The neural net computer simulations which will be presented here are based on the acceptance of a set of assumptions that for the last twenty years have been expressed in the fields of information processing, neurophysiology and cognitive sciences. First of all, neural networks and their dynamical behaviors in terms of attractors is the natural way adopted by the brain to encode information. Any information item to be stored in the neural net should be coded in some way or another in one of the dynamical attractors of the brain and retrieved by stimulating the net so as to trap its dynamics in the desired item's basin of attraction. The second view shared by neural net researchers is to base the learning of the synaptic matrix on a local Hebbian mechanism. The last assumption is the presence of chaos and the benefit gained by its presence. Chaos, although very simply produced, inherently possesses an infinite amount of cyclic regimes that can be exploited for coding information. Moreover, the network randomly wanders around these unstable regimes in a spontaneous way, thus rapidly proposing alternative responses to external stimuli and being able to easily switch from one of these potential attractors to another in response to any coming stimulus.<p><p>In this thesis, it is shown experimentally that the more information is to be stored in robust cyclic attractors, the more chaos appears as a regime in the back, erratically itinerating among brief appearances of these attractors. Chaos does not appear to be the cause but the consequence of the learning. However, it appears as an helpful consequence that widens the net's encoding capacity. To learn the information to be stored, an unsupervised Hebbian learning algorithm is introduced. By leaving the semantics of the attractors to be associated with the feeding data unprescribed, promising results have been obtained in term of storing capacity. / Doctorat en sciences appliquées / info:eu-repo/semantics/nonPublished

Advances in unsupervised learning and deep neural networks have led to increased performance in a number of domains, and to the ability to draw strong comparisons between the biological method of self-organization conducted by the brain and computational mechanisms. This thesis aims to use real-world data to tackle two areas in the domain of computer vision which have biological equivalents: feature detection and motion tracking. The aforementioned advances have allowed efficient learning of feature representations directly from large sets of unlabeled data instead of using traditional handcrafted features. The first part of this thesis evaluates such representations by comparing regularization and preprocessing methods which incorporate local neighbouring information during training on a single-layer neural network. The networks are trained and tested on the Hollywood2 video dataset, as well as the static CIFAR-10, STL-10, COIL-100, and MNIST image datasets. The induction of topography or simple image blurring via Gaussian filters during training produces better discriminative features as evidenced by the consistent and notable increase in classification results that they produce. In the visual domain, invariant features are desirable such that objects can be classified despite transformations. It is found that most of the compared methods produce more invariant features, however, classification accuracy does not correlate to invariance. The second, and paramount, contribution of this thesis is a biologically-inspired model to explain the emergence of motion tracking behaviour in early development using unsupervised learning. The model’s self-organization is biased by an original concept called retinal constancy, which measures how similar visual contents are between successive frames. In the proposed two-layer deep network, when exposed to real-world video, the first layer learns to encode visual motion, and the second layer learns to relate that motion to gaze movements, which it perceives and creates through bi-directional nodes. This is unique because it uses general machine learning algorithms, and their inherent generative properties, to learn from real-world data. It also implements a biological theory and learns in a fully unsupervised manner. An analysis of its parameters and limitations is conducted, and its tracking performance is evaluated. Results show that this model is able to successfully follow targets in real-world video, despite being trained without supervision on real-world video.

