than pure tones yielded conflicting results. Here, we used functional magnetic resonance imaging (fMRI) to systematically investigate prediction error to subjective expectations in auditory cortical fields Te1.0, Te1.1, Te1.2, and Te3, and two types of stimuli: pure tones and frequency modulated (FM) sweeps. Our results show that prediction error is elicited with respect to the participants’ expectations independently of stimulus repetition and similarly expressed across auditory fields. Moreover, despite the radically different strategies underlying the decoding of pure tones and FM-sweeps, both stimulus modalities were encoded as prediction error in most fields of AC. Altogether, our results provide unequivocal evidence that predictive coding is the general encoding mechanism in AC.

Cotutelle entre l’université de Montréal et d’Aix-Marseille / Cette thèse vise à comprendre les fondements et les fonctions des calculs probabilistes impliqués dans les processus visuels. Nous nous appuyons sur une double stratégie, qui implique le développement de modèles dans le cadre du codage prédictif selon le principe de l'énergie libre. Ces modèles servent à définir des hypothèses claires sur la fonction neuronale, qui sont testées à l'aide d'enregistrements extracellulaires du cortex visuel primaire. Cette région du cerveau est principalement impliquée dans les calculs sur les unités élémentaires des entrées visuelles naturelles, sous la forme de distributions d'orientations. Ces distributions probabilistes, par nature, reposent sur le traitement de la moyenne et de la variance d'une entrée visuelle. Alors que les premières ont fait l'objet d'un examen neurobiologique approfondi, les secondes ont été largement négligées. Cette thèse vise à combler cette lacune. Nous avançons l'idée que la connectivité récurrente intracorticale est parfaitement adaptée au traitement d'une telle variance d'entrées, et nos contributions à cette idée sont multiples. (1) Nous fournissons tout d'abord un examen informatique de la structure d'orientation des images naturelles et des stratégies d'encodage neuronal associées. Un modèle empirique clairsemé montre que le code neuronal optimal pour représenter les images naturelles s'appuie sur la variance de l'orientation pour améliorer l'efficacité, la performance et la résilience. (2) Cela ouvre la voie à une étude expérimentale des réponses neurales dans le cortex visuel primaire du chat à des stimuli multivariés. Nous découvrons de nouveaux types de neurones fonctionnels, dépendants de la couche corticale, qui peuvent être liés à la connectivité récurrente. (3) Nous démontrons que ce traitement de la variance peut être compris comme un graphe dynamique pondéré conditionné par la variance sensorielle, en utilisant des enregistrements du cortex visuel primaire du macaque. (4) Enfin, nous soutenons l'existence de calculs de variance (prédictifs) en dehors du cortex visuel primaire, par l'intermédiaire du noyau pulvinar du thalamus. Cela ouvre la voie à des études sur les calculs de variance en tant que calculs neuronaux génériques soutenus par la récurrence dans l'ensemble du cortex. / This thesis aims to understand the foundations and functions of the probabilistic computations involved in visual processes. We leverage a two-fold strategy, which involves the development of models within the framework of predictive coding under the free energy principle. These models serve to define clear hypotheses of neuronal function, which are tested using extracellular recordings of the primary visual cortex. This brain region is predominantly involved in computations on the elementary units of natural visual inputs, in the form of distributions of oriented edges. These probabilistic distributions, by nature, rely on processing both the mean and variance of a visual input. While the former have undergone extensive neurobiological scrutiny, the latter have been largely overlooked. This thesis aims to bridge this knowledge gap. We put forward the notion that intracortical recurrent connectivity is optimally suited for processing such variance of inputs, and our contributions to this idea are multi-faceted. (1) We first provide a computational examination of the orientation structure of natural images and associated neural encoding strategies. An empirical sparse model shows that the optimal neural code for representing natural images relies on orientation variance for increased efficiency, performance, and resilience. (2) This paves the way for an experimental investigation of neural responses in the cat's primary visual cortex to multivariate stimuli. We uncover novel, cortical-layer-dependent, functional neuronal types that can be linked to recurrent connectivity. (3) We demonstrate that this variance processing can be understood as a dynamical weighted graph conditioned on sensory variance, using macaque primary visual cortex recordings. (4) Finally, we argue for the existence of (predictive) variance computations outside the primary visual cortex, through the Pulvinar nucleus of the thalamus. This paves the way for studies on variance computations as generic weighting of neural computations, supported by recurrence throughout the entire cortex.

<p>Sectra Communications AB has developed a terminal for encrypted communication called the Tiger platform. During voice communication delays have sometimes been experienced resulting in conversational complications.</p><p>A solution to this problem, as was proposed by Sectra, would be to introduce voice activity detection, which means a separation of speech parts and non-speech parts of the input signal, to the Tiger platform. By only transferring the speech parts to the receiver, the bandwidth needed should be dramatically decreased. A lower bandwidth needed implies that the delays slowly should disappear. The problem is then to come up with a method that manages to distinguish the speech parts from the input signal. Fortunately a lot of theory on the subject has been done and numerous voice activity methods exist today.</p><p>In this thesis the theory of voice activity detection has been studied. A review of voice activity detectors that exist on the market today followed by an evaluation of some of these was performed in order to select a suitable candidate for the Tiger platform. This evaluation would later become the foundation for the selection of a voice activity detector for implementation.</p><p>Finally, the implementation of the chosen voice activity detector, including a comfort noise generator, was done on the platform. This implementation was based on the special requirements of the platform. Tests of the implementation in office environments show that possible delays are steadily being reduced during periods of speech inactivity, while the active speech quality is preserved.</p>