Glass Dataset Research Articles

Minimum interpretation by autoencoder-based serial and enhanced mutual information production

The present paper aims to propose an information-theoretic method for interpreting the inference mechanism of neural networks. The new method aims to interpret the inference mechanism minimally by disentangling complex information into simpler and easily interpretable information. This disentanglement of complex information can be realized by maximizing mutual information between input patterns and the corresponding neurons. However, because the use of mutual information has faced difficulty in computation, we use the well-known autoencoder to increase mutual information by re-interpreting the sparsity constraint, which is considered a device to increase mutual information. The computational procedures to increase mutual information are decomposed into the serial operation of equal use of neurons and specific responses to input patterns. The specific responses are realized by enhancing the results by the equal use of neurons. The method was applied to three data sets: the glass, office equipment, and pulsar data sets. With all three data sets, we could observe that, when the number of neurons was forced to increase, mutual information could be increased. Then, collective weights, or average collectively treated weights, showed that the method could extract the simple and linear relations between inputs and targets, making it possible to interpret the inference mechanism minimally.

The Late Holocene tephra record of the central Mediterranean Sea: Mapping occurrences and new potential isochrons for the 4.4–2.0 ka time interval

ABSTRACTFive cores from the southern Tyrrhenian and Ionian seas were studied for their tephra and cryptotephra content in the 4.4–2.0 ka time interval. The chronological framework for each core was obtained by accelerator mass spectrometry 14C dating, the occurrence of distinct marker tephra and stratigraphic correlation with adjacent records. Tephrochronology allowed us to correlate the analyzed deposits with tephra markers associated with Somma‐Vesuvius (79 ad), Ischia Island (Cretaio), Mt Etna (FG, FL and FS) and Campi Flegrei (Astroni‐Agnano Monte Spina) events. For the first time in the marine setting, a large single glass data set is provided for the Late Holocene Etnean marker beds including the FS tephra (ca. 4.3 ka). Moreover, unknown deposits from Lipari (ca. 2.2–2.0 ka) and Vulcano (3.6–3.3 ka) are also recognized at more distal sites than previously reported. These results contribute to improve the high‐resolution tephrostratigraphic framework of the central Mediterranean Sea. They also provide new insights into the chemical composition and dispersal pattern of tephras that can be used as inter‐archive tools for regional and ‘local’ stratigraphic correlations and for addressing paleoclimate research.

Internal and collective interpretation for improving human interpretability of multi-layered neural networks

The present paper aims to propose a new type of information-theoretic method to interpret the inference mechanism of neural networks. We interpret the internal inference mechanism for itself without any external methods such as symbolic or fuzzy rules. In addition, we make interpretation processes as stable as possible. This means that we interpret the inference mechanism, considering all internal representations, created by those different conditions and patterns. To make the internal interpretation possible, we try to compress multi-layered neural networks into the simplest ones without hidden layers. Then, the natural information loss in the process of compression is complemented by the introduction of a mutual information augmentation component. The method was applied to two data sets, namely, the glass data set and the pregnancy data set. In both data sets, information augmentation and compression methods could improve generalization performance. In addition, compressed or collective weights from the multi-layered networks tended to produce weights, ironically, similar to the linear correlation coefficients between inputs and targets, while the conventional methods such as the logistic regression analysis failed to do so.

Feasibility of informing syndrome-level empiric antibiotic recommendations using publicly available antibiotic resistance datasets.

Background: Antibiotics are most often prescribed empirically, meaning that they are used to treat infection syndromes prior to identification of the causative bacteria and their susceptibility to antibiotics. The effectiveness of antibiotic therapies is now compromised by the emergence and spread of antibiotic-resistant bacteria. Guidelines on empiric antibiotic therapy are a key component of effective clinical care for infection syndromes, as treatment needs to be informed by knowledge of likely aetiology and bacterial resistance patterns. Methods: We used open-access antimicrobial resistance (AMR) surveillance datasets, including the newly available ATLAS dataset from Pfizer, to derive a composite index of antibiotic resistance for common infection syndromes. Results: We developed a framework that integrated data on antibiotic prescribing guidelines, aetiology of infections, access to and cost of antibiotics, with antibiotic susceptibilities from global AMR surveillance datasets to create an empirical prescribing index. The results are presented in an interactive web app to allow users to visualise underlying resistance rates to first-line empiric antibiotics for their infection syndromes and countries of interest. Conclusions: We found that whilst an index for empiric antibiotic therapy based on resistance data can technically be created, the ATLAS dataset in its current form can only inform on a limited number of infection syndromes. Other open-access AMR surveillance datasets (ECDC Surveillance Atlas, CDDEP ResistanceMap and WHO GLASS datasets) are largely limited to bacteraemia-derived specimens and cannot directly inform treatment of other infection syndromes. With improving data availability on international rates of AMR and better understanding of infection aetiology, our approach may prove useful for informing empiric prescribing decisions in settings with limited local AMR surveillance data. Syndrome-level resistance could be a more clinically relevant measure of resistance to inform on the appropriateness of empiric antibiotic therapies at the country-level.

