Generative models of the human connectome.

The human connectome represents a network map of the brain's wiring diagram and the pattern into which its connections are organized is thought to play an important role in cognitive function. The generative rules that shape the topology of the human connectome remain incompletely understood. Earlier work in model organisms has suggested that wiring rules based on geometric relationships (distance) can account for many but likely not all topological features. Here we systematically explore a family of generative models of the human connectome that yield synthetic networks designed according to different wiring rules combining geometric and a broad range of topological factors. We find that a combination of geometric constraints with a homophilic attachment mechanism can create synthetic networks that closely match many topological characteristics of individual human connectomes, including features that were not included in the optimization of the generative model itself. We use these models to investigate a lifespan dataset and show that, with age, the model parameters undergo progressive changes, suggesting a rebalancing of the generative factors underlying the connectome across the lifespan.

Deep generative models for facial keypoints detection

A new area of machine learning research called deep learning, has moved machine learning closer to one of its original goals: artificial intelligence and general learning algorithm. The key idea is to pretrain models in completely unsupervised way and finally they can be fine-tuned for the task at hand using supervised learning. In this thesis, a general introduction to deep learning models and algorithms are given and these methods are applied to facial keypoints detection. The task is to predict the positions of 15 keypoints on grayscale face images. Each predicted keypoint is specified by an (x,y) real-valued pair in the space of pixel indices. In experiments, we pretrained deep belief networks (DBN) and finally performed a discriminative fine-tuning. We varied the depth and size of an architecture. We tested both deterministic and sampled hidden activations and the effect of additional unlabeled data on pretraining. The experimental results show that our model provides better results than publicly available benchmarks for the dataset.

Using generative models for handwritten digit recognition

We describe a method of recognizing handwritten digits by fitting generative models that are built from deformable B-splines with Gaussian ``ink generators'' spaced along the length of the spline. The splines are adjusted using a novel elastic matching procedure based on the Expectation Maximization (EM) algorithm that maximizes the likelihood of the model generating the data. This approach has many advantages. (1) After identifying the model most likely to have generated the data, the system not only produces a classification of the digit but also a rich description of the instantiation parameters which can yield information such as the writing style. (2) During the process of explaining the image, generative models can perform recognition driven segmentation. (3) The method involves a relatively small number of parameters and hence training is relatively easy and fast. (4) Unlike many other recognition schemes it does not rely on some form of pre-normalization of input images, but can handle arbitrary scalings, translations and a limited degree of image rotation. We have demonstrated our method of fitting models to images does not get trapped in poor local minima. The main disadvantage of the method is it requires much more computation than more standard OCR techniques.

On the generative equations of fractal self-similarity in granular media and the related PSD models

The physical appearance of granular media suggests the existence of geometrical scale invariance. The paper discuss how this physico-empirical property can be mathematically encoded leading to different generative models: a smooth one encoded by a differential equation and another encoded by an equation coming from a measure theoretical property.

Modeling associations between genetic markers using Bayesian networks

Motivation: Understanding the patterns of association between polymorphisms at different loci in a population ( linkage disequilibrium, LD) is of fundamental importance in various genetic studies. Many coefficients were proposed for measuring the degree of LD, but they provide only a static view of the current LD structure. Generative models (GMs) were proposed to go beyond these measures, giving not only a description of the actual LD structure but also a tool to help understanding the process that generated such structure. GMs based in coalescent theory have been the most appealing because they link LD to evolutionary factors. Nevertheless, the inference and parameter estimation of such models is still computationally challenging. Results: We present a more practical method to build GM that describe LD. The method is based on learning weighted Bayesian network structures from haplotype data, extracting equivalence structure classes and using them to model LD. The results obtained in public data from the HapMap database showed that the method is a promising tool for modeling LD. The associations represented by the learned models are correlated with the traditional measure of LD D`. The method was able to represent LD blocks found by standard tools. The granularity of the association blocks and the readability of the models can be controlled in the method. The results suggest that the causality information gained by our method can be useful to tell about the conservability of the genetic markers and to guide the selection of subset of representative markers.