In dieser Arbeit werden zwei wesentliche neue Ergebnisse vorgestellt. Das erste bezieht sich auf die olfaktorische Kodierung und das zweite auf das sensorische Gedaechtnis. Beide Phaenomene werden am Beispiel des Gehirns der Honigbiene untersucht. In Bezug auf die olfaktorische Kodierung zeige ich, dass die neuronale Dynamik waehrend der Stimulation im Antennallobus duftspezifische Trajektorien beschreibt, die in duftspezifischen Attraktoren enden. Das Zeitinterval, in dem diese Attraktoren erreicht werden, betraegt unabhaengig von der Identitaet und der Konzentration des Duftes ungefaehr 800 ms. Darueber hinaus zeige ich, dass Support-Vektor Maschinen, und insbesondere Perzeptronen, ein realistisches und biologisches Model der Wechselwirkung zwischen dem Antennallobus (dem kodierenden Netwerk) und dem Pilzkoerper (dem dekodierenden Netzwerk) darstellen. Dieses Model kann sowohl Reaktionszeiten von ca. 300 ms als auch die Invarianz der Duftwahrnehmung gegenueber der Duftkonzentration erklaeren. In Bezug auf das sensorische Gedaechtnis zeige ich, dass eine einzige Stimulation ohne Belohnung dem Hebbschen Postulat folgend Veraenderungen der paarweisen Korrelationen zwischen Glomeruli induziert. Ich zeige, dass diese Veranderungen der Korrelationen bei 2/3 der Bienen ausreichen, um den letzten Stimulus zu bestimmen. In der zweiten Minute nach der Stimulation ist eine erfolgreiche Bestimmung des Stimulus nur bei 1/3 der Bienen moeglich. Eine Hauptkomponentenanalyse der spontanen Aktivitaet laesst erkennen, dass das dominante Muster des Netzwerks waehrend der spontanen Aktivitaet nach, aber nicht vor der Stimulation das duftinduzierte Aktivitaetsmuster bei 2/3 der Bienen nachbildet. Man kann deshalb die duftinduzierten (Veraenderungen der) Korrelationen als Spuren eines Kurzzeitgedaechtnisses bzw. als Hebbsche "Reverberationen" betrachtet werden. / Two major novel results are reported in this work. The first concerns olfactory coding and the second concerns sensory memory. Both phenomena are investigated in the brain of the honeybee as a model system. Considering olfactory coding I demonstrate that the neural dynamics in the antennal lobe describe odor-specific trajectories during stimulation that converge to odor-specific attractors. The time interval to reach these attractors is, regardless of odor identity and concentration, approximately 800 ms. I show that support-vector machines and, in particular perceptrons provide a realistic and biological model of the interaction between the antennal lobe (coding network) and the mushroom body (decoding network). This model can also account for reaction-times of about 300 ms and for concentration invariance of odor perception. Regarding sensory memory I show that a single stimulation without reward induces changes of pairwise correlation between glomeruli in a Hebbian-like manner. I demonstrate that those changes of correlation suffice to retrieve the last stimulus presented in 2/3 of the bees studied. Succesful retrieval decays to 1/3 of the bees within the second minute after stimulation. In addition, a principal-component analysis of the spontaneous activity reveals that the dominant pattern of the network during the spontaneous activity after, but not before stimulation, reproduces the odor-induced activity pattern in 2/3 of the bees studied. One can therefore consider the odor-induced (changes of) correlation as traces of a short-term memory or as Hebbian reverberations.

A substantial number of works have aimed at modeling the receptive field properties of the primary visual cortex (V1). Their evaluation criterion is usually the similarity of the model response properties to the recorded responses from biological organisms. However, as several algorithms were able to demonstrate some degree of similarity to biological data based on the existing criteria, we focus on the robustness against loss of information in the form of occlusions as an additional constraint for better understanding the algorithmic level of early vision in the brain. We try to investigate the influence of competition mechanisms on the robustness. Therefore, we compared four methods employing different competition mechanisms, namely, independent component analysis, non-negative matrix factorization with sparseness constraint, predictive coding/biased competition, and a Hebbian neural network with lateral inhibitory connections. Each of those methods is known to be capable of developing receptive fields comparable to those of V1 simple-cells. Since measuring the robustness of methods having simple-cell like receptive fields against occlusion is difficult, we measure the robustness using the classification accuracy on the MNIST hand written digit dataset. For this we trained all methods on the training set of the MNIST hand written digits dataset and tested them on a MNIST test set with different levels of occlusions. We observe that methods which employ competitive mechanisms have higher robustness against loss of information. Also the kind of the competition mechanisms plays an important role in robustness. Global feedback inhibition as employed in predictive coding/biased competition has an advantage compared to local lateral inhibition learned by an anti-Hebb rule.