Most existing autonomous driving perception approaches rely on the Direct perception method with camera sensors, yet they often overlook the valuable 3D spatial data provided by other sensors, such as LiDAR. This Master thesis investigates enhancing affordance learning through a multimodal fusion transformer, aiming to refine AV perception and scene interpretation by effectively integrating multi-sensor data. Our approach introduces a two-stage network architecture: the first stage employs a backbone to fuse sensor data and to extract features, while the second stage employs a Taskblock MLP network to predict both classification affordances (junction, red light, pedestrian, and vehicle hazards) and regression affordances (relative angle, lateral distance, and target vehicle distance). We utilized the TransFuser backbone, based on Imitation Learning, to integrate image and LiDAR BEV data using a self-attention mechanism and to extract the feature map. Our results are compared against image-only architectures like Latent TransFuser and other sensor fusion backbones. Integration with the OmniOpt 2 tool, developed by ScaDS.AI, facilitates hyperparameter optimization, enhancing the model performance. We assessed our model's effectiveness using the CARLA Town02 and as well as the real-world KITTI-360 datasets, demonstrating significant improvements in affordance prediction accuracy and reliability. This advancement underscores the potential of combining LiDAR and image data via transformer-based fusion to create safer and more efficient autonomous driving systems.:List of Figures ix List of Tables xi Abbreviations xiii 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Autonmous Driving: Overview . . . . . . . . . . . . . . . . . . . . . . . 1 1.1.1 From highly automated to autonomous . . . . . . . . . . . . . . 1 1.1.2 Autonomy levels . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.1.3 Perception systems . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Three Paradigms for autonomous driving . . . . . . . . . . . . . . . . . 4 1.3 Sensor Fusion: Global context capture . . . . . . . . . . . . . . . . . . . 5 1.4 Research Questions and Methods . . . . . . . . . . . . . . . . . . . . . . 5 1.4.1 Research Questions (RQ) . . . . . . . . . . . . . . . . . . . . . . 5 1.4.2 Research Methods (RM) . . . . . . . . . . . . . . . . . . . . . . . 6 1.5 Structure of the work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2 Research Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.1 Affordance Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.2 Multi-Modal Autonomous Driving . . . . . . . . . . . . . . . . . . . . . 9 2.3 Sensor Fusion Methods for Object Detection and Motion Forecasting . . 10 2.4 Attention for Autonomous Driving . . . . . . . . . . . . . . . . . . . . . 11 3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3.1 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3.1.1 Problem setting A . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3.1.2 Problem setting B . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.2 Input and Output parametrization . . . . . . . . . . . . . . . . . . . . . 15 3.2.1 Input Representation . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.2.2 Output Representation . . . . . . . . . . . . . . . . . . . . . . . . 18 3.3 Definition of affordances . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.4 Proposed Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.5 Detailed overview of the Proposed Architecture . . . . . . . . . . . . . . 20 3.5.1 Stage1: TransFuser Backbone - Multimodal fusion transformer . 21 3.5.2 Fused Feature extraction . . . . . . . . . . . . . . . . . . . . . . 23 3.5.3 Annotations extraction . . . . . . . . . . . . . . . . . . . . . . . 24 3.5.4 Stage2: Task-Block MLP Network architecture . . . . . . . . . . 29 3.6 Loss Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 3.6.1 Stage1: Loss Function . . . . . . . . . . . . . . . . . . . . . . . . 30 3.6.2 Stage2: Loss Function . . . . . . . . . . . . . . . . . . . . . . . . 31 3.6.3 Total Loss Function . . . . . . . . . . . . . . . . . . . . . . . . . 31 3.7 Other Backbone Architectures . . . . . . . . . . . . . . . . . . . . . . . . 32 3.7.1 Latent TransFuser . . . . . . . . . . . . . . . . . . . . . . . . . . 32 3.7.2 Geometric Fusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.7.3 Late Fusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.8 Hyperparameter Optimization: OmniOpt 2 . . . . . . . . . . . . . . . . 34 4 Training and Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 4.1 Dataset definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 4.1.1 Types of Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 4.1.2 Overview of Dataset Distribution . . . . . . . . . . . . . . . . . . 36 4.2 Implementation Details . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.3 Training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 4.3.1 Stage 1: Backbone architecture training . . . . . . . . . . . . . . 38 4.3.2 Stage 2: TaskBlock MLP training . . . . . . . . . . . . . . . . . 39 4.3.3 Traning Parameter Study . . . . . . . . . . . . . . . . . . . . . . 41 4.4 Loss curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.4.1 Stage 1 Loss curve . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.4.2 Stage 2 Loss curve . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.5 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 4.5.1 Preparation of a optimization project . . . . . . . . . . . . . . . 