Search

The bond strength between ultra-high-performance concrete (UHPC) and normal-strength concrete (NC) plays an important role in governing the composite specimens’ overall behaviors. Unfortunately, there are still no widely accepted formulas targeting UHPC–NC interfacial strength, either in their specifications or in research papers. To this end, this study constructs an experimental database, consisting of 563 and 338 specimens for splitting and slant shear tests, respectively. Moreover, an additional 35 specimens for “improved” slant shear tests were performed, which could circumvent concrete crushing and trigger interfacial debonding. Additionally, for the first time in our tests, the effect of casting sequence on UHPC–NC bond strength was identified. Based on the database, an artificial neural network (ANN) model is proposed with the following inputs: namely, the normal stress perpendicular to the interface, the interface roughness, and the compressive strengths of the UHPC and NC materials. Based on the ANN analyses, the explicit expression of UHPC–NC bond strength is proposed, which significantly lowers the prediction error. To be fully compatible with the specifications, the conventional shear-friction formula is modified. By splitting the total force into adhesion and friction forces, the modified formula additionally takes the casting sequence into account. Although sacrificing accuracy to some extent compared to the ANN model, the modified formula relies on a solid physical basis and its accuracy is enhanced significantly compared to the existing formulas in specifications or research papers.

Tutor: Antonio Agudo Self-driving cars have gained popularity in recent years. The increase in computational power coupled with big data has enabled the application of artificial intelligence in such complex areas as driving. Specifically, it is the field of deep learning that has revolutionized these applications. The use of models based on artificial neural networks has achieved great results in a short period of time. In this project we adapt a conventional radio-controlled car into a small-scale autonomous system. The car is able to drive autonomously thanks to the information provided by an RGB monocular camera. We analyze the whole process from assembly, data acquisition, training and test bench. Els cotxes autònoms han guanyat popularitat aquests darrers anys. L’augment del poder computacional juntament amb el big data ha permès aplicar la intel·ligència artificial en àmbits tan complexos com és el de la conducció. Concretament, és el camp de l’aprenentatge profund el que ha revolucionat aquestes aplicacions. L’ús de models basats en xarxes neuronals artificials ha aconseguit grans resultats en un breu període de temps. En aquest projecte adaptem un cotxe radio-control convencional per convertir-lo en un sistema autònom a petita escala. El cotxe és capaç de conduir de manera autònoma gràcies a la informació que li proporciona una càmera monocular RGB. Analitzem tot el procés des de l’ensamblatge, l’adquisició de dades, l’entrenament i el banc de proves. Los coches autónomos han ganado popularidad en los últimos años. El aumento del poder computacional junto con el big data ha permitido aplicar la inteligencia artificial en ámbitos tan complejos como es el de la conducción. Concretamente, es el campo del aprendizaje profundo el que ha revolucionado estas aplicaciones. El uso de modelos basados en redes neuronales artificiales ha conseguido grandes resultados en un breve periodo de tiempo. En este proyecto adaptamos un coche radio-control convencional para convertirlo en un sistema autónomo a pequeña escala. El coche es capaz de conducir de manera autónoma gracias a la información que le proporciona una cámara monocular RGB. Analizamos todo el proceso desde el ensamblaje, la adquisición de datos, el entrenamiento y el banco de pruebas.