An intelligent quality-based approach to fusing multi-source possibilistic information

Recently, Yager and Petry were proposing a quality-based methodology to combine data provided by multiple probabilistic sources to improve the quality of information for decision-makers. This paper offers a sort of companion paper that adapts this methodology to possibilistic sources. Possibility theory is particularly well suited to cope with incomplete information from poor-data sources. The methodology and algorithms used for the probabilistic approach are adapted for the possibilistic case. Both approaches are then compared by the means of a numerical example and four experimental benchmark datasets: one, the IRIS data set, being data-poorer than the three other ones (Diabetes dataset, Glass dataset and Liver-disorder dataset). A vector representation is introduced for a possibility distribution as in the probabilistic case and, the Gini's formulation of entropy is being used. However, the Gini's entropy has to be used differently than with the probabilistic case. This has an impact on the selection of subsets. A fusion scheme is designed to select the best-quality subsets according to two information quality factors: quantity of information and source credibility. Results obtained from comparison of both approaches on the four experimental benchmarks confirm the superiority of the possibilistic approach in the presence of information scarcity or incompleteness.

Analysis of Accuracy and Epoch on Back-propagation BFGS Quasi-Newton

Back-propagation is one of the learning algorithms on artificial neural networks that have been widely used to solve various problems, such as pattern recognition, prediction and classification. The Back-propagation architecture will affect the outcome of learning processed. BFGS Quasi-Newton is one of the functions that can be used to change the weight of back-propagation. This research tested some back-propagation architectures using classical back-propagation and back-propagation with BFGS. There are 7 architectures that have been tested on glass dataset with various numbers of neurons, 6 architectures with 1 hidden layer and 1 architecture with 2 hidden layers. BP with BFGS improves the convergence of the learning process. The average improvement convergence is 98.34%. BP with BFGS is more optimal on architectures with smaller number of neurons with decreased epoch number is 94.37% with the increase of accuracy about 0.5%.

New findings of Late Glacial Etna pumice fall deposits in NE Sicily and implications for distal tephra correlations in the Mediterranean area

Pumice fall deposits found in the Nebrodi Mountains and along the Alcantara River, close to the town of Randazzo (NE Sicily), have been studied to derive information about their volcanic source and age. The geochemical Na-alkaline affinity of juvenile products, benmoreite to trachyte, clearly indicates they originated from Etna volcano (Sicily). Major (EMPA) and trace (LA-ICP-MS) element compositional data on matrix glasses indicate that the investigated deposits have a compositional affinity consistent with the tephra deposits of unit D produced by the Ellittico caldera-forming eruptions between ca. 17 and 19 cal ka BP. Furthermore, their compositions correspond to the distal tephra equivalent found in some lacustrine and marine cores in Central Italy (Y-1, TM-11), Tyrrhenian Sea (Et-1; MD10α) and Adriatic Sea (Pal94-66-358; Pal94-8-353). We applied the principal components analysis (PCA), a statistical tool able to reduce the variability of a complex system, to compare the compositions of the proximal samples with the possible distal counterparts found in drill cores of the Mediterranean area. On the basis of northward dispersal of the studied deposits and their geochemical features, we suggest they represent a previously unreported sub-Plinian/Plinian eruption of Ellittico volcano producing medial-distal pumice fall deposits in the Nebrodi Mountains and close to Randazzo, named here the D1c layer. The discovery of these deposits helps solve the problem of distal correlations of the northerly dispersed tephra from Etna related to unit D, for which no definitive attribution with proximal units was given in previous studies. The results presented here add to the knowledge of the eruptive history of the volcano and contribute to expanding the proximal geochemical glass dataset for distal tephra correlation in the Mediterranean region during the Late Glacial period.