43 5 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 5.1 Quantitative Insights into Regression-Based Affordance Predictions . . . 45 5.1.1 Comparative Analysis of Error Metrics against each Backbone . 45 5.1.2 Graphical Analysis of error metrics performance for Transfuser . 47 5.2 Quantitative Insights into Classification-Based Affordance Predictions . 48 5.2.1 Comparative Analysis of Classification Performance Metrics against each Backbone . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 5.2.2 Graphical Analysis of classification performance for TransFuser . 50 5.3 OmniOpt2 Hyper-optimization results . . . . . . . . . . . . . . . . . . . 52 5.4 Affordance Prediction Dashboard . . . . . . . . . . . . . . . . . . . . . . 53 6 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 6.1 Evaluation with CARLA Test dataset . . . . . . . . . . . . . . . . . . . 55 6.1.1 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 6.2 Evaluation with real world: The KITTI Dataset . . . . . . . . . . . . . 56 6.2.1 Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 6.2.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 A Ablation Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 A.1 Latent Transfuser with MLP . . . . . . . . . . . . . . . . . . . . . . . . 61 A.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 A.2.1 Comparative Analysis of Error Metrics in Latent Transfuser with Transformer and MLP . . . . . . . . . . . . . . . . . . . . . . . . 61 A.2.2 Comparative Analysis of Classification Performance Metrics in Latent Transfuser with Transformer and MLP . . . . . . . . . . 62

Submitted by Caroline Xavier (caroline.xavier@pucrs.br) on 2017-06-30T13:59:55Z No. of bitstreams: 1 DIS_GICELDA_JULIA_DAL_BO_COMPLETO.pdf: 3328714 bytes, checksum: c1dae6f11752d09fdfaddc8e2f46a7ce (MD5) / Made available in DSpace on 2017-06-30T13:59:55Z (GMT). No. of bitstreams: 1 DIS_GICELDA_JULIA_DAL_BO_COMPLETO.pdf: 3328714 bytes, checksum: c1dae6f11752d09fdfaddc8e2f46a7ce (MD5) Previous issue date: 2017-03-09 / This dissertation explores structures of power from the point of view of empowerment and resistance in a shared economy model, having Uber (since its arrival in the country) as object of study. To this end, it relies on a qualitative method of interpretativist approach through a single case study. By identifying the agents' discourses of the urban mobility socio-technical system, in the voice of media vehicles and specialized websites, this research proposes to analyze how power structures are organized. The study lists three perspectives, an ?economic model?, a ?business model? and ?sustainable development?, built to group the ?two sides of the same coin?, empowerment and resistance, proposing a ?conversation? between them. It presents a view of the relationship with the collectivity through two mechanisms, also of opposing forces: ?quest for legitimacy? and ?reinforcement to the status quo?. In a complementary way, it proposes a discussion section about uberism and its variations. Among the aspects listed in this section, issues of regulation and characteristics of the so-called Uber economy stand out. It brings, as theoretical basis, concepts of the Transition Theory for Sustainable Development, the Multiple Level Perspective (MLP) and an applied study that links sharing economy to the notion of socio-technical systems. With this, it seeks to demonstrate the ?game? of forces between empowerment and resistance in the Brazilian context of the socio-technical system of urban mobility. A combination of technological trajectories that have been destabilizing and pressuring the existing socio-technical structures pointed out for the existence of transitions to sustainable development. / Esta disserta??o explora estruturas de poder sob o ponto de vista de empoderamento e de resist?ncia em um modelo de economia compartilhada, tendo a Uber (desde sua chegada ao pa?s) como objeto de estudo. Para tal, apoia-se em um m?todo qualitativo de abordagem interpretativista por meio de um estudo de caso ?nico. Prop?e-se a analisar como est?o organizadas as estruturas de poder identificando os discursos de agentes do sistema sociotecnico de mobilidade urbana na voz dos ve?culos de m?dia e de sites especializados. O estudo elenca tr?s perspectivas, modelo econ?mico, modelo de neg?cio e desenvolvimento sustent?vel, constru?das de forma a agrupar os ?dois lados da mesma moeda?, empoderamento e resist?ncia, propondo uma ?conversa? entre eles. Apresenta um olhar da rela??o com a coletividade atrav?s de dois mecanismos, tamb?m de for?as contr?rias: ?busca por legitimidade? e ?refor?o ao status quo?. De forma complementar, prop?e uma se??o de discuss?o acerca de uberismo e suas varia??es. Dentre os aspectos elencados nesta se??o, destacam-se quest?es de regula??o e caracter?sticas da chamada Uber economia. Como suporte te?rico apresenta conceitos da Teoria das Transi??es para o Desenvolvimento Sustent?vel, da Perspectiva de M?ltiplos N?veis (MLP) e de um estudo aplicado que vincula economia compartilhada ? no??o de sistema sociot?cnico. Com isso, busca demonstrar o ?jogo? de for?as entre empoderamento e resist?ncia no contexto brasileiro do sistema sociot?cnico de mobilidade urbana. Os resultados apontaram em dire??o a exist?ncia de transi??es para o desenvolvimento sustent?vel, permeadas por uma combina??o de trajet?rias que tem desestabilizado e pressionado as estruturas sociot?cnicas vigentes.

