Fast 2D/3D object representation with growing neural gas

This work presents the design of a real-time system to model visual objects with the use of self-organising networks. The architecture of the system addresses multiple computer vision tasks such as image segmentation, optimal parameter estimation and object representation. We first develop a framework for building non-rigid shapes using the growth mechanism of the self-organising maps, and then we define an optimal number of nodes without overfitting or underfitting the network based on the knowledge obtained from information-theoretic considerations. We present experimental results for hands and faces, and we quantitatively evaluate the matching capabilities of the proposed method with the topographic product. The proposed method is easily extensible to 3D objects, as it offers similar features for efficient mesh reconstruction.

Combining Taxonomies using Word2vec

Taxonomies have gained a broad usage in a variety of fields due to their extensibility, as well as their use for classification and knowledge organization. Of particular interest is the digital document management domain in which their hierarchical structure can be effectively employed in order to organize documents into content-specific categories. Common or standard taxonomies (e.g., the ACM Computing Classification System) contain concepts that are too general for conceptualizing specific knowledge domains. In this paper we introduce a novel automated approach that combines sub-trees from general taxonomies with specialized seed taxonomies by using specific Natural Language Processing techniques. We provide an extensible and generalizable model for combining taxonomies in the practical context of two very large European research projects. Because the manual combination of taxonomies by domain experts is a highly time consuming task, our model measures the semantic relatedness between concept labels in CBOW or skip-gram Word2vec vector spaces. A preliminary quantitative evaluation of the resulting taxonomies is performed after applying a greedy algorithm with incremental thresholds used for matching and combining topic labels.

New developments on VCA unmixing algorithm

Hyperspectral sensors are being developed for remote sensing applications. These sensors produce huge data volumes which require faster processing and analysis tools. Vertex component analysis (VCA) has become a very useful tool to unmix hyperspectral data. It has been successfully used to determine endmembers and unmix large hyperspectral data sets without the use of any a priori knowledge of the constituent spectra. Compared with other geometric-based approaches VCA is an efficient method from the computational point of view. In this paper we introduce new developments for VCA: 1) a new signal subspace identification method (HySime) is applied to infer the signal subspace where the data set live. This step also infers the number of endmembers present in the data set; 2) after the projection of the data set onto the signal subspace, the algorithm iteratively projects the data set onto several directions orthogonal to the subspace spanned by the endmembers already determined. The new endmember signature corresponds to these extreme of the projections. The capability of VCA to unmix large hyperspectral scenes (real or simulated), with low computational complexity, is also illustrated.

Datasets, features, learning, and models in visual recognition

Visual recognition is a fundamental research topic in computer vision. This dissertation explores datasets, features, learning, and models used for visual recognition. In order to train visual models and evaluate different recognition algorithms, this dissertation develops an approach to collect object image datasets on web pages using an analysis of text around the image and of image appearance. This method exploits established online knowledge resources (Wikipedia pages for text; Flickr and Caltech data sets for images). The resources provide rich text and object appearance information. This dissertation describes results on two datasets. The first is Berg’s collection of 10 animal categories; on this dataset, we significantly outperform previous approaches. On an additional set of 5 categories, experimental results show the effectiveness of the method. Images are represented as features for visual recognition. This dissertation introduces a text-based image feature and demonstrates that it consistently improves performance on hard object classification problems. The feature is built using an auxiliary dataset of images annotated with tags, downloaded from the Internet. Image tags are noisy. The method obtains the text features of an unannotated image from the tags of its k-nearest neighbors in this auxiliary collection. A visual classifier presented with an object viewed under novel circumstances (say, a new viewing direction) must rely on its visual examples. This text feature may not change, because the auxiliary dataset likely contains a similar picture. While the tags associated with images are noisy, they are more stable when appearance changes. The performance of this feature is tested using PASCAL VOC 2006 and 2007 datasets. This feature performs well; it consistently improves the performance of visual object classifiers, and is particularly effective when the training dataset is small. With more and more collected training data, computational cost becomes a bottleneck, especially when training sophisticated classifiers such as kernelized SVM. This dissertation proposes a fast training algorithm called Stochastic Intersection Kernel Machine (SIKMA). This proposed training method will be useful for many vision problems, as it can produce a kernel classifier that is more accurate than a linear classifier, and can be trained on tens of thousands of examples in two minutes. It processes training examples one by one in a sequence, so memory cost is no longer the bottleneck to process large scale datasets. This dissertation applies this approach to train classifiers of Flickr groups with many group training examples. The resulting Flickr group prediction scores can be used to measure image similarity between two images. Experimental results on the Corel dataset and a PASCAL VOC dataset show the learned Flickr features perform better on image matching, retrieval, and classification than conventional visual features. Visual models are usually trained to best separate positive and negative training examples. However, when recognizing a large number of object categories, there may not be enough training examples for most objects, due to the intrinsic long-tailed distribution of objects in the real world. This dissertation proposes an approach to use comparative object similarity. The key insight is that, given a set of object categories which are similar and a set of categories which are dissimilar, a good object model should respond more strongly to examples from similar categories than to examples from dissimilar categories. This dissertation develops a regularized kernel machine algorithm to use this category dependent similarity regularization. Experiments on hundreds of categories show that our method can make significant improvement for categories with few or even no positive examples.

