Guiding Monte Carlo tree searches with neural networks in the game of go

A dissertation submitted in fulfillment of the requirements to the degree of Master in Computer Science and Computer Engineering

Exotic dynamics in a firing rate model of neural tissue with threshold accommodation

Many of the equations describing the dynamics of neural systems are written in terms of firing rate functions, which themselves are often taken to be threshold functions of synaptic activity. Dating back to work by Hill in 1936 it has been recognized that more realistic models of neural tissue can be obtained with the introduction of state-dependent dynamic thresholds. In this paper we treat a specific phenomenological model of threshold accommodation that mimics many of the properties originally described by Hill. Importantly we explore the consequences of this dynamic threshold at the tissue level, by modifying a standard neural field model of Wilson-Cowan type. As in the case without threshold accommodation classical Mexican-Hat connectivity is shown to allow for the existence of spatially localized states (bumps) in both one and two dimensions. Importantly an analysis of bump stability in one dimension, using recent Evans function techniques, shows that bumps may undergo instabilities leading to the emergence of both breathers and traveling waves. Moreover, a similar analysis for traveling pulses leads to the conditions necessary to observe a stable traveling breather. In the regime where a bump solution does not exist direct numerical simulations show the possibility of self-replicating bumps via a form of bump splitting. Simulations in two space dimensions show analogous localized and traveling solutions to those seen in one dimension. Indeed dynamical behavior in this neural model appears reminiscent of that seen in other dissipative systems that support localized structures, and in particular those of coupled cubic complex Ginzburg-Landau equations. Further numerical explorations illustrate that the traveling pulses in this model exhibit particle like properties, similar to those of dispersive solitons observed in some three component reaction-diffusion systems. A preliminary account of this work first appeared in S Coombes and M R Owen, Bumps, breathers, and waves in a neural network with spike frequency adaptation, Physical Review Letters 94 (2005), 148102(1-4).

Tutoria da anatomia mamária feminina utilizando uma rede neural artificial interactive activation and competition orientada a serviço

Dissertação (mestrado)—Universidade de Brasília, Faculdade Gama, Programa de Pós-Graduação em Engenharia Biomédica, 2015.

Delays in activity based neural networks

In this paper we study the effect of two distinct discrete delays on the dynamics of a Wilson-Cowan neural network. This activity based model describes the dynamics of synaptically interacting excitatory and inhibitory neuronal populations. We discuss the interpretation of the delays in the language of neurobiology and show how they can contribute to the generation of network rhythms. First we focus on the use of linear stability theory to show how to destabilise a fixed point, leading to the onset of oscillatory behaviour. Next we show for the choice of a Heaviside nonlinearity for the firing rate that such emergent oscillations can be either synchronous or anti-synchronous depending on whether inhibition or excitation dominates the network architecture. To probe the behaviour of smooth (sigmoidal) nonlinear firing rates we use a mixture of numerical bifurcation analysis and direct simulations, and uncover parameter windows that support chaotic behaviour. Finally we comment on the role of delays in the generation of bursting oscillations, and discuss natural extensions of the work in this paper.

A Learning Function for Parameter Reduction in Spiking Neural Networks with Radial Basis Function

Spiking neural networks - networks that encode information in the timing of spikes - are arising as a new approach in the artificial neural networks paradigm, emergent from cognitive science. One of these new models is the pulsed neural network with radial basis function, a network able to store information in the axonal propagation delay of neurons. Learning algorithms have been proposed to this model looking for mapping input pulses into output pulses. Recently, a new method was proposed to encode constant data into a temporal sequence of spikes, stimulating deeper studies in order to establish abilities and frontiers of this new approach. However, a well known problem of this kind of network is the high number of free parameters - more that 15 - to be properly configured or tuned in order to allow network convergence. This work presents for the first time a new learning function for this network training that allow the automatic configuration of one of the key network parameters: the synaptic weight decreasing factor.

Mutual Information-based RBM Neural Networks

(Deep) neural networks are increasingly being used for various computer vision and pattern recognition tasks due to their strong ability to learn highly discriminative features. However, quantitative analysis of their classication ability and design philosophies are still nebulous. In this work, we use information theory to analyze the concatenated restricted Boltzmann machines (RBMs) and propose a mutual information-based RBM neural networks (MI-RBM). We develop a novel pretraining algorithm to maximize the mutual information between RBMs. Extensive experimental results on various classication tasks show the eectiveness of the proposed approach.