Training deep convolutional architectures for vision

Les tâches de vision artificielle telles que la reconnaissance d’objets demeurent irrésolues à ce jour. Les algorithmes d’apprentissage tels que les Réseaux de Neurones Artificiels (RNA), représentent une approche prometteuse permettant d’apprendre des caractéristiques utiles pour ces tâches. Ce processus d’optimisation est néanmoins difficile. Les réseaux profonds à base de Machine de Boltzmann Restreintes (RBM) ont récemment été proposés afin de guider l’extraction de représentations intermédiaires, grâce à un algorithme d’apprentissage non-supervisé. Ce mémoire présente, par l’entremise de trois articles, des contributions à ce domaine de recherche. Le premier article traite de la RBM convolutionelle. L’usage de champs réceptifs locaux ainsi que le regroupement d’unités cachées en couches partageant les même paramètres, réduit considérablement le nombre de paramètres à apprendre et engendre des détecteurs de caractéristiques locaux et équivariant aux translations. Ceci mène à des modèles ayant une meilleure vraisemblance, comparativement aux RBMs entraînées sur des segments d’images. Le deuxième article est motivé par des découvertes récentes en neurosciences. Il analyse l’impact d’unités quadratiques sur des tâches de classification visuelles, ainsi que celui d’une nouvelle fonction d’activation. Nous observons que les RNAs à base d’unités quadratiques utilisant la fonction softsign, donnent de meilleures performances de généralisation. Le dernière article quand à lui, offre une vision critique des algorithmes populaires d’entraînement de RBMs. Nous montrons que l’algorithme de Divergence Contrastive (CD) et la CD Persistente ne sont pas robustes : tous deux nécessitent une surface d’énergie relativement plate afin que leur chaîne négative puisse mixer. La PCD à "poids rapides" contourne ce problème en perturbant légèrement le modèle, cependant, ceci génère des échantillons bruités. L’usage de chaînes tempérées dans la phase négative est une façon robuste d’adresser ces problèmes et mène à de meilleurs modèles génératifs.

Génération et reconnaissance de rythmes au moyen de réseaux de neurones à réservoir

Les fichiers sons qui accompagne mon document sont au format midi. Le programme que nous avons développés pour ce travail est en language Python.

Algorithmes d’apprentissage profonds supervisés et non-supervisés: applications et résultats théoriques

La liste des domaines touchés par l’apprentissage machine s’allonge rapidement. Au fur et à mesure que la quantité de données disponibles augmente, le développement d’algorithmes d’apprentissage de plus en plus puissants est crucial. Ce mémoire est constitué de trois parties: d’abord un survol des concepts de bases de l’apprentissage automatique et les détails nécessaires pour l’entraînement de réseaux de neurones, modèles qui se livrent bien à des architectures profondes. Ensuite, le premier article présente une application de l’apprentissage machine aux jeux vidéos, puis une méthode de mesure performance pour ceux-ci en tant que politique de décision. Finalement, le deuxième article présente des résultats théoriques concernant l’entraînement d’architectures profondes nonsupervisées. Les jeux vidéos sont un domaine particulièrement fertile pour l’apprentissage automatique: il estf facile d’accumuler d’importantes quantités de données, et les applications ne manquent pas. La formation d’équipes selon un critère donné est une tˆache commune pour les jeux en lignes. Le premier article compare différents algorithmes d’apprentissage à des réseaux de neurones profonds appliqués à la prédiction de la balance d’un match. Ensuite nous présentons une méthode par simulation pour évaluer les modèles ainsi obtenus utilisés dans le cadre d’une politique de décision en ligne. Dans un deuxième temps nous présentons une nouvelleméthode pour entraîner des modèles génératifs. Des résultats théoriques nous indiquent qu’il est possible d’entraîner par rétropropagation des modèles non-supervisés pouvant générer des échantillons qui suivent la distribution des données. Ceci est un résultat pertinent dans le cadre de la récente littérature scientifique investiguant les propriétés des autoencodeurs comme modèles génératifs. Ces résultats sont supportés avec des expériences qualitatives préliminaires ainsi que quelques résultats quantitatifs.

Hierarchical Prediction Machines and Big Data Analytics

An emerging consensus in cognitive science views the biological brain as a hierarchically-organized predictive processing system. This is a system in which higher-order regions are continuously attempting to predict the activity of lower-order regions at a variety of (increasingly abstract) spatial and temporal scales. The brain is thus revealed as a hierarchical prediction machine that is constantly engaged in the effort to predict the flow of information originating from the sensory surfaces. Such a view seems to afford a great deal of explanatory leverage when it comes to a broad swathe of seemingly disparate psychological phenomena (e.g., learning, memory, perception, action, emotion, planning, reason, imagination, and conscious experience). In the most positive case, the predictive processing story seems to provide our first glimpse at what a unified (computationally-tractable and neurobiological plausible) account of human psychology might look like. This obviously marks out one reason why such models should be the focus of current empirical and theoretical attention. Another reason, however, is rooted in the potential of such models to advance the current state-of-the-art in machine intelligence and machine learning. Interestingly, the vision of the brain as a hierarchical prediction machine is one that establishes contact with work that goes under the heading of 'deep learning'. Deep learning systems thus often attempt to make use of predictive processing schemes and (increasingly abstract) generative models as a means of supporting the analysis of large data sets. But are such computational systems sufficient (by themselves) to provide a route to general human-level analytic capabilities? I will argue that they are not and that closer attention to a broader range of forces and factors (many of which are not confined to the neural realm) may be required to understand what it is that gives human cognition its distinctive (and largely unique) flavour. The vision that emerges is one of 'homomimetic deep learning systems', systems that situate a hierarchically-organized predictive processing core within a larger nexus of developmental, behavioural, symbolic, technological and social influences. Relative to that vision, I suggest that we should see the Web as a form of 'cognitive ecology', one that is as much involved with the transformation of machine intelligence as it is with the progressive reshaping of our own cognitive capabilities.