Orientador: Prof. Dr. Gustavo Sousa Pavani Dissertação (mestrado) - Universidade Federal do ABC, Programa de Pós-Graduação em Engenharia da Informação, Santo André, 2021. Neste trabalho, busca-se realizar uma análise técnico-econômica para se avaliar os ganhos do compartilhamento de infraestrutura na implantação da rede móvel 5G no Brasil. Com efeito, são avaliadas arquiteturas de compartilhamento da rede de acesso de rádio e de redes neutras para alcançar este objetivo tendo como base a rede móvel já instalada com a tecnologia 4G. Nesse contexto, são apresentados os possíveis ganhos de compartilhamento de infraestrutura para as Estações Rádio Base (ERBs) e para as redes de backhaul. Também é apresentada uma ferramenta de visualização para as ERBs 4G implantadas no Brasil e para a hipotética rede de backhaul necessária para atendimento dos requisitos do 5G. In this work, we propose a technical-economic analysis to assess the improvement of infrastructure sharing in the deployment of the 5G mobile network in Brazil. In effect, we evaluate the architectures for Radio Access Network (RAN) sharing and neutral networking by taking into consideration the mobile network already deployed for the 4G technology. In this context, we present the possible infrastructure sharing improvements for base stations and backhaul networks. We also present a visualization tool for the deployed 4G base stations and the hypothetical backhaul network needed to fulfill the 5G requirements.

Linear feature extraction is crucial for special objects in semantic segmentation networks, such as slot marking and lanes. The objects with linear characteristics have global contextual information dependency. It is very difficult to capture the complete information of these objects in semantic segmentation tasks. To improve the linear feature extraction ability of the semantic segmentation network, we propose introducing the dilated convolution with vertical and horizontal kernels (DVH) into the task of feature extraction in semantic segmentation networks. Meanwhile, we figure out the outcome if we put the different vertical and horizontal kernels on different places in the semantic segmentation networks. Our networks are trained on the basis of the SS dataset, the TuSimple lane dataset and the Massachusetts Roads dataset. These datasets consist of slot marking, lanes, and road images. The research results show that our method improves the accuracy of the slot marking segmentation of the SS dataset by 2%. Compared with other state-of-the-art methods, our UnetDVH-Linear (v1) obtains better accuracy on the TuSimple Benchmark Lane Detection Challenge with a value of 97.53%. To prove the generalization of our models, road segmentation experiments were performed on aerial images. Without data argumentation, the segmentation accuracy of our model on the Massachusetts roads dataset is 95.3%. Moreover, our models perform better than other models when training with the same loss function and experimental settings. The experiment result shows that the dilated convolution with vertical and horizontal kernels will enhance the neural network on linear feature extraction.

Transcriptional regulation is crucial for all biological processes and well investigated at the molecular level for a wide range of organisms. However, it is quite unclear how innovations, such as the activity of a novel regulatory element, evolve. In the case of transcription factor (TF) binding, both a novel TF and a novel-binding site would need to evolve concertedly. Since promiscuous functions have recently been identified as important intermediate steps in creating novel specific functions in many areas such as enzyme evolution and protein–protein interactions, we ask here how promiscuous binding of TFs to TF-binding sites (TFBSs) affects the robustness and evolvability of this tightly regulated system. Specifically, we investigate the binding behavior of several hundred TFs from different species at unprecedented breadth. Our results illustrate multiple aspects of TF-binding interactions, ranging from correlations between the strength of the interaction bond and specificity, to preferences regarding TFBS nucleotide composition in relation to both domains and binding specificity. We identified a subset of high A/T binding motifs. Motifs in this subset had many functionally neutral one-error mutants, and were bound by multiple different binding domains. Our results indicate that, especially for some TF–TFBS associations, low binding specificity confers high degrees of evolvability, that is that few mutations facilitate rapid changes in transcriptional regulation, in particular for large and old TF families. In this study we identify binding motifs exhibiting behavior indicating high evolutionary potential for innovations in transcriptional regulation.

The problem of RNA design attempts to construct RNA sequences that perform a predefined biological function, identified by several additional constraints. One of the foremost objective of RNA design is that the designed RNA sequence should adopt a predefined target secondary structure preferentially to any alternative structure, according to a given metrics and folding model. It was observed in several works that some secondary structures are undesignable, i.e. no RNA sequence can fold into the target structure while satisfying some criterion measuring how preferential this folding is compared to alternative conformations.In this paper, we show that the proportion of designable secondary structures decreases exponentially with the size of the target secondary structure, for various popular combinations of energy models and design objectives. This exponential decay is, at least in part, due to the existence of undesignable motifs, which can be generically constructed, and jointly analyzed to yield asymptotic upper-bounds on the number of designable structures.