Validation of AltiKa Matching Pursuit Rain Flag

The major drawback of Ka band, operating frequency of the AltiKa altimeter on board SARAL, is its sensitivity to atmospheric liquid water. Even light rain or heavy clouds can strongly attenuate the signal and distort the signal leading to erroneous geophysical parameters estimates. A good detection of the samples affected by atmospheric liquid water is crucial. As AltiKa operates at a single frequency, a new technique based on the detection by a Matching Pursuit algorithm of short scale variations of the slope of the echo waveform plateau has been developed and implemented prelaunch in the ground segment. As the parameterization of the detection algorithm was defined using Jason-1 data, the parameters were re-estimated during the cal-val phase, during which the algorithm was also updated. The measured sensor signal-to-noise ratio is significantly better than planned, the data loss due to attenuation by rain is significantly smaller than expected (<0.1%). For cycles 2 to 9, the flag detects about 9% of 1Hz data, 5.5% as rainy and 3.5 % as backscatter bloom (or sigma0 bloom). The results of the flagging process are compared to independent rain data from microwave radiometers to evaluate its performances in term of detection and false alarms.

A Hybrid Genetic Algorithm to Solve a Logt-Sizing and Scheduling Problem

Abstract: This paper reports a lot-sizing and scheduling problem, which minimizes inventory and backlog costs on m parallel machines with sequence-dependent set-up times over t periods. Problem solutions are represented as product subsets ordered and/or unordered for each machine m at each period t. The optimal lot sizes are determined applying a linear program. A genetic algorithm searches either over ordered or over unordered subsets (which are implicitly ordered using a fast ATSP-type heuristic) to identify an overall optimal solution. Initial computational results are presented, comparing the speed and solution quality of the ordered and unordered genetic algorithm approaches.

Development of a Real-Time Hierarchical 3D Path Planning Algorithm for Unmanned Aerial Vehicles

Unmanned aerial vehicles (UAVs) frequently operate in partially or entirely unknown environments. As the vehicle traverses the environment and detects new obstacles, rapid path replanning is essential to avoid collisions. This thesis presents a new algorithm called Hierarchical D* Lite (HD*), which combines the incremental algorithm D* Lite with a novel hierarchical path planning approach to replan paths sufficiently fast for real-time operation. Unlike current hierarchical planning algorithms, HD* does not require map corrections before planning a new path. Directional cost scale factors, path smoothing, and Catmull-Rom splines are used to ensure the resulting paths are feasible. HD* sacrifices optimality for real-time performance. Its computation time and path quality are dependent on the map size, obstacle density, sensor range, and any restrictions on planning time. For the most complex scenarios tested, HD* found paths within 10% of optimal in under 35 milliseconds.

The Motif Tracking Algorithm

The search for patterns or motifs in data represents a problem area of key interest to finance and economic researchers. In this paper we introduce the Motif Tracking Algorithm, a novel immune inspired pattern identification tool that is able to identify unknown motifs of a non specified length which repeat within time series data. The power of the algorithm comes from the fact that it uses a small number of parameters with minimal assumptions regarding the data being examined or the underlying motifs. Our interest lies in applying the algorithm to financial time series data to identify unknown patterns that exist. The algorithm is tested using three separate data sets. Particular suitability to financial data is shown by applying it to oil price data. In all cases the algorithm identifies the presence of a motif population in a fast and efficient manner due to the utilisation of an intuitive symbolic representation. The resulting population of motifs is shown to have considerable potential value for other applications such as forecasting and algorithm seeding.

Motion compensation with minimal residue dispersion matching criteria

Dissertação (mestrado)—Universidade de Brasília, Faculdade de Tecnoloigia, 2016.