Computer generation and quantitative morphometric analysis of virtual neurons

An important goal in computational neuroanatomy is the complete and accurate simulation of neuronal morphology. We are developing computational tools to model three-dimensional dendritic structures based on sets of stochastic rules. This paper reports an extensive, quantitative anatomical characterization of simulated motoneurons and Purkinje cells. We used several local and global algorithms implemented in the L-Neuron and ArborVitae programs to generate sets of virtual neurons. Parameters statistics for all algorithms were measured from experimental data, thus providing a compact and consistent description of these morphological classes. We compared the emergent anatomical features of each group of virtual neurons with those of the experimental database in order to gain insights on the plausibility of the model assumptions, potential improvements to the algorithms, and non-trivial relations among morphological parameters. Algorithms mainly based on local constraints (e.g., branch diameter) were successful in reproducing many morphological properties of both motoneurons and Purkinje cells (e.g. total length, asymmetry, number of bifurcations). The addition of global constraints (e.g., trophic factors) improved the angle-dependent emergent characteristics (average Euclidean distance from the soma to the dendritic terminations, dendritic spread). Virtual neurons systematically displayed greater anatomical variability than real cells, suggesting the need for additional constraints in the models. For several emergent anatomical properties, a specific algorithm reproduced the experimental statistics better than the others did. However, relative performances were often reversed for different anatomical properties and/or morphological classes. Thus, combining the strengths of alternative generative models could lead to comprehensive algorithms for the complete and accurate simulation of dendritic morphology.

Emotion and anticipation in an enactive framework for cognition (response to Andy Clark)

Undeniably, anticipation plays a crucial role in cognition. By what means, to what extent, and what it achieves remain open questions. In a recent BBS target article, Clark (in press) depicts an integrative model of the brain that builds on hierarchical Bayesian models of neural processing (Rao and Ballard, 1999; Friston, 2005; Brown et al., 2011), and their most recent formulation using the free-energy principle borrowed from thermodynamics (Feldman and Friston, 2010; Friston, 2010; Friston et al., 2010). Hierarchical generative models of cognition, such as those described by Clark, presuppose the manipulation of representations and internal models of the world, in as much detail as is perceptually available. Perhaps surprisingly, Clark acknowledges the existence of a “virtual version of the sensory data” (p. 4), but with no reference to some of the historical debates that shaped cognitive science, related to the storage, manipulation, and retrieval of representations in a cognitive system (Shanahan, 1997), or accounting for the emergence of intentionality within such a system (Searle, 1980; Preston and Bishop, 2002). Instead of demonstrating how this Bayesian framework responds to these foundational questions, Clark describes the structure and the functional properties of an action-oriented, multi-level system that is meant to combine perception, learning, and experience (Niedenthal, 2007).

Sviluppo di un approccio congiunto fuzzy-Bayesiano per l'analisi e la modellizzazione degli ecosistemi: applicazione ad ecosistemi marini costieri.

Nell’attuale contesto di aumento degli impatti antropici e di “Global Climate Change” emerge la necessità di comprenderne i possibili effetti di questi sugli ecosistemi inquadrati come fruitori di servizi e funzioni imprescindibili sui quali si basano intere tessiture economiche e sociali. Lo studio previsionale degli ecosistemi si scontra con l’elevata complessità di questi ultimi in luogo di una altrettanto elevata scarsità di osservazioni integrate. L’approccio modellistico appare il più adatto all’analisi delle dinamiche complesse degli ecosistemi ed alla contestualizzazione complessa di risultati sperimentali ed osservazioni empiriche. L’approccio riduzionista-deterministico solitamente utilizzato nell’implementazione di modelli non si è però sin qui dimostrato in grado di raggiungere i livelli di complessità più elevati all’interno della struttura eco sistemica. La componente che meglio descrive la complessità ecosistemica è quella biotica in virtù dell’elevata dipendenza dalle altre componenti e dalle loro interazioni. In questo lavoro di tesi viene proposto un approccio modellistico stocastico basato sull’utilizzo di un compilatore naive Bayes operante in ambiente fuzzy. L’utilizzo congiunto di logica fuzzy e approccio naive Bayes è utile al processa mento del livello di complessità e conseguentemente incertezza insito negli ecosistemi. I modelli generativi ottenuti, chiamati Fuzzy Bayesian Ecological Model(FBEM) appaiono in grado di modellizare gli stati eco sistemici in funzione dell’ elevato numero di interazioni che entrano in gioco nella determinazione degli stati degli ecosistemi. Modelli FBEM sono stati utilizzati per comprendere il rischio ambientale per habitat intertidale di spiagge sabbiose in caso di eventi di flooding costiero previsti nell’arco di tempo 2010-2100. L’applicazione è stata effettuata all’interno del progetto EU “Theseus” per il quale i modelli FBEM sono stati utilizzati anche per una simulazione a lungo termine e per il calcolo dei tipping point specifici dell’habitat secondo eventi di flooding di diversa intensità.