Safety and reliability are prime concerns in aircraft performance due to the involved costs and risk to lives. Despite the best efforts in design methodology, quality evaluation in production and structural integrity assessment in-service, attainment of one hundred percent safety through development and use of a suitable in-flight health monitoring system is still a farfetched goal. And, evolution of such a system requires, first, identification of an appropriate Technique and next its adoption to meet the challenges posed by newer materials (advanced composites), complex structures and the flight environment. In fact, a quick survey of the available Non-Destructive Evaluation (NDE) techniques suggests Acoustic Emission (AE) as the only available method. High merit in itself could be a weakness - Noise is the worst enemy of AE. So, while difficulties are posed due to the insufficient understanding of the basic behavior of composites, growth and interaction of defects and damage under a specified load condition, high in-flight noise further complicates the issue making the developmental task apparently formidable and challenging. Development of an in-flight monitoring system based on AE to function as an early warning system needs addressing three aspects, viz., the first, discrimination of AE signals from noise data, the second, extraction of required information from AE signals for identification of sources (source characterization) and quantification of its growth, and the third, automation of the entire process. And, a quick assessment of the aspects involved suggests that Artificial Neural Networks (ANN) are ideally suited for solving such a complex problem. A review of the available open literature while indicates a number of investigations carried out using noise elimination and source characterization methods such as frequency filtering and statistical pattern recognition but shows only sporadic attempts using ANN. This may probably be due to the complex nature of the problem involving investigation of a large number of influencing parameters, amount of effort and time to be invested, and facilities required and multi-disciplinary nature of the problem. Hence as stated in the foregoing, the need for such a study cannot be over emphasized. Thus, this thesis is an attempt addressing the issue of analysis and automation of complex sets of AE data such as AE signals mixed with in-flight noise thus forming the first step towards in-flight monitoring using AE. An ANN can in fact replace the traditional algorithmic approaches used in the past. ANN in general are model free estimators and derive their computational efficiency due to large connectivity, massive parallelism, non-linear analog response and learning capabilities. They are better suited than the conventional methods (statistical pattern recognition methods) due to their characteristics such as classification, pattern matching, learning, generalization, fault tolerance and distributed memory and their ability to process unstructured data sets which may be carrying incomplete information at times and hence chosen as the tool. Further, in the current context, the set of investigations undertaken were in the absence of sufficient a priori information and hence clustering of signals generated by AE sources through self-organizing maps is more appropriate. Thus, in the investigations carried out under the scope of this thesis, at first a hybrid network named "NAEDA" (Neural network for Acoustic Emission Data Analysis) using Kohonen self-organizing feature map (KSOM) and multi-layer perceptron (MLP) that learns on back propagation learning rule was specifically developed with innovative data processing techniques built into the network. However, for accurate pattern recognition, multi-layer back propagation NN needed to be trained with source and noise clusters as input data. Thus, in addition to optimizing the network architecture and training parameters, preprocessing of input data to the network and multi-class clustering and classification proved to be the corner stones in obtaining excellent identification accuracy. Next, in-flight noise environment of an aircraft was generated off line through carefully designed simulation experiments carried out in the laboratory (Ex: EMI, friction, fretting and other mechanical and hydraulic phenomena) based on the in-flight noise survey carried out by earlier investigators. From these experiments data was acquired and classified into their respective classes through MLP. Further, these noises were mixed together and clustered through KSOM and then classified into their respective clusters through MLP resulting in an accuracy of 95%- 100% Subsequently, to evaluate the utility of NAEDA for source classification and characterization, carbon fiber reinforced plastic (CFRP) specimens were subjected to spectrum loading simulating typical in-flight load and AE signals were acquired continuously up to a maximum of three designed lives and in some cases up to failure. Further, AE signals with similar characteristics