Identification of the Changbaishan ‘Millennium’ (B-Tm) eruption deposit in the Lake Suigetsu (SG06) sedimentary archive, Japan: Synchronisation of hemispheric-wide palaeoclimate archives

The B-Tm tephra, dispersed during the highly explosive Changbaishan ‘Millennium’ eruption (ca. 940–950 CE) and a key marker layer within the Greenland ice cores, has now been identified in the Lake Suigetsu (SG06) sedimentary sequence, central Japan. The major element geochemistry of the volcanic glasses within this tephra layer are compared to a new glass dataset from the distal type-locality (Tomakomai Port, Hokkaido) and other published ‘Millennium’ eruption/B-Tm deposits, to verify this correlation. The discovery of the B-Tm tephra in the Lake Suigetsu record provides, to date, the most southerly identification of this ash and, crucially, the first direct tie-point between this high-resolution, mid-latitude palaeoclimate archive and the Greenland ice cores. These findings present significant encouragement for on-going research into the tephrostratigraphy of East Asia, focusing on the identification of widely-dispersed tephra layers which can facilitate the synchronisation of disparate palaeoclimate archives and thus enable the assessment of spatio-temporal variations in past climatic change.

Design and Development of Data Classification Methodology for Uncertain Data

The classification of uncertain data has become one of the tedious processes in the data-mining. The uncertain dataset contains tuples with different and multiple data and thus to predict correct output class is a complex process. In this paper, we present two algorithms for classifying the uncertain data using the KNN classifier, which handles the uncertain dataset in two different ways to discover the corresponding class. In both algorithms, we split the database into two portions. The first portion is named as training dataset and the second portion is name as testing dataset. In the first algorithm, we used the properties of uncertain data to convert the uncertain data into certain data. The algorithm 1 initially converts the uncertain data to certain data and then it utilizes the KNN algorithm to classify data through the distance measure. The second algorithm converts the uncertain data through probability distribution function (pdf). The algorithm 2, initially calculates the N number of split point for each attributes of the training part of uncertain data then it calculates pdf with respect to the selected split point. The same process is applied for testing portion of uncertain data; subsequently algorithm 2 employs the KNN algorithm to classify the converted data. Finally, we compared our proposed algorithm with the UDT (Uncertain Decision Tree) algorithm with the four real datasets such as iris dataset, ionosphere dataset, breast cancer dataset, glass dataset and we proved that our proposed algorithms performed well than the UDT algorithm in terms of accuracy.

Integration of Particle Swarm Optimization and Immune Genetic Algorithm-Based Dynamic Clustering for Customer Clustering

This study intends to present a dynamic clustering (DC) approach based on particle swarm optimization (PSO) and immune genetic (IG) (DCPIG) algorithm, which is able to cluster the data into adequate clusters through data characteristics with pre-specified numbers of clusters. The proposed DCPIG algorithm is compared with three DC algorithms in the literature using Iris, Wine, Glass and Vowel benchmark data sets. The experiment results show that the DCPIG algorithm can achieve higher stability and accuracy than the other algorithms. In addition, the DCPIG algorithm is also applied to a real-world problem considering the customer clustering for a cyber flower shop. Lastly, we recommend different products and services to customers based on the clustering results.

Integration of Artificial Immune System and K-Means Algorithm for Customer Clustering

This study proposes a novel artificial immune system (AIS)-based clustering algorithm, which integrates with a K-means (AISK) algorithm for a customer clustering problem. Computational results using Iris, Glass, Wine, and Breast Cancer benchmark datasets indicate that the proposed AIS-based clustering algorithm is more accurate than some particle swarm optimization (PSO)-based clustering algorithms. In addition, the model evaluation results using a daily transaction database provided by a cyberstore also show that the proposed AISK algorithm is superior to PSO-based clustering algorithms.