What to do when K-means clustering fails:a simple yet principled alternative algorithm

The K-means algorithm is one of the most popular clustering algorithms in current use as it is relatively fast yet simple to understand and deploy in practice. Nevertheless, its use entails certain restrictive assumptions about the data, the negative consequences of which are not always immediately apparent, as we demonstrate. While more flexible algorithms have been developed, their widespread use has been hindered by their computational and technical complexity. Motivated by these considerations, we present a flexible alternative to K-means that relaxes most of the assumptions, whilst remaining almost as fast and simple. This novel algorithm which we call MAP-DP (maximum a-posteriori Dirichlet process mixtures), is statistically rigorous as it is based on nonparametric Bayesian Dirichlet process mixture modeling. This approach allows us to overcome most of the limitations imposed by K-means. The number of clusters K is estimated from the data instead of being fixed a-priori as in K-means. In addition, while K-means is restricted to continuous data, the MAP-DP framework can be applied to many kinds of data, for example, binary, count or ordinal data. Also, it can efficiently separate outliers from the data. This additional flexibility does not incur a significant computational overhead compared to K-means with MAP-DP convergence typically achieved in the order of seconds for many practical problems. Finally, in contrast to K-means, since the algorithm is based on an underlying statistical model, the MAP-DP framework can deal with missing data and enables model testing such as cross validation in a principled way. We demonstrate the simplicity and effectiveness of this algorithm on the health informatics problem of clinical sub-typing in a cluster of diseases known as parkinsonism.

Optimization of Pattern Matching Algorithms for Multi- and Many-Core Platforms

Image and video compression play a major role in the world today, allowing the storage and transmission of large multimedia content volumes. However, the processing of this information requires high computational resources, hence the improvement of the computational performance of these compression algorithms is very important. The Multidimensional Multiscale Parser (MMP) is a pattern-matching-based compression algorithm for multimedia contents, namely images, achieving high compression ratios, maintaining good image quality, Rodrigues et al. [2008]. However, in comparison with other existing algorithms, this algorithm takes some time to execute. Therefore, two parallel implementations for GPUs were proposed by Ribeiro [2016] and Silva [2015] in CUDA and OpenCL-GPU, respectively. In this dissertation, to complement the referred work, we propose two parallel versions that run the MMP algorithm in CPU: one resorting to OpenMP and another that converts the existing OpenCL-GPU into OpenCL-CPU. The proposed solutions are able to improve the computational performance of MMP by 3 and 2:7 , respectively. The High Efficiency Video Coding (HEVC/H.265) is the most recent standard for compression of image and video. Its impressive compression performance, makes it a target for many adaptations, particularly for holoscopic image/video processing (or light field). Some of the proposed modifications to encode this new multimedia content are based on geometry-based disparity compensations (SS), developed by Conti et al. [2014], and a Geometric Transformations (GT) module, proposed by Monteiro et al. [2015]. These compression algorithms for holoscopic images based on HEVC present an implementation of specific search for similar micro-images that is more efficient than the one performed by HEVC, but its implementation is considerably slower than HEVC. In order to enable better execution times, we choose to use the OpenCL API as the GPU enabling language in order to increase the module performance. With its most costly setting, we are able to reduce the GT module execution time from 6.9 days to less then 4 hours, effectively attaining a speedup of 45 .

POTENCIAL DOS DADOS DO SETOR CENSITÁRIO BRASILEIRO APLICADO AO MARKETING DE UM FAST FOOD DELIVERY

O projeto de pesquisa preocupou-se em identificar espacialmente o potencial sócio-econômico relacionado aos clientes da Rede Fast Food Delivery "China in Box" que se encontra instalado na região central do Município de Florianópolis - Santa Catarina - Brasil. A base de dados alfanumérica proveniente do IBGE (Setores Censitários) foi manipulada no ambiente SIG-ArcGIS e tornou-se a base para a tomada de decisões estratégicas de marketing pela empresa. O método desenvolvido e validado na pesquisa se configurou na estruturação do banco de dados geográfico referente ao cadastro dos clientes, o qual foi definido por um princípio de geocodificação. Posteriormente a estruturação do banco de dados no SIG, que contemplou as variáveis referentes ao nível de renda, escolaridade, faixa etária, quantidade de moradores, entre outras, foram empregadas funções de análise espacial, por meio da ferramenta Spatial Analyst. A aplicação desse recurso permitiu gerar, por exemplo, mapas temáticos que responderam e espacializaram os questionamentos de interesse ao plano estratégico de desenvolvimento da empresa e a tomada de decisão segundo os princípios do marketing.

