A comparison of automated keyphrase extraction techniques and of automatic evaluation vs. human evaluation

Keyphrases are added to documents to help identify the areas of interest they contain. However, in a significant proportion of papers author selected keyphrases are not appropriate for the document they accompany: for instance, they can be classificatory rather than explanatory, or they are not updated when the focus of the paper changes. As such, automated methods for improving the use of keyphrases are needed, and various methods have been published. However, each method was evaluated using a different corpus, typically one relevant to the field of study of the method’s authors. This not only makes it difficult to incorporate the useful elements of algorithms in future work, but also makes comparing the results of each method inefficient and ineffective. This paper describes the work undertaken to compare five methods across a common baseline of corpora. The methods chosen were Term Frequency, Inverse Document Frequency, the C-Value, the NC-Value, and a Synonym based approach. These methods were analysed to evaluate performance and quality of results, and to provide a future benchmark. It is shown that Term Frequency and Inverse Document Frequency were the best algorithms, with the Synonym approach following them. Following these findings, a study was undertaken into the value of using human evaluators to judge the outputs. The Synonym method was compared to the original author keyphrases of the Reuters’ News Corpus. The findings show that authors of Reuters’ news articles provide good keyphrases but that more often than not they do not provide any keyphrases.

Nonstationary feature extraction techniques for automatic classification of impact acoustic signals

Condition monitoring of wooden railway sleepers applications are generallycarried out by visual inspection and if necessary some impact acoustic examination iscarried out intuitively by skilled personnel. In this work, a pattern recognition solutionhas been proposed to automate the process for the achievement of robust results. Thestudy presents a comparison of several pattern recognition techniques together withvarious nonstationary feature extraction techniques for classification of impactacoustic emissions. Pattern classifiers such as multilayer perceptron, learning cectorquantization and gaussian mixture models, are combined with nonstationary featureextraction techniques such as Short Time Fourier Transform, Continuous WaveletTransform, Discrete Wavelet Transform and Wigner-Ville Distribution. Due to thepresence of several different feature extraction and classification technqies, datafusion has been investigated. Data fusion in the current case has mainly beeninvestigated on two levels, feature level and classifier level respectively. Fusion at thefeature level demonstrated best results with an overall accuracy of 82% whencompared to the human operator.

Análise da precipitação no semi-árido potiguar

The semiarid rainfall regime is northeastern Brazil is highly variable. Climate processes associated with rainfall are complex and their effects may represent extreme situations of drought or floods, which can have adverse effects on society and the environment. The regional economy has a significant agricultural component, which is strongly influenced by weather conditions. Maximum precipitation analysis is traditionally performed using the intensity-duration-frequency (IDF) probabilistic approach. Results from such analysis are typically used in engineering projects involving hydraulic structures such as drainage network systems and road structures. On the other hand, precipitation data analysis may require the adoption of some kind of event identification criteria. The minimum inter-event duration (IMEE) is one of the most used criteria. This study aims to analyze the effect of the IMEE on the obtained rain event properties. For this purpose, a nine-year precipitation time series (2002- 2011) was used. This data was obtained from an automatic raingauge station, installed in an environmentally protected area, Ecological Seridó Station. The results showed that adopted IMEE values has an important effect on the number of events, duration, event height, mean rainfall rate and mean inter-event duration. Furthermore, a higher occurrence of extreme events was observed for small IMEE values. Most events showed average rainfall intensity higher than 2 mm.h-1 regardless of IMEE. The storm coefficient of advance was, in most cases, within the first quartile of the event, regardless of the IMEE value. Time series analysis using partial time series made it possible to adjust the IDF equations to local characteristics

Fingerprint Recognition: Enhancement, Feature Extraction and Automatic Evaluation of Algorithms

The identification of people by measuring some traits of individual anatomy or physiology has led to a specific research area called biometric recognition. This thesis is focused on improving fingerprint recognition systems considering three important problems: fingerprint enhancement, fingerprint orientation extraction and automatic evaluation of fingerprint algorithms. An effective extraction of salient fingerprint features depends on the quality of the input fingerprint. If the fingerprint is very noisy, we are not able to detect a reliable set of features. A new fingerprint enhancement method, which is both iterative and contextual, is proposed. This approach detects high-quality regions in fingerprints, selectively applies contextual filtering and iteratively expands like wildfire toward low-quality ones. A precise estimation of the orientation field would greatly simplify the estimation of other fingerprint features (singular points, minutiae) and improve the performance of a fingerprint recognition system. The fingerprint orientation extraction is improved following two directions. First, after the introduction of a new taxonomy of fingerprint orientation extraction methods, several variants of baseline methods are implemented and, pointing out the role of pre- and post- processing, we show how to improve the extraction. Second, the introduction of a new hybrid orientation extraction method, which follows an adaptive scheme, allows to improve significantly the orientation extraction in noisy fingerprints. Scientific papers typically propose recognition systems that integrate many modules and therefore an automatic evaluation of fingerprint algorithms is needed to isolate the contributions that determine an actual progress in the state-of-the-art. The lack of a publicly available framework to compare fingerprint orientation extraction algorithms, motivates the introduction of a new benchmark area called FOE (including fingerprints and manually-marked orientation ground-truth) along with fingerprint matching benchmarks in the FVC-onGoing framework. The success of such framework is discussed by providing relevant statistics: more than 1450 algorithms submitted and two international competitions.