Fuzzy Hyperline Segment Neural Network Pattern Classifier with Different Distance Metrics

The Fuzzy Hyperline Segment Neural Network (FHLSNN) pattern classifier utilizes fuzzy set as pattern classes in which each fuzzy set is a union of fuzzy set hyperline segments. The Euclidean distance metric is used to compute the distances to decide the degree of membership function. In this paper, the use of other various distance metrics such as Manhattan, Squared Euclidean, Canberra and Chebyshew distance metrics is proposed. The performance of FHLSNN pattern classifier is evaluated with various benchmark databases such as Glass, Wine, PID, and Iris data set and real handwritten database. The FHLSNN pattern classifier is evaluated for generalization performance under recognition rate, training time and testing time. From the result analysis, the performance of classifier is based on the distance metrics as well the database used is verified. This analysis will help to select a suitable distance metric for fuzzy neural network classifier for particular application. General Terms Pattern Recognition, Neural Network, and Distance metrics

Decision Support System Using Fuzzy Min-Max Neural Network with the Modified Genetic Algorithm

Pattern recognition and classification has been an active research area for many decades and has got a surge in the recent past. The decision support system based on this has been explored in many areas including the business and market research. A two-phase pattern classification and rule extraction system is proposed using fuzzy min-max neural network and modified genetic algorithm in this paper. The first phase consists of a modified fuzzy min–max (FMM) neural-network-based pattern classifier, while the second phase consists of a genetic-algorithm (GA)-based rule extractor. We have redesigned the pruning equation (confidence factor) in the later part of first stage and the rule extraction is carried out using proposed Oppositional Genetic Algorithm with gene amplification operator. The experimentation is carried out using Tea dataset and Glass dataset. The evaluation metric used is the accuracy and we have also compared to existing methods. From the results we can see that our proposed technique has achieved better accuracy values

Double Competition for Information-Theoretic SOM

In this paper, we propose a new type of informationtheoretic method for the self-organizing maps (SOM), taking into account competition between competitive (output) neurons as well as input neurons. The method is called ”double competition”, as it considers competition between outputs as well as input neurons. By increasing information in input neurons, we expect to obtain more detailed information on input patterns through the information-theoretic method. We applied the informationtheoretic methods to two well-known data sets from the machine learning database, namely, the glass and dermatology data sets. We found that the information-theoretic method with double competition explicitly separated the different classes. On the other hand, without considering input neurons, class boundaries could not be explicitly identified. In addition, without considering input neurons, quantization and topographic errors were inversely related. This means that when the quantization errors decreased, topographic errors inversely increased. However, with double competition, this inverse relation between quantization and topographic errors was neutralized. Experimental results show that by incorporating information in input neurons, class structure could be clearly identified without degrading the map quality to severely.

Study of Various Decision Tree Pruning Methods with their Empirical Comparison in WEKA

is important problem in data mining. Given a data set, classifier generates meaningful description for each class. Decision trees are most effective and widely used classification methods. There are several algorithms for induction of decision trees. These trees are first induced and then prune subtrees with subsequent pruning phase to improve accuracy and prevent overfitting. In this paper, various pruning methods are discussed with their features and also effectiveness of pruning is evaluated. Accuracy is measured for diabetes and glass dataset with various pruning factors. The experiments are shown for this two datasets for measuring accuracy and size of the tree.

Geochemistry of the Phlegraean Fields (Italy) proximal sources for major Mediterranean tephras: Implications for the dispersal of Plinian and co-ignimbritic components of explosive eruptions

Volcanic activity at Phlegraean Fields, Italy, produced several major marker tephras over a 50ka period. The caldera forming eruptions of the Campanian Ignimbrite (CI) and Neapolitan Yellow Tuff (NYT) are of particular importance for tephrostratigraphy in Europe. Other key eruptions from this source include the Pomici Principali (PP) and the Tufi Biancastri eruptions. We combine analyses of fresh glasses from proximal locations (i.e., juvenile clasts in proximal flow and fall deposits) with data for key tephra layers from Lago Grande di Monticchio, 120km to the east. The micron-beam major (EMPA) and trace (LA-ICP-MS) element glass dataset allows us to: (a) distinguish between tephra units produced from the Phlegraean Fields before and during the CI eruption (CI-series), and before and during the NYT and PP eruptions (NYT-series/PP); (b) discriminate between the CI and the geochemically similar Pre-CI pyroclastic deposits; (c) separate the NYT from Pre-NYT tephra units, although both major and trace elements do show significant overlap. The complex compositional overlap between Pre-NYT tephras may present a problem for tephra correlations in the 14–39ka time window and may have resulted in incorrect proximal–distal and distal–distal correlations. The diagnostic chemical criteria detailed herein permits more accurate matching of distal tephras with their proximal equivalents and hence will improve chronostratigraphy of distal settings and give insight into tephra dispersal. We show that the dispersal of PP tephra was more limited than previously thought. The surge/fall (Lower Member) and subsequent pyroclastic density current (Upper Member) phases of the NYT eruption can be recognised in distal settings. Both the NYT Lower and Upper Members are found in distal localities to the east of the Phlegraean Fields, however the Lower Member is found in the absence of the Upper Member in locations to the far north of Phlegraean Fields. Chemical compositions of the Plinian and ignimbrite phases of the CI eruption overlap extensively, but can be distinguished on a plot of Zr–Th.