Controle preditivo neural aplicado à processos petroquímicos

A pesquisa tem como objetivo desenvolver uma estrutura de controle preditivo neural, com o intuito de controlar um processo de pH, caracterizado por ser um sistema SISO (Single Input - Single Output). O controle de pH é um processo de grande importância na indústria petroquímica, onde se deseja manter constante o nível de acidez de um produto ou neutralizar o afluente de uma planta de tratamento de fluidos. O processo de controle de pH exige robustez do sistema de controle, pois este processo pode ter ganho estático e dinâmica nãolineares. O controlador preditivo neural envolve duas outras teorias para o seu desenvolvimento, a primeira referente ao controle preditivo e a outra a redes neurais artificiais (RNA s). Este controlador pode ser dividido em dois blocos, um responsável pela identificação e outro pelo o cálculo do sinal de controle. Para realizar a identificação neural é utilizada uma RNA com arquitetura feedforward multicamadas com aprendizagem baseada na metodologia da Propagação Retroativa do Erro (Error Back Propagation). A partir de dados de entrada e saída da planta é iniciado o treinamento offline da rede. Dessa forma, os pesos sinápticos são ajustados e a rede está apta para representar o sistema com a máxima precisão possível. O modelo neural gerado é usado para predizer as saídas futuras do sistema, com isso o otimizador calcula uma série de ações de controle, através da minimização de uma função objetivo quadrática, fazendo com que a saída do processo siga um sinal de referência desejado. Foram desenvolvidos dois aplicativos, ambos na plataforma Builder C++, o primeiro realiza a identificação, via redes neurais e o segundo é responsável pelo controle do processo. As ferramentas aqui implementadas e aplicadas são genéricas, ambas permitem a aplicação da estrutura de controle a qualquer novo processo

Network Construction and Graph Theoretical Analysis of Functional Language Networks in Pediatric Epilepsy

This dissertation introduces a new approach for assessing the effects of pediatric epilepsy on the language connectome. Two novel data-driven network construction approaches are presented. These methods rely on connecting different brain regions using either extent or intensity of language related activations as identified by independent component analysis of fMRI data. An auditory description decision task (ADDT) paradigm was used to activate the language network for 29 patients and 30 controls recruited from three major pediatric hospitals. Empirical evaluations illustrated that pediatric epilepsy can cause, or is associated with, a network efficiency reduction. Patients showed a propensity to inefficiently employ the whole brain network to perform the ADDT language task; on the contrary, controls seemed to efficiently use smaller segregated network components to achieve the same task. To explain the causes of the decreased efficiency, graph theoretical analysis was carried out. The analysis revealed no substantial global network feature differences between the patient and control groups. It also showed that for both subject groups the language network exhibited small-world characteristics; however, the patient’s extent of activation network showed a tendency towards more random networks. It was also shown that the intensity of activation network displayed ipsilateral hub reorganization on the local level. The left hemispheric hubs displayed greater centrality values for patients, whereas the right hemispheric hubs displayed greater centrality values for controls. This hub hemispheric disparity was not correlated with a right atypical language laterality found in six patients. Finally it was shown that a multi-level unsupervised clustering scheme based on self-organizing maps, a type of artificial neural network, and k-means was able to fairly and blindly separate the subjects into their respective patient or control groups. The clustering was initiated using the local nodal centrality measurements only. Compared to the extent of activation network, the intensity of activation network clustering demonstrated better precision. This outcome supports the assertion that the local centrality differences presented by the intensity of activation network can be associated with focal epilepsy.