were grouped into individual clusters through self-organizing map and labeled as belonging to appropriate failure modes, there by generating the class configuration. Then MLP was trained with this class information, which resulted in automatic identification and classification of failure modes with an accuracy of 95% - 100%. In addition, extraneous noise generated during the experiments was acquired and classified so as to evaluate the presence or absence of such data in the AE data acquired from the CFRP specimens. In the next stage, noise and signals were mixed together at random and were reclassified into their respective classes through supervised training of multi-layer back propagation NN. Initially only noise was discriminated from the AE signals from CFRP failure modes and subsequently both noise discrimination and failure mode identification and classification was carried out resulting in an accuracy of 95% - 100% in most of the cases. Further, extraneous signals mentioned above were classified which indicated the presence of such signals in the AE signals obtained from the CFRP specimen. Thus, having established the basis for noise identification and AE source classification and characterization, two specific examples were considered to evaluate the utility and efficiency of NAEDA. In the first, with the postulation that different basic failure modes in composites have unique AE signatures, the difference in damage generation and progression can be clearly characterized under different loading conditions. To examine this, static compression tests were conducted on a different set of CFRP specimens till failure with continuous AE monitoring and the resulting AE signals were classified through already trained NAEDA. The results obtained shows that the total number of signals obtained were very less when compared to fatigue tests and the specimens failed with hardly any damage growth. Further, NAEDA was able to discriminate the"noise and failure modes in CFRP specimen with the same degree of accuracy with which it has classified such signals obtained from fatigue tests. In the second example, with the same postulate of unique AE signatures for different failure modes, the differences in the complexion of the damage growth and progression should become clearly evident when one considers specimens with different lay up sequences. To examine this, the data was reclassified on the basis of differences in lay up sequences from specimens subjected to fatigue. The results obtained clearly confirmed the postulation. As can be seen from the summary of the work presented in the foregoing paragraphs, the investigations undertaken within the scope of this thesis involve elaborate experimentation, development of tools, acquisition of extensive data and analysis. Never the less, the results obtained were commensurate with the efforts and have been fruitful. Of the useful results that have been obtained, to state in specific, the first is, discrimination of simulated noise sources achieved with significant success but for some overlapping which is not of major concern as far as noises are concerned. Therefore they are grouped into required number of clusters so as to achieve better classification through supervised NN. This proved to be an innovative measure in supervised classification through back propagation NN. The second is the damage characterization in CFRP specimens, which involved imaginative data processing techniques that proved their worth in terms of optimization of various training parameters and resulted in accurate identification through clustering. Labeling of clusters is made possible by marking each signal starting from clustering to final classification through supervised neural network and is achieved through phenomenological correlation combined with ultrasonic imaging. Most rewarding of all is the identification of failure modes (AE signals) mixed in noise into their respective classes. This is a direct consequence of innovative data processing, multi-class clustering and flexibility of grouping various noise signals into suitable number of clusters. Thus, the results obtained and presented in this thesis on NN approach to AE signal analysis clearly establishes the fact that methods and procedures developed can automate detection and identification of failure modes in CFRP composites under hostile environment, which could lead to the development of an in-flight monitoring system.