Transcriptional regulation is crucial for all biological processes and well investigated at the molecular level for a wide range of organisms. However, it is quite unclear how innovations, such as the activity of a novel regulatory element, evolve. In the case of transcription factor (TF) binding, both a novel TF and a novel-binding site would need to evolve concertedly. Since promiscuous functions have recently been identified as important intermediate steps in creating novel specific functions in many areas such as enzyme evolution and protein–protein interactions, we ask here how promiscuous binding of TFs to TF-binding sites (TFBSs) affects the robustness and evolvability of this tightly regulated system. Specifically, we investigate the binding behavior of several hundred TFs from different species at unprecedented breadth. Our results illustrate multiple aspects of TF-binding interactions, ranging from correlations between the strength of the interaction bond and specificity, to preferences regarding TFBS nucleotide composition in relation to both domains and binding specificity. We identified a subset of high A/T binding motifs. Motifs in this subset had many functionally neutral one-error mutants, and were bound by multiple different binding domains. Our results indicate that, especially for some TF–TFBS associations, low binding specificity confers high degrees of evolvability, that is that few mutations facilitate rapid changes in transcriptional regulation, in particular for large and old TF families. In this study we identify binding motifs exhibiting behavior indicating high evolutionary potential for innovations in transcriptional regulation.

Background Wright’s metaphor of the fitness landscape has shaped and conditioned our view of the adaptation of populations for almost a century. Since its inception, and including criticism raised by Wright himself, the concept has been surrounded by controversy. Among others, the debate stems from the intrinsic difficulty to capture important features of the space of genotypes, such as its high dimensionality or the existence of abundant ridges, in a visually appealing two-dimensional picture. Two additional currently widespread observations come to further constrain the applicability of the original metaphor: the very skewed distribution of phenotype sizes (which may actively prevent, due to entropic effects, the achievement of fitness maxima), and functional promiscuity (i.e. the existence of secondary functions which entail partial adaptation to environments never encountered before by the population). Results Here we revise some of the shortcomings of the fitness landscape metaphor and propose a new “scape” formed by interconnected layers, each layer containing the phenotypes viable in a given environment. Different phenotypes within a layer are accessible through mutations with selective value, while neutral mutations cause displacements of populations within a phenotype. A different environment is represented as a separated layer, where phenotypes may have new fitness values, other phenotypes may be viable, and the same genotype may yield a different phenotype, representing genotypic promiscuity. This scenario explicitly includes the many-to-many structure of the genotype-to-phenotype map. A number of empirical observations regarding the adaptation of populations in the light of adaptive multiscapes are reviewed. Conclusions Several shortcomings of Wright’s visualization of fitness landscapes can be overcome through adaptive multiscapes. Relevant aspects of population adaptation, such as neutral drift, functional promiscuity or environment-dependent fitness, as well as entropic trapping and the concomitant impossibility to reach fitness peaks are visualized at once. Adaptive multiscapes should aid in the qualitative understanding of the multiple pathways involved in evolutionary dynamics. Reviewers This article was reviewed by Eugene Koonin and Ricard Solé.