Optimal control of non-linear systems through hybrid cell-mapping/artificial-neural-networks techniques

In this paper, a real-time optimal control technique for non-linear plants is proposed. The control system makes use of the cell-mapping (CM) techniques, widely used for the global analysis of highly non-linear systems. The CM framework is employed for designing approximate optimal controllers via a control variable discretization. Furthermore, CM-based designs can be improved by the use of supervised feedforward artificial neural networks (ANNs), which have proved to be universal and efficient tools for function approximation, providing also very fast responses. The quantitative nature of the approximate CM solutions fits very well with ANNs characteristics. Here, we propose several control architectures which combine, in a different manner, supervised neural networks and CM control algorithms. On the one hand, different CM control laws computed for various target objectives can be employed for training a neural network, explicitly including the target information in the input vectors. This way, tracking problems, in addition to regulation ones, can be addressed in a fast and unified manner, obtaining smooth, averaged and global feedback control laws. On the other hand, adjoining CM and ANNs are also combined into a hybrid architecture to address problems where accuracy and real-time response are critical. Finally, some optimal control problems are solved with the proposed CM, neural and hybrid techniques, illustrating their good performance.

Digital soil mapping using reference area and artificial neural networks.

Digital soil mapping is an alternative for the recognition of soil classes in areas where pedological surveys are not available. The main aim of this study was to obtain a digital soil map using artificial neural networks (ANN) and environmental variables that express soillandscape relationships. This study was carried out in an area of 11,072 ha located in the Barra Bonita municipality, state of São Paulo, Brazil. A soil survey was obtained from a reference area of approximately 500 ha located in the center of the area studied. With the mapping units identified together with the environmental variables elevation, slope, slope plan, slope profile, convergence index, geology and geomorphic surfaces, a supervised classification by ANN was implemented. The neural network simulator used was the Java NNS with the learning algorithm "back propagation." Reference points were collected for evaluating the performance of the digital map produced. The occurrence of soils in the landscape obtained in the reference area was observed in the following digital classification: medium-textured soils at the highest positions of the landscape, originating from sandstone, and clayey loam soils in the end thirds of the hillsides due to the greater presence of basalt. The variables elevation and slope were the most important factors for discriminating soil class through the ANN. An accuracy level of 82% between the reference points and the digital classification was observed. The methodology proposed allowed for a preliminary soil classification of an area not previously mapped using mapping units obtained in a reference area

Robotic Sensing and Manipulation of Deformable Linear Objects with Learning-based methods

Nowadays robotic applications are widespread and most of the manipulation tasks are efficiently solved. However, Deformable-Objects (DOs) still represent a huge limitation for robots. The main difficulty in DOs manipulation is dealing with the shape and dynamics uncertainties, which prevents the use of model-based approaches (since they are excessively computationally complex) and makes sensory data difficult to interpret. This thesis reports the research activities aimed to address some applications in robotic manipulation and sensing of Deformable-Linear-Objects (DLOs), with particular focus to electric wires. In all the works, a significant effort was made in the study of an effective strategy for analyzing sensory signals with various machine learning algorithms. In the former part of the document, the main focus concerns the wire terminals, i.e. detection, grasping, and insertion. First, a pipeline that integrates vision and tactile sensing is developed, then further improvements are proposed for each module. A novel procedure is proposed to gather and label massive amounts of training images for object detection with minimal human intervention. Together with this strategy, we extend a generic object detector based on Convolutional-Neural-Networks for orientation prediction. The insertion task is also extended by developing a closed-loop control capable to guide the insertion of a longer and curved segment of wire through a hole, where the contact forces are estimated by means of a Recurrent-Neural-Network. In the latter part of the thesis, the interest shifts to the DLO shape. Robotic reshaping of a DLO is addressed by means of a sequence of pick-and-place primitives, while a decision making process driven by visual data learns the optimal grasping locations exploiting Deep Q-learning and finds the best releasing point. The success of the solution leverages on a reliable interpretation of the DLO shape. For this reason, further developments are made on the visual segmentation.