Οι σύγχρονες τάσεις της Επιστήμης Σχεδιασμού των Υπολογιστικών Συστημάτων έχουν υιοθετήσει την χρήση των Κρυφών Μνημών ή Μνημών Cache, αποβλέποντας στην απόκρυψη της Καθυστέρησης της Κύριας Μνήμης των Συστημάτων (Memory Latency) και την γεφύρωση του χάσματος της απόδοσης του Επεξεργαστή και της Κύριας Μνήμης (Processor – Memory Performance Gap). Οι Μνήμες Cache έτσι έχουν αποκτήσει αδιαμφισβήτητα πρωτεύοντα ρόλο στην Ιεραρχία Μνήμης των Ηλεκτρονικών Υπολογιστών. Οι νέες τάσεις Σχεδιασμού ανέδειξαν την Έννοια του Παραλληλισμού σε πρωτεύοντα ρόλο. Αρχικά διερευνήθηκε ο Παραλληλισμός Επιπέδου Εντολών, ωστόσο η αύξηση της Απόδοσης των Υπολογιστών σύντομα έφτασε ένα μέγιστο. Την τελευταία δεκαετία το κέντρο του ενδιαφέροντος των σχεδιαστών έχει και πάλι μετατοπιστεί, καθώς ένας νέος τύπος Επεξεργαστών έχει εισέλθει στο προσκήνιο, οι Πολυπύρηνοι Επεξεργαστές, ή όπως είναι αλλιώς γνωστοί on-chip Multiprocessors (CMP). Αυτές οι εξελίξεις, σε συνδυασμό με την ολοένα αυξανόμενη πολυπλοκότητα της “συμπεριφοράς” των εκτελούμενων Εφαρμογών, ώθησαν το σχεδιαστικό ενδιαφέρον προς την εκμετάλλευση ενός νεοσύστατου τύπου Παραλληλισμού. Ο Παραλληλισμός Επιπέδου Μνήμης ή Memory Level Parallelism (MLP) αποτελεί τα τελευταία χρόνια, το πλέον ισχυρό μέσο αύξησης της απόδοσης των Υπολογιστικών Συστημάτων και μαζί με τους Πολυπύρηνους Επεξεργαστές θα κυριαρχήσει στο προσκήνιο των εξελίξεων τα επόμενα χρόνια. Σκοπός της παρούσας Διπλωματικής Εργασίας είναι η ανάπτυξη ενός Στατιστικού – Πιθανοτικού Μοντέλου για μελέτη και πρόβλεψη των φαινομένων που αναπτύσσονται σε Μνήμες Cache, στις οποίες αποθηκεύονται δεδομένα από εκτελούμενες Εφαρμογές, με έντονο Παραλληλισμό Επιπέδου Μνήμης. Θα οριστεί ένας Εκτιμητής του Φόρτου που επιβάλλεται στο Σύστημα, από φαινόμενα Παραλληλισμού Επιπέδου Μνήμης (MLP). Στην συνέχεια, με βάση το Μοντέλο που αναπτύσσουμε, θα διερευνηθεί ένα ικανοποιητικό σύνολο Εφαρμογών, και θα εξαχθεί μια Εκτίμηση – Πρόβλεψη για τον Φόρτο (MLP) του Συστήματος. Εφόσον οι Προβλέψεις μας κριθούν επιτυχής, το Μοντέλο Πρόβλεψης Φόρτου MLP που αναπτύξαμε, μπορεί να αποτελέσει χρήσιμο Εργαλείο στα χέρια των Σχεδιαστών που ασχολούνται με την αύξηση της Απόδοσης των Σύγχρονων Υπολογιστικών Συστημάτων. / -

O principal objetivo do presente trabalho é propor, desenvolver, testar e apresentar um método para a classificação do grau de desenvolvimento da capacidade cognitiva espacial de diferentes indivíduos. 37 alunos de graduação tiveram seus eletroencefalogramas (EEGs) capturados enquanto estavam engajados em tarefas de rotação mental de imagens tridimensionais. Seu grau de desenvolvimento da capacidade cognitiva espacial foi avaliado utilizando-se um teste psicológico BPR-5. O maior expoente de Lyapunov (LLE) foi calculado a partir de cada um dos 8 canais dos EEGs capturados. OS LLEs foram então utilizados como tuplas de entrada para 5 diferentes classificadores: i) perceptron de múltiplas camadas, ii) rede neural artificial de funções de base radial, iii) perceptron votado, iv) máquinas de vetor de suporte, e v) k-vizinhos. O melhor resultado foi obtido utilizando-se uma RBF com 4 clusters e a função de kernel Puk. Também foi realizada uma análise estatística das diferenças de atividade cerebral, baseando-se nos LLEs calculados, entre os dois grupos de interesse: SI+ (indivíduos com um suposto maior grau de desenvolvimento da sua capacidade cognitiva espacial) e SI- (grupo de controle) durante a realização de tarefas de rotação mental de imagens tridimensionais. Uma diferença média de 16% foi encontrada entre os dois grupos. O método de classificação proposto pode vir a contribuir e a interagir com outros processos na analise e no estudo da capacidade cognitiva espacial humana, assim como no entendimento da inteligência humana como um todo. Um melhor entendimento e avaliação das capacidades cognitivas de um indivíduo podem sugerir a este elementos de motivação, facilidade ou de inclinações naturais suas, podendo, provavelmente, afetar as decisões da sua vida e carreira de uma forma positiva. / The main objective of the present work is to propose, develop, test, and show a method for classifying the spatial cognition degree of development on different individuals. Thirty-Seven undergraduate students had their electroencephalogram (EEG) recorded while engaged in 3-D images mental rotation tasks. Their spatial cognition degree of development was evaluated using a BPR-5 psychological test. The Largest Lyapunov Exponent (LLE) was calculated from each of the 8 electrodes recorded in each EEG. The LLEs were used as input for five different classifiers: i) multi-layer perceptron artificial neural network, ii) radial base functions artificial neural network, iii) voted perceptron artificial neural network, iv) support vector machines, and v) K-Nearest Neighbors. The best result was achieved by using a RBF with 4 clusters and Puk kernel function. Also a statistical analysis of the brain activity, based in the calculated LLEs, differences between two interest groups: SI+ (participants with an alleged higher degree of development of their spatial cognition) and SI- (control group) during the performing of mental rotation of tridimensional images tasks was done.. An average difference of 16% was found between both groups. The proposed classification method can contribute and interact with other processes in the analysis and study of human spatial cognition, as in the understanding of the human intelligence at all. A better understanding and evaluation of the cognitive capabilities of an individual could suggest him elements of motivation, ease or natural inclinations, possibly affecting the decisions of his life and carrier positively.