Automatic extraction of the mid-facial plane for cranio-maxillofacial surgery planning

Recently developed computer applications provide tools for planning cranio-maxillofacial interventions based on 3-dimensional (3D) virtual models of the patient's skull obtained from computed-tomography (CT) scans. Precise knowledge of the location of the mid-facial plane is important for the assessment of deformities and for planning reconstructive procedures. In this work, a new method is presented to automatically compute the mid-facial plane on the basis of a surface model of the facial skeleton obtained from CT. The method matches homologous surface areas selected by the user on the left and right facial side using an iterative closest point optimization. The symmetry plane which best approximates this matching transformation is then computed. This new automatic method was evaluated in an experimental study. The study included experienced and inexperienced clinicians defining the symmetry plane by a selection of landmarks. This manual definition was systematically compared with the definition resulting from the new automatic method: Quality of the symmetry planes was evaluated by their ability to match homologous areas of the face. Results show that the new automatic method is reliable and leads to significantly higher accuracy than the manual method when performed by inexperienced clinicians. In addition, the method performs equally well in difficult trauma situations, where key landmarks are unreliable or absent.

Automatic extraction of femur contours from calibrated fluoroscopic images

Automatic identification and extraction of bone contours from X-ray images is an essential first step task for further medical image analysis. In this paper we propose a 3D statistical model based framework for the proximal femur contour extraction from calibrated X-ray images. The automatic initialization is solved by an estimation of Bayesian network algorithm to fit a multiple component geometrical model to the X-ray data. The contour extraction is accomplished by a non-rigid 2D/3D registration between a 3D statistical model and the X-ray images, in which bone contours are extracted by a graphical model based Bayesian inference. Preliminary experiments on clinical data sets verified its validity

Effective acoustical monitoring tools for Mannophryne lamarcai

We performed the field-work during the dry (March 2014) and reainy season (May and June 2014) at the species type locality: Cerro Socopó, located at central-west region between Falcón, Lara and Zulia states, Venezuela. Socopó is a small and isolated mountain (1.571 m) belonging to the Ziruma mountains, and represents a relict of tropical mountain forest surrounded by semi-arid vegetation and grassland. This area is home to 312 species of vertebrates, including endangered and endemic amphibians species like Mannophryne lamarcai, Leptodactylus magistris and Dendropsophus amicorum. These forest and species are severely threatened by cattle ranch and illegal timber extraction, with forest formations only above 1000 meters. Despite this, no legal protected figure has been established in the area. We identified a 2.5 km secondary road transect within the study area based on the following criteria: 1) it cover different habitat types (streams and lagoons); and 2) it within the altitudinal gradient described for the specie (1,040 to 1,363 m). We identified three sampling points throughout the transect located in the vicinity of wetland habits: socopo1, socopo2 and socopo4. We did two types of recordings: 1) high quality recordings to characterize the advertisement call for M. lamarcai, non described to date (socopo4), and 2) recordings in different sampling points to evaluate call detectors performance in different acoustic scenarios (in all three localities).

Automatic extraction of syntactic semantic patterns for multilingual resources

In this paper we present an automatic system for the extraction of syntactic semantic patterns applied to the development of multilingual processing tools. In order to achieve optimum methods for the automatic treatment of more than one language, we propose the use of syntactic semantic patterns. These patterns are formed by a verbal head and the main arguments, and they are aligned among languages. In this paper we present an automatic system for the extraction and alignment of syntactic semantic patterns from two manually annotated corpora, and evaluate the main linguistic problems that we must deal with in the alignment process.

Real-time 3D semi-local surface patch extraction using GPGPU

Feature vectors can be anything from simple surface normals to more complex feature descriptors. Feature extraction is important to solve various computer vision problems: e.g. registration, object recognition and scene understanding. Most of these techniques cannot be computed online due to their complexity and the context where they are applied. Therefore, computing these features in real-time for many points in the scene is impossible. In this work, a hardware-based implementation of 3D feature extraction and 3D object recognition is proposed to accelerate these methods and therefore the entire pipeline of RGBD based computer vision systems where such features are typically used. The use of a GPU as a general purpose processor can achieve considerable speed-ups compared with a CPU implementation. In this work, advantageous results are obtained using the GPU to accelerate the computation of a 3D descriptor based on the calculation of 3D semi-local surface patches of partial views. This allows descriptor computation at several points of a scene in real-time. Benefits of the accelerated descriptor have been demonstrated in object recognition tasks. Source code will be made publicly available as contribution to the Open Source Point Cloud Library.

Graph-based Semi-Supervised Learning in Acoustic Modeling for Automatic Speech Recognition

Thesis (Ph.D.)--University of Washington, 2016-06