It has been argued that much of evolution takes place in the absence of fitness gradients. Such periods of evolution can be analysed by examining the mutational network formed by sequences of equal fitness, that is, the neutral network. It has been demonstrated that, in large populations under a high mutation rate, the population distribution over the neutral network and average mutational robustness are given by the principal eigenvector and eigen- value, respectively, of the network's adjacency matrix. However, little progress has been made towards understanding the manner in which the topology of the neutral network influences the resulting population distribution and robustness. In this work, we build on recent results from spectral graph theory and utilize numerical methods to enhance our understanding of how populations distribute themselves over neutral networks. We demonstrate that, in the presence of certain topological features, the population will undergo an exploration catastrophe and become confined to a small portion of the network. We further derive approximations, in terms of mutational biases, for the population distribution and average robustness in networks with a homogeneous structure. The applicability of these results is explored, first, by a detailed review of the literature in both evolutionary computing and biology concerning the structure of neutral networks. This is extended by studying the actual and predicted population distribution over the neutral networks of H1N1 and H3N2 influenza haemagglutinin during seasons between 2005 and 2016. It is shown that, in some instances, these populations experience an exploration catastrophe. These results provide insight into the behaviour of populations on neutral networks, demonstrating that neutrality does not necessarily lead to an exploration of genotype/phenotype space or an associated increase in population diversity. Moreover, they provide a plausible explanation for conflicting results concerning the relationship between robustness and evolvability.

Millions of people went from Romania to make a fortune in foreign countries and a few of them came back and make their life easier at home. There are some preconditions for each of them and there have some expectations of the outcome of the time spent. Because these factors are not related in a straight way, the analysis of the problem needs knowledge discovery. We choose artificial neural networks to recognize some patterns related to this problem. In neural network analyses, sometimes the results are good, but we cannot transform it in adequate conclusions. These results are fitted to predict the outcomes of new data. Neural networks mimic the functions of the human brain. In comparison with the fuzzy sets, the artificial neural network is the hardware, and the fuzzy set model is the software approach of, sometimes, the same problem. We have a brunch of problem, for that the neural network model cannot be substituted by the fuzzy model, although the results of the fuzzy approach are most human-readable. Our problem is maybe closer to the fuzzy system, but this time we want to verify the neural network approach. We use this analysis to discover patterns between the inputs of a human that goes outboard to work in foreign country for a period of time and the outputs of the work, like material realization, education and professional profit.

This paper is concerned with the fixed-time consensus tracking problem for multi-AUV (autonomous underwater vehicle) systems with uncertain parameters and external disturbances. Firstly, a fixed-time terminal sliding mode is proposed, which can avoid the singularity problem. Then, a continuous distributed consensus tracking control law is designed based on Neutral Network approximation technique, which can guarantee the consensus tracking errors converge to the desired regions in fixed time. A simulation example is given to show the effectiveness of proposed methods.

Linear principal component analysis (PCA) can be extended to a nonlinear PCA by using artificial neural networks. But the benefit of curved components requires a careful control of the model complexity. Moreover, standard techniques for model selection, including cross-validation and more generally the use of an independent test set, fail when applied to nonlinear PCA because of its inherent unsupervised characteristics. This paper presents a new approach for validating the complexity of nonlinear PCA models by using the error in missing data estimation as a criterion for model selection. It is motivated by the idea that only the model of optimal complexity is able to predict missing values with the highest accuracy. While standard test set validation usually favours over-fitted nonlinear PCA models, the proposed model validation approach correctly selects the optimal model complexity., 12 pages, 5 figures

International audience; We analyze here the evolutionary consequences of selection with delay in a population genetics context. In the classical works on evolutionary dynamics, an individual produces off-springs in direct proportion to its fitness, a process in which mutations may occur. In the present scenario of delayed selection, individuals that acquire deleterious mutations can still reproduce unharmed for several generations. During this time delay, the damage passed on to off-springs can potentially be repaired by subsequent compensatory mutations. In the absence of such a repair, the individual becomes sterile. Here we study the population-genetic effects of such a time delay by means of both numerical simulations and theoretical modeling. The results show that delayed selection lowers the extinction threshold, endangering the survival of the population. Surprisingly, however, no traces of this delay effect are encountered in the sequence diversity of the population. These conclusions suggest that delayed selection is hard to detect in genetic data and thus could be a wide-spread but rarely detected phenomenon.