O principal objetivo do presente trabalho é propor, desenvolver, testar e apresentar um método para a classificação do grau de desenvolvimento da capacidade cognitiva espacial de diferentes indivíduos. 37 alunos de graduação tiveram seus eletroencefalogramas (EEGs) capturados enquanto estavam engajados em tarefas de rotação mental de imagens tridimensionais. Seu grau de desenvolvimento da capacidade cognitiva espacial foi avaliado utilizando-se um teste psicológico BPR-5. O maior expoente de Lyapunov (LLE) foi calculado a partir de cada um dos 8 canais dos EEGs capturados. OS LLEs foram então utilizados como tuplas de entrada para 5 diferentes classificadores: i) perceptron de múltiplas camadas, ii) rede neural artificial de funções de base radial, iii) perceptron votado, iv) máquinas de vetor de suporte, e v) k-vizinhos. O melhor resultado foi obtido utilizando-se uma RBF com 4 clusters e a função de kernel Puk. Também foi realizada uma análise estatística das diferenças de atividade cerebral, baseando-se nos LLEs calculados, entre os dois grupos de interesse: SI+ (indivíduos com um suposto maior grau de desenvolvimento da sua capacidade cognitiva espacial) e SI- (grupo de controle) durante a realização de tarefas de rotação mental de imagens tridimensionais. Uma diferença média de 16% foi encontrada entre os dois grupos. O método de classificação proposto pode vir a contribuir e a interagir com outros processos na analise e no estudo da capacidade cognitiva espacial humana, assim como no entendimento da inteligência humana como um todo. Um melhor entendimento e avaliação das capacidades cognitivas de um indivíduo podem sugerir a este elementos de motivação, facilidade ou de inclinações naturais suas, podendo, provavelmente, afetar as decisões da sua vida e carreira de uma forma positiva. / The main objective of the present work is to propose, develop, test, and show a method for classifying the spatial cognition degree of development on different individuals. Thirty-Seven undergraduate students had their electroencephalogram (EEG) recorded while engaged in 3-D images mental rotation tasks. Their spatial cognition degree of development was evaluated using a BPR-5 psychological test. The Largest Lyapunov Exponent (LLE) was calculated from each of the 8 electrodes recorded in each EEG. The LLEs were used as input for five different classifiers: i) multi-layer perceptron artificial neural network, ii) radial base functions artificial neural network, iii) voted perceptron artificial neural network, iv) support vector machines, and v) K-Nearest Neighbors. The best result was achieved by using a RBF with 4 clusters and Puk kernel function. Also a statistical analysis of the brain activity, based in the calculated LLEs, differences between two interest groups: SI+ (participants with an alleged higher degree of development of their spatial cognition) and SI- (control group) during the performing of mental rotation of tridimensional images tasks was done.. An average difference of 16% was found between both groups. The proposed classification method can contribute and interact with other processes in the analysis and study of human spatial cognition, as in the understanding of the human intelligence at all. A better understanding and evaluation of the cognitive capabilities of an individual could suggest him elements of motivation, ease or natural inclinations, possibly affecting the decisions of his life and carrier positively.

