Scenario-Driven Automatic Pattern Recognition in Nowcasting

In this thesis, the development of one of the pattern recognition algorithms for the Belle II experiment based on conformal and Legendre transformations is presented. In order to optimize the performance of the algorithm (CPU time and efficiency) specialized processing steps have been introduced. To show achieved results, Monte-Carlo based efficiency measurements of the tracking algorithms in the Central Drift Chamber (CDC) has been done.

Use of pattern recognition in photoacoustic imaging to identify neuronal ensembles in the prefrontal cortex of rats undergoing conditioned fear learning.

Online learning has become a trend for the current generation of students who have been exposed to advanced information and communication technology. Smart education can use pattern recognition. Manual assessments are subjective and inconsistent. To overcome these problems, pattern recognition can be used in the non-verbal aspect assessment system. This study describes pattern recognition in online learning about the functions, modalities, and algorithms and specifically related to the recognition system of non-verbal aspects of teaching skills. The literature study was carried out through the stages of planning, selection, extraction, and selection. There are 86 articles reviewed. The first result is the functions of implementing pattern recognition in online learning are engagement recognition, attention detection, emotion recognition, learning behavior, learning activity recognition, authentication, teaching training, etc. using four classifications of modality: visual, audio, biosignal, behavioral, and CNN as the most widely used learning algorithm. Secondly, all modalities (except behavioral) and CNN algorithm can be used for assessing teaching skills. Early development of the non-verbal aspect assessment system can use Facial Expression Recognition (FER) and Hand Gesture Recognition (HGR). The future analysis needs to focus on technology characteristics, the meaningfulness of the content, and the proper teaching mode. In the end, hopefully, prospective teachers will acquire technology that can make it easier for them to practice teaching and get objective assessments.

Wildfires have increased in frequency, duration and size in the western United States (U.S.) over the past decades. These trends are projected to continue, with negative consequences for air quality across the U.S. Wildfires emit large quantities of particles and gases that serve as air pollutants and their precursors, and can lead to severe air quality conditions over large spatial and long temporal scales. Characterization of the chemical constituents in smoke as a function of combustion conditions, fuel type, and fuel component is an important step towards improving the prediction of air quality effects from fires and evaluating mitigation strategies. Building on the comprehensive characterization of gaseous non-methane organic compounds (NMOCs) identified in laboratory and field studies, a supervised pattern recognition algorithm was developed that successfully identified unique chemical speciation profiles among similar fuel types common in western coniferous forests. The algorithm was developed using laboratory data from single fuel species and tested on simplified synthetic fuel mixtures. The fuel types in the synthetic mixtures were differentiated but as the relative mixing proportions became more similar, the differentiation became poorer. Using the results from the pattern recognition algorithm, a classification model based on linear discriminant analysis was trained to differentiate smoke samples based on the contribution(s) of dominant fuel type(s). The classification model was applied to field data and despite the complexity of contributing fuels, and the presence of fuels "unknown" to the classifier, the dominant sources/fuel types were identified correctly. The pattern recognition and classification algorithms are a promising approach for identifying the types of fuels contributing to smoke samples and facilitating selection of appropriate chemical speciation profiles for predictive air quality modeling, using a highly reduced suite of measured NMOCs. Utility and performance of the pattern recognition and classification algorithms can be improved by expanding the training and test sets to include data from a broader range of single and mixed fuel types.

Shock signal features must be extracted for use in pattern recognition or fault diagnosis. In this work, we proposed a method for the feature extraction of shock signals, which are vibration signals that change faster and have larger amplitude ranges than general signals. First, we proposed the concepts of amplitude density for monotonic functions and piecewise monotonic functions. On the basis of these concepts, we then proposed the concept of the upper limit of density integral (ULDI), which was adopted to obtain signal features. Then, we introduced two types of serious fault cracks to the latch sheet of an automatic gun mechanism that is used on warships. Next, we applied the proposed method to extract the features of shock signals from data acquired when the automatic gun mechanism fired with normal and two fault patterns. Finally, we verified the effectiveness of our proposed method by applying the features that it extracted to a support vector machine (SVM). Our proposed method provided good results and was superior to the traditional statistics-based feature extraction method when applied to a SVM for classification. In addition, the former method demonstrated better generalisation than the latter. Thus, our method is an efficient approach for extracting shock signal features in pattern recognition and fault diagnosis.