International audience; Undoubtedly, viruses represent a major threat faced by human and veterinary medicines and by agronomy. The rapid evolution of viruses enables them to escape from natural immunities and from state-of-the-art antiviral treatments, with new viruses periodically emerging with deadly consequences. Viruses have also become powerful and are increasingly used tools in the field of experimental evolution. A growing body of evidence points that the evolution of viruses is mainly determined by key features such as their compacted genomes, enormous population sizes, and short generation times. In addition, RNA viruses also present large selection coefficients, antagonistic epistasis, and high mutation rates. Most of this knowledge comes from studies that have used either bacteriophages or animal viruses in cell cultures as experimental systems. However, plant viruses provide almost identical advantages for evolutionary studies and, in addition, offer an invaluable tool for studying the interplay between viruses and pluricellular hosts. Without seeking to be exhaustive, here we summarize some peculiarities of plant viruses and review recent experiments that have explored important questions on evolution, such as the role of deleterious mutation and neutrality, the effect of different transmission modes in the evolution of virulence, and the heterogeneous selective constraints imposed by multiple hosts.

In this thesis, a new mathematical formalism for analyzing evolutionary dynamics is developed. This formalism combines ideas and methods from statistical mechanics, mathematical population genetics, and dynamical systems theory to describe the dynamics of evolving populations. In particular, the work shows how the maximum entropy formalism of statistical mechanics can be extended to apply to simple evolutionary systems, such that "macroscopic" equations of motion can be constructed from an underlying "microscopic" evolutionary dynamics. More specifically, the thesis studies "epochal evolution"; a dynamical phenomenon which is frequently observed in evolutionary dynamics. In epochal evolution, some macroscopic state variables that describe the evolving population exhibit an alternation of periods of stasis (epochs) and sudden transitions (innovations). For populations evolving in a constant selective environment there are two main mechanisms that may bring about epochal evolution. In the first mechanism the population is imagined to evolve on a "fitness landscape" that assigns a fitness to each point in a space of genotypes. Metastability then occurs when the population gets stuck around a "local optimum" in the fitness landscape. An innovation takes place when a rare sequence of mutations creates a lineage of individuals that crosses a valley of low fitness toward a local optimum of higher fitness. In the second mechanism, the genotype space is thought to decompose into a relatively small number of "neutral subbasins": large connected sets of equal fitness genotypes. In this view, an epoch corresponds to a time period in which the fittest members of the population diffuse through a neutral subbasin under mutation until one of them discovers a (rare) connection to a neutral subbasin of higher fitness. In somewhat different methodology, in the first mechanism the metastability is caused by a "fitness barrier", whereas in the second mechanism it is caused by an "entropy barrier". In this thesis, the evolutionary dynamics in the presence of both fitness and entropy barriers is studied, although the focus is on fitness functions with entropy barriers. For a large class of simple fitness functions we derive, using the maximum entropy methodology, equations of motion on the level of fitness distributions from the underlying microscopic dynamics of selection and mutation on genotypes. In the "thermodynamic limit" of infinite population sizes the population follows these equations of motion precisely while for finite populations the population only follows these equations of motion on average at each time step. From this formulation of the finite population dynamics we determine explicitly the locations of epochs in the space of fitness distributions, the stochastic dynamics within and between the epochs, the average durations of epochs, and the stability of epochs. The results also bear directly on the dynamics of evolutionary search algorithms such as genetic algorithms. In two chapters of this thesis, our mathematical model is used to derive optimal parameter settings for evolutionary search on a wide class of fitness functions. The analysis suggests that optimal evolutionary search occurs in a parameter regime where the highest fitness strings in the population are only marginally stable. We also studied in detail the way in which an evolving population will spread through a neutral subbasin or neutral network. As the theory shows, the limit distribution of the population over a neutral network is independent of almost all evolutionary parameters and is determined solely by the topology of the neutral network. Additionally, this distribution is concentrated at those areas of the neutral network where the "neutrality" is largest. This implies that under neutral evolution the population evolves to increase its "mutational robustness" and that the mutational robustness that it attains is only dependent on the topology of the neutral network on which the population evolves. Finally, we studied and compared the barrier crossing times for fitness and entropy barriers. The results show that entropy barrier crossing in evolving population takes place on much shorter time scales than fitness barrier crossing. This suggests that in evolution, the escape from a metastable state is most likely to occur along neutral paths in genotype space.