O principal objetivo do presente trabalho é propor, desenvolver, testar e apresentar um método para a classificação do grau de desenvolvimento da capacidade cognitiva espacial de diferentes indivíduos. 37 alunos de graduação tiveram seus eletroencefalogramas (EEGs) capturados enquanto estavam engajados em tarefas de rotação mental de imagens tridimensionais. Seu grau de desenvolvimento da capacidade cognitiva espacial foi avaliado utilizando-se um teste psicológico BPR-5. O maior expoente de Lyapunov (LLE) foi calculado a partir de cada um dos 8 canais dos EEGs capturados. OS LLEs foram então utilizados como tuplas de entrada para 5 diferentes classificadores: i) perceptron de múltiplas camadas, ii) rede neural artificial de funções de base radial, iii) perceptron votado, iv) máquinas de vetor de suporte, e v) k-vizinhos. O melhor resultado foi obtido utilizando-se uma RBF com 4 clusters e a função de kernel Puk. Também foi realizada uma análise estatística das diferenças de atividade cerebral, baseando-se nos LLEs calculados, entre os dois grupos de interesse: SI+ (indivíduos com um suposto maior grau de desenvolvimento da sua capacidade cognitiva espacial) e SI- (grupo de controle) durante a realização de tarefas de rotação mental de imagens tridimensionais. Uma diferença média de 16% foi encontrada entre os dois grupos. O método de classificação proposto pode vir a contribuir e a interagir com outros processos na analise e no estudo da capacidade cognitiva espacial humana, assim como no entendimento da inteligência humana como um todo. Um melhor entendimento e avaliação das capacidades cognitivas de um indivíduo podem sugerir a este elementos de motivação, facilidade ou de inclinações naturais suas, podendo, provavelmente, afetar as decisões da sua vida e carreira de uma forma positiva. / The main objective of the present work is to propose, develop, test, and show a method for classifying the spatial cognition degree of development on different individuals. Thirty-Seven undergraduate students had their electroencephalogram (EEG) recorded while engaged in 3-D images mental rotation tasks. Their spatial cognition degree of development was evaluated using a BPR-5 psychological test. The Largest Lyapunov Exponent (LLE) was calculated from each of the 8 electrodes recorded in each EEG. The LLEs were used as input for five different classifiers: i) multi-layer perceptron artificial neural network, ii) radial base functions artificial neural network, iii) voted perceptron artificial neural network, iv) support vector machines, and v) K-Nearest Neighbors. The best result was achieved by using a RBF with 4 clusters and Puk kernel function. Also a statistical analysis of the brain activity, based in the calculated LLEs, differences between two interest groups: SI+ (participants with an alleged higher degree of development of their spatial cognition) and SI- (control group) during the performing of mental rotation of tridimensional images tasks was done.. An average difference of 16% was found between both groups. The proposed classification method can contribute and interact with other processes in the analysis and study of human spatial cognition, as in the understanding of the human intelligence at all. A better understanding and evaluation of the cognitive capabilities of an individual could suggest him elements of motivation, ease or natural inclinations, possibly affecting the decisions of his life and carrier positively.

Made available in DSpace on 2014-06-12T15:56:06Z (GMT). No. of bitstreams: 2 arquivo2738_1.pdf: 2219821 bytes, checksum: e0060e817bd6a925ad67e0971641acff (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2011 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Classificadores têm sido largamente aplicados nos mais diversos campos científicos e industriais, em geral obtendo bons desempenhos. Entretanto, quando aplicados a problemas cuja quantidade de dados disponível para o treinamento é limitada (bases de dados pequenas) ou quando estes dados apresentam um desbalanceamento entre as classes (bases de dados desbalanceadas), a maioria dos classificadores obtém um desempenho pobre. O poder de generalização do classificador é reduzido quando bases de dados pequenas são utilizadas durante o processo de treinamento, enquanto que em bases de dados desbalanceadas, as classes com maior representatividade e menor importância tendem a ser favorecidas. Inerentes a diversos problemas do mundo real, conjuntos de dados pequenos e desbalanceados representam uma limitação a ser superada por algoritmos de aprendizagem para produção de classificadores precisos e confiáveis. Neste trabalho é proposta uma abordagem baseada na adição de ruído Gaussiano durante o processo de treinamento de uma rede neural MultiLayer Perceptron (MLP) com o intuito de contornar as limitações referentes às bases de dados pequenas e/ou desbalanceadas, possibilitando a rede neural obter um alto poder de generalização A metodologia proposta pode ser dividida em duas etapas principais. Na primeira, um estudo acerca da correlação entre as variáveis é realizado. Este estudo envolve avaliar a correlação entre as variáveis por meio do coeficiente de correlação de Pearson e a descorrelação das variáveis através do método Análise de Componentes Principais (ACP). Na segunda, ruídos derivados a partir de uma distribuição Gaussiana são inseridos nas variáveis de entrada. Para validar a abordagem proposta foram utilizadas três bases públicas de um conhecido benchmark da comunidade de redes neurais, Proben1. Os resultados experimentais indicam que a abordagem proposta obtém um desempenho estatisticamente melhor (95% de confiança) que o método de treinamento convencional, principalmente quando utilizado o método PCA para descorrelação das variáveis antes da aplicação de ruído