Analysis of slip distribution of large earthquakes oriented to tsunamigenesis characterisation

The present thesis focuses on the on-fault slip distribution of large earthquakes in the framework of tsunami hazard assessment and tsunami warning improvement. It is widely known that ruptures on seismic faults are strongly heterogeneous. In the case of tsunamigenic earthquakes, the slip heterogeneity strongly influences the spatial distribution of the largest tsunami effects along the nearest coastlines. Unfortunately, after an earthquake occurs, the so-called finite-fault models (FFM) describing the coseismic on-fault slip pattern becomes available over time scales that are incompatible with early tsunami warning purposes, especially in the near field. Our work aims to characterize the slip heterogeneity in a fast, but still suitable way. Using finite-fault models to build a starting dataset of seismic events, the characteristics of the fault planes are studied with respect to the magnitude. The patterns of the slip distribution on the rupture plane, analysed with a cluster identification algorithm, reveal a preferential single-asperity representation that can be approximated by a two-dimensional Gaussian slip distribution (2D GD). The goodness of the 2D GD model is compared to other distributions used in literature and its ability to represent the slip heterogeneity in the form of the main asperity is proven. The magnitude dependence of the 2D GD parameters is investigated and turns out to be of primary importance from an early warning perspective. The Gaussian model is applied to the 16 September 2015 Illapel, Chile, earthquake and used to compute early tsunami predictions that are satisfactorily compared with the available observations. The fast computation of the 2D GD and its suitability in representing the slip complexity of the seismic source make it a useful tool for the tsunami early warning assessments, especially for what concerns the near field.

Motion Cueing Algorithm Development in a 2DOF Driving Simulator: a Driving Behaviour-Centered Approach

I simulatori di guida sono strumenti altamente tecnologici che permettono di svolgere attività di ricerca in vari ambiti quali la psicologia, la medicina e l’ingegneria. Tuttavia, affinché i dati ottenuti mediante le simulazioni siano rapportabili alla loro controparte reale, la fedeltà delle componenti del simulatore di guida deve essere elevata. Questo lavoro tratta del miglioramento del sistema di restituzione del movimento nel simulatore a due gradi di libertà (2DOF) SIMU-LACET Driving Simulator, costruito e sviluppato presso il laboratorio LEPSIS dell’IFSTTAR (Istituto Francese delle Scienze e Tecnologie dei Trasporti, dello Sviluppo e delle Reti), in particolare nella sua sede di Parigi – Marne-la-Vallée. Si è deciso di andare a riprogettare la parte software del sistema di restituzione del movimento (motion cueing), operando su due elementi principali: lo scale factor (fattore di scala) applicato agli impulsi dinamici provenienti dal modello veicolare e i Motion Cueing Algorihms (MCA, algoritmi di restituzione del movimento), questo per entrambi i gradi di libertà. Si è quindi intervenuti sul modello esistente implementato in MATLAB-Simulink nello specifico blocco del motion cueing sul surge (traslazione longitudinale) e sul yaw (imbardata). Riguardo lo scale factor, è stata introdotta una metodologia per creare uno scale factor non lineare in forma esponenziale, tale da migliorare la restituzione degli impulsi meno ampi, pur rispettando i limiti fisici della piattaforma di movimento. Per quanto concerne il MCA, si sono vagliate diverse transfer function dell’algoritmo classico. La scelta finale dei MCA e la validazione del motion cueig in genere è stata effettuata mediante due esperimenti ed il giudizio dei soggetti che vi hanno partecipato. Inoltre, in virtù dei risultati del primo esperimento, si è investigata l’influenza che la strategia in merito al cambio delle marce avesse sulla percezione del movimento da parte del guidatore.

Economic and Environmental Optimization of Ornamental Stone Block Cutting with a 3D Algorithm

This thesis focuses on finding the optimum block cutting dimensions in terms of the environmental and economic factors by using a 3D algorithm for a limestone quarry in Foggia, Italy. The environmental concerns of quarrying operations are mainly: energy consumption, material waste, and pollution. The main economic concerns are the block recovery, the selling prices, and the production costs. Fractures adversely affect the block recovery ratio. With a fracture model, block production can be optimized. In this research, the waste volume produced by quarrying was minimised to increase the recovery ratio and ensure economic benefits. SlabCutOpt is a software developed at DICAM–University of Bologna for block cutting optimization which tests different cutting angles on the x-y-z planes to offer up alternative cutting methods. The program tests several block sizes and outputs the optimal result for each entry. By using SlabCutOpt, ten different block dimensions were analysed, the results indicated the maximum number of non-intersecting blocks for each dimension. After analysing the outputs, the block named number 1 with the dimensions ‘1mx1mx1m’ had the highest recovery ratio as 43% and the total Relative Money Value (RMV) with a value of 22829. Dimension number 1, also had the lowest waste volume, with a value of 3953.25 m3, for the total bench. For cutting the total bench volume of 6932.25m3, the diamond wire cutter had the lowest dust emission values for the block with the dimension ‘2mx2mx2m’, with a value of 24m3. When compared with the Eco-Label standards, block dimensions having surface area values lower than 15m2, were found to fit the natural resource waste criteria of the label, as the threshold required 25% of minimum recovery [1]. Due to the relativity of production costs, together with the Eco-Label threshold, the research recommends the selection of the blocks with a surface area value between 6m2 and 14m2.