In this thesis, a new mathematical formalism for analyzing evolutionary dynamics is developed. This formalism combines ideas and methods from statistical mechanics, mathematical population genetics, and dynamical systems theory to describe the dynamics of evolving populations. In particular, the work shows how the maximum entropy formalism of statistical mechanics can be extended to apply to simple evolutionary systems, such that "macroscopic" equations of motion can be constructed from an underlying "microscopic" evolutionary dynamics. More specifically, the thesis studies "epochal evolution"; a dynamical phenomenon which is frequently observed in evolutionary dynamics. In epochal evolution, some macroscopic state variables that describe the evolving population exhibit an alternation of periods of stasis (epochs) and sudden transitions (innovations). For populations evolving in a constant selective environment there are two main mechanisms that may bring about epochal evolution. In the first mechanism the population is imagined to evolve on a "fitness landscape" that assigns a fitness to each point in a space of genotypes. Metastability then occurs when the population gets stuck around a "local optimum" in the fitness landscape. An innovation takes place when a rare sequence of mutations creates a lineage of individuals that crosses a valley of low fitness toward a local optimum of higher fitness. In the second mechanism, the genotype space is thought to decompose into a relatively small number of "neutral subbasins": large connected sets of equal fitness genotypes. In this view, an epoch corresponds to a time period in which the fittest members of the population diffuse through a neutral subbasin under mutation until one of them discovers a (rare) connection to a neutral subbasin of higher fitness. In somewhat different methodology, in the first mechanism the metastability is caused by a "fitness barrier", whereas in the second mechanism it is caused by an "entropy barrier". In this thesis, the evolutionary dynamics in the presence of both fitness and entropy barriers is studied, although the focus is on fitness functions with entropy barriers. For a large class of simple fitness functions we derive, using the maximum entropy methodology, equations of motion on the level of fitness distributions from the underlying microscopic dynamics of selection and mutation on genotypes. In the "thermodynamic limit" of infinite population sizes the population follows these equations of motion precisely while for finite populations the population only follows these equations of motion on average at each time step. From this formulation of the finite population dynamics we determine explicitly the locations of epochs in the space of fitness distributions, the stochastic dynamics within and between the epochs, the average durations of epochs, and the stability of epochs. The results also bear directly on the dynamics of evolutionary search algorithms such as genetic algorithms. In two chapters of this thesis, our mathematical model is used to derive optimal parameter settings for evolutionary search on a wide class of fitness functions. The analysis suggests that optimal evolutionary search occurs in a parameter regime where the highest fitness strings in the population are only marginally stable. We also studied in detail the way in which an evolving population will spread through a neutral subbasin or neutral network. As the theory shows, the limit distribution of the population over a neutral network is independent of almost all evolutionary parameters and is determined solely by the topology of the neutral network. Additionally, this distribution is concentrated at those areas of the neutral network where the "neutrality" is largest. This implies that under neutral evolution the population evolves to increase its "mutational robustness" and that the mutational robustness that it attains is only dependent on the topology of the neutral network on which the population evolves. Finally, we studied and compared the barrier crossing times for fitness and entropy barriers. The results show that entropy barrier crossing in evolving population takes place on much shorter time scales than fitness barrier crossing. This suggests that in evolution, the escape from a metastable state is most likely to occur along neutral paths in genotype space.

Consideration of the eigenproblem of the synaptic matrix in Hopfield's model of neural networks suggests to introduce a matrix built up from an orthogonal set, orthogonal to the original memories. With this new scheme, capacity storage is significantly enhanced and robustness at least conserved.