The Neuroscience of Visual Hallucinations

Each year, some two million people in the United Kingdom experience visual hallucinations. Infrequent, fleeting visual hallucinations, often around sleep, are a usual feature of life. In contrast, consistent, frequent, persistent hallucinations during waking are strongly associated with clinical disorders; in particular delirium, eye disease, psychosis, and dementia. Research interest in these disorders has driven a rapid expansion in investigatory techniques, new evidence, and explanatory models. In parallel, a move to generative models of normal visual function has resolved the theoretical tension between veridical and hallucinatory perceptions. From initial fragmented areas of investigation, the field has become increasingly coherent over the last decade. Controversies and gaps remain, but for the first time the shapes of possible unifying models are becoming clear, along with the techniques for testing these. This book provides a comprehensive survey of the neuroscience of visual hallucinations and the clinical techniques for testing these. It brings together the very latest evidence from cognitive neuropsychology, neuroimaging, neuropathology, and neuropharmacology, placing this within current models of visual perception. Leading researchers from a range of clinical and basic science areas describe visual hallucinations in their historical and scientific context, combining introductory information with up-to-date discoveries. They discuss results from the main investigatory techniques applied in a range of clinical disorders. The final section outlines future research directions investigating the potential for new understandings of veridical and hallucinatory perceptions, and for treatments of problematic hallucinations. Fully comprehensive, this is an essential reference for clinicians in the fields of the psychology and psychiatry of hallucinations, as well as for researchers in departments, research institutes and libraries. It has strong foundations in neuroscience, cognitive science, optometry, psychiatry, psychology, clinical medicine, and philosophy. With its lucid explanation and many illustrations, it is a clear resource for educators and advanced undergraduate and graduate students.

Asymptotic theory of time varying networks with memory and heterogeneous activation pattern

In this thesis work we develop a new generative model of social networks belonging to the family of Time Varying Networks. The importance of correctly modelling the mechanisms shaping the growth of a network and the dynamics of the edges activation and inactivation are of central importance in network science. Indeed, by means of generative models that mimic the real-world dynamics of contacts in social networks it is possible to forecast the outcome of an epidemic process, optimize the immunization campaign or optimally spread an information among individuals. This task can now be tackled taking advantage of the recent availability of large-scale, high-quality and time-resolved datasets. This wealth of digital data has allowed to deepen our understanding of the structure and properties of many real-world networks. Moreover, the empirical evidence of a temporal dimension in networks prompted the switch of paradigm from a static representation of graphs to a time varying one. In this work we exploit the Activity-Driven paradigm (a modeling tool belonging to the family of Time-Varying-Networks) to develop a general dynamical model that encodes fundamental mechanism shaping the social networks' topology and its temporal structure: social capital allocation and burstiness. The former accounts for the fact that individuals does not randomly invest their time and social interactions but they rather allocate it toward already known nodes of the network. The latter accounts for the heavy-tailed distributions of the inter-event time in social networks. We then empirically measure the properties of these two mechanisms from seven real-world datasets and develop a data-driven model, analytically solving it. We then check the results against numerical simulations and test our predictions with real-world datasets, finding a good agreement between the two. Moreover, we find and characterize a non-trivial interplay between burstiness and social capital allocation in the parameters phase space. Finally, we present a novel approach to the development of a complete generative model of Time-Varying-Networks. This model is inspired by the Kaufman's adjacent possible theory and is based on a generalized version of the Polya's urn. Remarkably, most of the complex and heterogeneous feature of real-world social networks are naturally reproduced by this dynamical model, together with many high-order topological properties (clustering coefficient, community structure etc.).

The High-Redshift Quasar Luminosity Function from Multi-Epoch Imaging Surveys

Thesis (Ph.D.)--University of Washington, 2016-08