Concepts and methods for efficient decentralized learning in federated settings

Deep Neural Networks (DNNs) have revolutionized a wide range of applications beyond traditional machine learning and artificial intelligence fields, e.g., computer vision, healthcare, natural language processing and others. At the same time, edge devices have become central in our society, generating an unprecedented amount of data which could be used to train data-hungry models such as DNNs. However, the potentially sensitive or confidential nature of gathered data poses privacy concerns when storing and processing them in centralized locations. To this purpose, decentralized learning decouples model training from the need of directly accessing raw data, by alternating on-device training and periodic communications. The ability of distilling knowledge from decentralized data, however, comes at the cost of facing more challenging learning settings, such as coping with heterogeneous hardware and network connectivity, statistical diversity of data, and ensuring verifiable privacy guarantees. This Thesis proposes an extensive overview of decentralized learning literature, including a novel taxonomy and a detailed description of the most relevant system-level contributions in the related literature for privacy, communication efficiency, data and system heterogeneity, and poisoning defense. Next, this Thesis presents the design of an original solution to tackle communication efficiency and system heterogeneity, and empirically evaluates it on federated settings. For communication efficiency, an original method, specifically designed for Convolutional Neural Networks, is also described and evaluated against the state-of-the-art. Furthermore, this Thesis provides an in-depth review of recently proposed methods to tackle the performance degradation introduced by data heterogeneity, followed by empirical evaluations on challenging data distributions, highlighting strengths and possible weaknesses of the considered solutions. Finally, this Thesis presents a novel perspective on the usage of Knowledge Distillation as a mean for optimizing decentralized learning systems in settings characterized by data heterogeneity or system heterogeneity. Our vision on relevant future research directions close the manuscript.

Data driven regularization models of non-linear ill-posed inverse problems in imaging

Imaging technologies are widely used in application fields such as natural sciences, engineering, medicine, and life sciences. A broad class of imaging problems reduces to solve ill-posed inverse problems (IPs). Traditional strategies to solve these ill-posed IPs rely on variational regularization methods, which are based on minimization of suitable energies, and make use of knowledge about the image formation model (forward operator) and prior knowledge on the solution, but lack in incorporating knowledge directly from data. On the other hand, the more recent learned approaches can easily learn the intricate statistics of images depending on a large set of data, but do not have a systematic method for incorporating prior knowledge about the image formation model. The main purpose of this thesis is to discuss data-driven image reconstruction methods which combine the benefits of these two different reconstruction strategies for the solution of highly nonlinear ill-posed inverse problems. Mathematical formulation and numerical approaches for image IPs, including linear as well as strongly nonlinear problems are described. More specifically we address the Electrical impedance Tomography (EIT) reconstruction problem by unrolling the regularized Gauss-Newton method and integrating the regularization learned by a data-adaptive neural network. Furthermore we investigate the solution of non-linear ill-posed IPs introducing a deep-PnP framework that integrates the graph convolutional denoiser into the proximal Gauss-Newton method with a practical application to the EIT, a recently introduced promising imaging technique. Efficient algorithms are then applied to the solution of the limited electrods problem in EIT, combining compressive sensing techniques and deep learning strategies. Finally, a transformer-based neural network architecture is adapted to restore the noisy solution of the Computed Tomography problem recovered using the filtered back-projection method.

Directly training spiking neural networks for cyber-physical systems: from supervised to reinforcement learning

Spiking Neural Networks (SNNs) are bio-inspired Artificial Neural Networks (ANNs) utilizing discrete spiking signals, akin to neuron communication in the brain, making them ideal for real-time and energy-efficient Cyber-Physical Systems (CPSs). This thesis explores their potential in Structural Health Monitoring (SHM), leveraging low-cost MEMS accelerometers for early damage detection in motorway bridges. The study focuses on Long Short-Term SNNs (LSNNs), although their complex learning processes pose challenges. Comparing LSNNs with other ANN models and training algorithms for SHM, findings indicate LSNNs' effectiveness in damage identification, comparable to ANNs trained using traditional methods. Additionally, an optimized embedded LSNN implementation demonstrates a 54% reduction in execution time, but with longer pre-processing due to spike-based encoding. Furthermore, SNNs are applied in UAV obstacle avoidance, trained directly using a Reinforcement Learning (RL) algorithm with event-based input from a Dynamic Vision Sensor (DVS). Performance evaluation against Convolutional Neural Networks (CNNs) highlights SNNs' superior energy efficiency, showing a 6x decrease in energy consumption. The study also investigates embedded SNN implementations' latency and throughput in real-world deployments, emphasizing their potential for energy-efficient monitoring systems. This research contributes to advancing SHM and UAV obstacle avoidance through SNNs' efficient information processing and decision-making capabilities within CPS domains.