Evaluation of a multiple gas mixture with a simple MOSFET gas sensor array and pattern recognition

Despite the appearance of advanced multi-degrees of freedom (DoF) robotic hands during the past decade, prosthetic control lacks a powerful interface to facilitate all its functionalities in a manner that is acceptable for a majority of users [1]. In this article, we explore the feasibility of using a sensing technique called force myography (FMG) as an alternative or synergist to the traditional surface electromyography (sEMG) technique as a human-machine interface (HMI) for the control of a multi-DoF prosthetic hand, bebionic 3 by Ottobock, Austin, Texas. In this article, we present a prosthetic prototype developed for the Cybathlon 2016, a championship for racing pilots with disabilities using assistive robotic devices. The design of the prototype is discussed and the effect of two factors on its control is analyzed. These factors are 1) the impact of a multisensory approach and 2) the placement of FMG sensor strips within the prosthetic inner socket. Analysis is performed by comparing resulting pattern recognition accuracies. Results show that the use of both sensing modalities (FMG and EMG) together produced the highest pattern recognition accuracy (81.1%) for ten classes of motion (four wrist movements and six grip patterns). We demonstrated that FMG has the potential to be an HMI for control of upper-limb-powered prostheses. FMG also illustrates the potential for intuitive control through the use of pattern recognition. A multisensory approach could assist in increasing robustness of the HMI for prosthetic control.

We describe the use of pattern recognition and multivariate regression in the assessment of complex mixtures by correlating chemical fingerprints to the mutagenicity of the mixtures. Mixtures were 20 organic extracts of exhaust particles, each containing 102-170 individual compounds such as polycyclic aromatic hydrocarbons (PAHs), nitro-PAHs, oxy-PAHs, and saturated hydrocarbons. Mixtures were characterized by full-scan GC-MS (gas chromatography-mass spectrometry). Data were resolved into peaks and spectra for individual compounds by an automated curve resolution procedure. Resolved chromatograms were integrated, resulting in a predictor matrix that was used as input to a principal component analysis to evaluate similarities between mixtures (i.e., classification). Furthermore, partial least-squares projections to latent structures were used to correlate the GC-MS data to mutagenicity, as measured in the Ames Salmonella assay (i.e., calibration). The best model (high r2 and Q2) identifies the variables that co-vary with the observed mutagenicity. These variables may subsequently be identified in more detail. Furthermore, the regression model can be used to predict mutagenicity from GC-MS chromatograms of other organic extracts. We emphasize that both chemical fingerprints as well as detailed data on composition can be used in pattern recognition.

Nowadays the use of pattern recognition for signal processing presents a novel approach in order to determine the position and number of flaws within a tested material. The aim of this work is to propose a pattern recognition technique in order to determine different types of flaws based upon pulse echoes using ultrasonic measurements. This approach is performed by the processing of an ultrasonic measurement using time frequency distribution technique and ART2 network. This approach is focused to understand how pattern recognition can be pursued into fault diagnosis.

Ten transition-metal ions are evaluated as chemical ionization reagents by use of pattern recognition. The data for the reactions of Sc/sup +/, Y/sup +/, La/sup +/, V/sup +/, Nb/sup +/, Ta/sup +/, Cr/sup +/, Fe/sup +/, Co/sup +/, and Cu/sup +/ with 11 ketones, 8 aldehydes, and 8 ethers were collected with a Fourier transform mass spectrometer equipped with a laser desorption ionization source. Selectivities of the metal ions for the three organic classes are quantitated and compared to 70-eV electron impact (EI) data. While a 100% recognition accuracy for the three classes is obtained for EI, Cu/sup +/, and Sc/sup +/ data, the fewest features are required for, and the best separation is produced with, the Sc/sup +/ data. The selected features are also examined for their chemical information, providing clues to the characteristic reactivities and reaction mechanisms for the metal ions with three functional groups.