SINGLE REDUCT GENERATION BASED ON RELATIVE INDISCERNIBILITY OF ROUGH SET THEORY

In real world everything is an object which represents particular classes. Every object can be fully described by its attributes. Any real world dataset contains large number of attributes and objects. Classifiers give poor performance when these huge datasets are given as input to it for proper classification. So from these huge dataset most useful attributes need to be extracted that contribute the maximum to the decision. In the paper, attribute set is reduced by generating reducts using the indiscernibility relation of Rough Set Theory (RST). The method measures similarity among the attributes using relative indiscernibility relation and computes attribute similarity set. Then the set is minimized and an attribute similarity table is constructed from which attribute similar to maximum number of attributes is selected so that the resultant minimum set of selected attributes (called reduct) cover all attributes of the attribute similarity table. The method has been applied on glass dataset collected from the UCI repository and the classification accuracy is calculated by various classifiers. The result shows the efficiency of the proposed method.

Single Reduct Generation Based on Relative Indiscernibility of Rough Set Theory

In real world everything is an object which represents particular classes. Every object can be fully described by its attributes. Any real world dataset contains large number of attributes and objects. Classifiers give poor performance when these huge datasets are given as input to it for proper classification. So from these huge dataset most useful attributes need to be extracted that contribute the maximum to the decision. In the paper, attribute set is reduced by generating reducts using the indiscernibility relation of Rough Set Theory (RST). The method measures similarity among the attributes using relative indiscernibility relation and computes attribute similarity set. Then the set is minimized and an attribute similarity table is constructed from which attribute similar to maximum number of attributes is selected so that the resultant minimum set of selected attributes (called reduct) cover all attributes of the attribute similarity table. The method has been applied on glass dataset collected from the UCI repository and the classification accuracy is calculated by various classifiers. The result shows the efficiency of the proposed method.

Class description using partial coverage of subspaces

A good description of a class should be (reasonably) accurate and interpretable. Previous works address this class-description problem by either analyzing the correlation of each attribute with the class, or by producing rules as in building a classifier. These solutions suffer from issues in accuracy and interpretability.A sentence is usually defined as a disjunction or conjunction of several terms, each of which specifies a constraint (range/set of values) on an attribute. From the data analysis point of view, a sentence specifies a subspace in the database. In this paper, we create a richer yet interpretable form of a sentence. Here, a sentence describes an object if any k attributes of that object satisfy the specified constraints, or in other words, the object is partially covered by the subspace. Since this simple enhancement subsumes rules used in previous solutions, descriptions based on such sentences are provably better.To that end, we design Pub, an algorithm that produces descriptions with our form of sentences. Theoretically, while constructing a sentence (within the description), Pub finds the optimal range/set of values for each attribute in linear time. Empirically, we show that Pub is efficient, and able to produce more accurate, concise and interpretable descriptions than current approaches on various real datasets. We also perform an illustrative case study on the Glass dataset, providing some useful insights.

Application of a hybrid of genetic algorithm and particle swarm optimization algorithm for order clustering

This study proposes an evolutionary-based clustering algorithm based on a hybrid of genetic algorithm (GA) and particle swarm optimization algorithm (PSOA) for order clustering in order to reduce surface mount technology (SMT) setup time. Simulational results via Iris, Glass, Vowel and Wine benchmark data sets indicate that the proposed evolutionary-based clustering algorithm is more accurate than the GA-based and PSOA-based clustering algorithms. In addition, the model evaluation results which use order information provided by an international industrial personal computer (PC) manufacturer show that the proposed algorithm is also superior to GA-based and PSOA-based clustering algorithms. Through order clustering, scheduling orders that belong to the same cluster together can reduce production time as well as machine idle time.