Made available in DSpace on 2014-06-12T15:58:28Z (GMT). No. of bitstreams: 2 arquivo4572_1.pdf: 1149011 bytes, checksum: 92aae8f6f9b5145bfcecb94d96dbbc0b (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2004 / Instituto Federal de Educação, Ciência e Tecnologia de Pernambuco / Um nariz artificial é um sistema modular composto de duas partes principais: um sistema sensor, formado de elementos que detectam odores e um sistema de reconhecimento de padrões que classifica os odores detectados. Redes neurais artificiais têm sido utilizadas como sistema de reconhecimento de padrões para narizes artificiais e vêm apresentando resultados promissores. Desde os anos 80, pesquisas para criação de narizes artificiais, que permitam detectar e classificar odores, vapores e gases automaticamente, têm tido avanços significativos. Esses equipamentos podem ser utilizados no monitoramento ambiental para controlar a qualidade do ar, na área de saúde para realizar diagnóstico de doenças e nas indústrias de alimentos para o controle de qualidade e o monitoramento de processos de produção. Esta dissertação investiga a utilização de quatro técnicas diferentes de redes neurais para criação de sistemas de reconhecimento de padrões em narizes artificiais. O trabalho está dividido em quatro partes principais: (1) introdução aos narizes artificiais, (2) redes neurais artificiais para sistema de reconhecimento de padrões, (3) métodos para medir o desempenho de sistemas de reconhecimento de padrões e comparar os resultados e (4) estudo de caso. Os dados utilizados para o estudo de caso, foram obtidos por um protótipo de nariz artificial composto por um arranjo de oito sensores de polímeros condutores, expostos a nove tipos diferentes de aguarrás. Foram adotadas as técnicas Multi-Layer Perceptron (MLP), Radial Base Function (RBF), Probabilistic Neural Network (PNN) e Time Delay Neural Network (TDNN) para criar os sistemas de reconhecimento de padrões. A técnica PNN foi investigada em detalhes, por dois motivos principais: esta técnica é indicada para realização de tarefas de classificação e seu treinamento é feito em apenas um passo, o que torna a etapa de criação dessas redes muito rápida. Os resultados foram comparados através dos valores dos erros médios de classificação utilizando o método estatístico de Teste de Hipóteses. As redes PNN correspondem a uma nova abordagem para criação de sistemas de reconhecimento de padrões de odor. Estas redes tiveram um erro médio de classificação de 1.1574% no conjunto de teste. Este foi o menor erro obtido entre todos os sistemas criados, entretanto mesmo com o menor erro médio de classificação, os testes de hipóteses mostraram que os classificadores criados com PNN não eram melhores do que os classificadores criados com a arquitetura RBF, que obtiveram um erro médio de classificação de 1.3889%. A grande vantagem de criar classificadores com a arquitetura PNN foi o pequeno tempo de treinamento dos mesmos, chegando a ser quase imediato. Porém a quantidade de nodos na camada escondida foi muito grande, o que pode ser um problema, caso o sistema criado deva ser utilizado em equipamentos com poucos recursos computacionais. Outra vantagem de criar classificadores com redes PNN é relativa à quantidade reduzida de parâmetros que devem ser analisados, neste caso apenas o parâmetro relativo à largura da função Gaussiana precisou ser investigado

Made available in DSpace on 2014-06-12T16:01:05Z (GMT). No. of bitstreams: 2 arquivo7154_1.pdf: 1370997 bytes, checksum: 1580b7b5979343826e4d0a3b88b57dac (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2005 / Esta dissertação propõe modificações na metodologia yamazaki para a otimização simultânea de arquiteturas e pesos de redes Multilayer Perceptron (MLP). O objetivo principal é propô-las em conjunto com as respectivas validações, visando tornar mais eficiente o processo de otimização. A base para o algoritmo híbrido de otimização são os algoritmos simulated annealing, tabu search e a metodologia yamazaki. As modificações são realizadas nos critérios de implementação tais como mecanismo de geração de vizinhança, esquema de esfriamento e função de custo. Um dos pontos principais desta dissertação é a criação de um novo mecanismo de geração de vizinhança visando aumentar o espaço de busca. O esquema de esfriamento é de grande importância na convergência do algoritmo. O custo de cada solução é medido como média ponderada entre o erro de classificação para o conjunto de treinamento e a porcentagem de conexões utilizadas pela rede. As bases de dados utilizadas nos experimentos são: classificação de odores provenientes de três safras de um mesmo vinho e classificação de gases. A fundamentação estatística para as conclusões observadas através dos resultados obtidos é realizada usando teste de hipóteses. Foi realizado um estudo do tempo de execução separando as fases de otimização global da fase de refinamento local. Concluiu-se que com o novo mecanismo de geração de vizinhança fez desnecessário o uso do backpropagation obtendo assim um alto ganho em tempo de execução. O algoritmo híbrido de otimização apresentou, para ambas as bases de dados, o menor valor da média do erro de classificação do conjunto de teste e o menor valor da quantidade de conexões. Além disso, o tempo de execução foi reduzido em média 46.72%