Persistent homology is a fairly new branch of computational topology which combines geometry and topology for an effective shape description of use in Pattern Recognition. In particular, it registers through Numbers the presence of holes and their persistence while a parameter (filtering function) is varied. In this paper, some recent developments in this field are integrated in a k-nearest neighbor search algorithm suited for an automatic retrieval of melanocytic lesions. Since long, dermatologists use five morphological parameters (A $$=$$= asymmetry, B $$=$$= boundary, C $$=$$= color, D $$=$$= diameter, E $$=$$= evolution) for assessing the malignancy of a lesion. The algorithm is based on a qualitative assessment of the segmented images by computing both 1 and 2-dimensional persistent Betti Number functions related to the ABCDE parameters and to the internal texture of the lesion. The results of a feasibility test on a set of 107 melanocytic lesions are reported in the section dedicated to the numerical experiments.

Classical strategies for fault classification have the drawback that they do not identify new fault scenarios online. Therefore, classification online becomes dependant on computation delays. In here, this problem is taken as a pattern recognition issue. The approach followed is based upon a fuzzy ART2 network. It consists of two modules, firstly the recognition of new scenarios is performed by the network. Secondly, the classification of every group of patterns is performed by a decision-making procedure. This work addresses the problem of fault classification online as a problem of pattern recognition rather than a fault detection approach. The use of pattern recognition presents the advantage of classification of recognized patterns as non fault scenario. The appearance of new patterns is taken as part of fault behaviour.

We approach sphere of influence graphs (SIGs) from a probabilistic perspective. Ordinary SIGs were first introduced by Toussaint as a type of proximity graph for use in pattern recognition, computer vision and other low-level vision tasks. A random sphere of influence graph (RSIG) is constructed as follows. Consider n points uniformly and independently distributed within the unit square in d dimensions. Around each point, X i , draw an open ball (‘sphere of influence’) with radius equal to the distance to X i 's nearest neighbour. Finally, draw an edge between two points if their spheres of influence intersect. Asymptotically exact values for the expected number of edges in a RSIG are determined for all values of d; previously, just upper and lower bounds were known for this quantity. A modification of the Azuma-Hoeffding exponential inequality is employed to exhibit the sharp concentration of the number of edges around its expected value.

In this work we have studied two classes of molecular groups trying to establish correlations between their electronic structure and correlated biological activity. The first group contains integrase inhibitors, which are necessary to HIV-1 replication; the second group contains Escherichia Coli inhibitors, one bacterium which caused diarrhea. Electronic Index Methodology (EIM) was used in the Structure-Activity Relationship (SAR) analysis. In order to have an independent statistical validation, the data produced from EIM were contrasted with more standard pattern recognition methodologies: Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA), and Artificial Neural Networks (ANN). Our results shown that EIM can be sucessfully use for pattern recognition. As in previous works for other organic families the EIM out performed PCA, HCA and ANN techniques presenting the same predictive power with a significant reduction of computational effort.

The background, development, performance assessment, and analysis of a novel pattern recognition algorithm that is applicable to any set of binary images are discussed. The efficacy of the algorithm when applied to the problem of fingerprint recognition is quantified. The conclusion was that the algorithm is relatively poor as a fingerprint identification algorithm, averaging an equal error rate of approximately 19% as calculated by the rules specified in the Year 2000 Fingerprint Verification Competition. The positive attributes of the algorithm were its ultra-fast matching times, orientation independence, lack of rejection events, relative insensitivity to resolution difference, and one-way transformations. The mechanism of algorithm operation as applied to fingerprints was investigated using integral geometry. This investigation showed that the algorithm was an indirect measure of ridge width, which explained the algorithm's relatively poor performance. Another set of experiments suggests that the algorithm may be well-suited to other pattern recognition problems, specifically cloud and precipitation particle recognition and camouflage recognition. In summary, the research extends the field of pattern recognition by developing, assessing the performance, and determining the mechanism of operation of a novel pattern recognition algorithm that is applicable to any set of binary images.