Search results for: Latent values/vectors

Authors: Kyung Bae Park, Sung Ho Ha

Abstract:

Online user-generated contents (UGC) significantly change the way customers behave (e.g., shop, travel), and a pressing need to handle the overwhelmingly plethora amount of various UGC is one of the paramount issues for management. However, a current approach (e.g., sentiment analysis) is often ineffective for leveraging textual information to detect the problems or issues that a certain management suffers from. In this paper, we employ text mining of Latent Dirichlet Allocation (LDA) on a popular online review site dedicated to complaint from users. We find that the employed LDA efficiently detects customer complaints, and a further inspection with the visualization technique is effective to categorize the problems or issues. As such, management can identify the issues at stake and prioritize them accordingly in a timely manner given the limited amount of resources. The findings provide managerial insights into how analytics on social media can help maintain and improve their reputation management. Our interdisciplinary approach also highlights several insights by applying machine learning techniques in marketing research domain. On a broader technical note, this paper illustrates the details of how to implement LDA in R program from a beginning (data collection in R) to an end (LDA analysis in R) since the instruction is still largely undocumented. In this regard, it will help lower the boundary for interdisciplinary researcher to conduct related research.

Keywords: Latent Dirichlet allocation, R program, text mining, topic model, user generated contents, visualization.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1216

Authors: Parvinder S. Sandhu, Hardeep Singh

Abstract:

Automatic reusability appraisal could be helpful in evaluating the quality of developed or developing reusable software components and in identification of reusable components from existing legacy systems; that can save cost of developing the software from scratch. But the issue of how to identify reusable components from existing systems has remained relatively unexplored. In this paper, we have mentioned two-tier approach by studying the structural attributes as well as usability or relevancy of the component to a particular domain. Latent semantic analysis is used for the feature vector representation of various software domains. It exploits the fact that FeatureVector codes can be seen as documents containing terms -the idenifiers present in the components- and so text modeling methods that capture co-occurrence information in low-dimensional spaces can be used. Further, we devised Neuro- Fuzzy hybrid Inference System, which takes structural metric values as input and calculates the reusability of the software component. Decision tree algorithm is used to decide initial set of fuzzy rules for the Neuro-fuzzy system. The results obtained are convincing enough to propose the system for economical identification and retrieval of reusable software components.

Keywords: Clustering, ID3, LSA, Neuro-fuzzy System, SVD

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1662

Authors: Rohit Gupta, Ravinder Agarwal

Abstract:

Surface electromyographic (sEMG) signal has the potential to identify the human activities and intention. This potential is further exploited to control the artificial limbs using the sEMG signal from residual limbs of amputees. The paper deals with the development of multichannel cost efficient sEMG signal interface for research application, along with evaluation of proposed class dependent statistical approach of the feature selection method. The sEMG signal acquisition interface was developed using ADS1298 of Texas Instruments, which is a front-end interface integrated circuit for ECG application. Further, the sEMG signal is recorded from two lower limb muscles for three locomotions namely: Plane Walk (PW), Stair Ascending (SA), Stair Descending (SD). A class dependent statistical approach is proposed for feature selection and also its performance is compared with 12 preexisting feature vectors. To make the study more extensive, performance of five different types of classifiers are compared. The outcome of the current piece of work proves the suitability of the proposed feature selection algorithm for locomotion recognition, as compared to other existing feature vectors. The SVM Classifier is found as the outperformed classifier among compared classifiers with an average recognition accuracy of 97.40%. Feature vector selection emerges as the most dominant factor affecting the classification performance as it holds 51.51% of the total variance in classification accuracy. The results demonstrate the potentials of the developed sEMG signal acquisition interface along with the proposed feature selection algorithm.

Keywords: Classifiers, feature selection, locomotion, sEMG.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1491

Authors: Safdar Muhammad Usman, Youji Kohda, Katsuhiro Umemoto

Abstract:

Electronic market place plays an important intermediary role for connecting dealers and retail customers. The main aim of this paper is to design a value co-creation model in used-car auctions. More specifically, the study has been designed in order to describe the process of value co-creation in used-car auctions, to explore the co-created values in used-car auctions, and finally conclude the paper indicating the future research directions. Our analysis shows that economic values as well as non-economic values are co-created in used-car auctions. In addition, this paper contributes to the academic society broadening the view of value co-creation in service science.

Keywords: Value co-creation, Used-car auctions, Non-economic values, Service science.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2473

Authors: Odón R. Sánchez-Ccoyllo, Elizabeth Ayma-Choque, Alan Llacza

Abstract:

Sulphur dioxide (SO₂) and surface-ozone (O₃) concentrations are associated with diseases. The objective of this research is to evaluate the effectiveness of the air-quality Weather Research and Forecasting model coupled to Chemistry (WRF-Chem) model with a horizontal resolution of 5 km x 5 km. For this purpose, the measurements of the hourly SO₂ and O₃ concentrations available in three air quality monitoring stations in Lima, Peru were used for the purpose of validating the simulations of the SO₂ and O₃ concentrations obtained with the WRF-Chem model in February 2018. For the quantitative evaluation of the simulations of these gases, statistical techniques were implemented, such as the average of the simulations; the average of the measurements; the Mean Bias (MeB); the Mean Error (MeE); and the Root Mean Square Error (RMSE). The results of these statistical metrics indicated that the simulated SO₂ and O₃ values over-predicted the SO₂ and O₃ measurements. For the SO₂ concentration, the MeB values varied from 0.58 to 26.35 µg/m³; the MeE values varied from 8.75 to 26.5 µg/m³; the RMSE values varied from 13.3 to 31.79 µg/m³; while for O₃ concentrations the statistical values of the MeB varied from 37.52 to 56.29 µg/m³; the MeE values varied from 37.54 to 56.70 µg/m³; the RMSE values varied from 43.05 to 69.56 µg/m³.

Keywords: Ground-ozone, Lima, Sulphur dioxide, WRF-Chem.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 364

Authors: Amirali Ebrahimi, T. A. G. Langrish

Abstract:

The effect of various humidities on process yields and degrees of crystallinity for spray-dried powders from spray drying of lactose with humid air in a straight-through system have been studied. It has been suggested by Williams–Landel–Ferry kinetics (WLF) that a higher particle temperature and lower glass-transition temperature would increase the crystallization rate of the particles during the spray-drying process. Freshly humidified air produced by a Buchi-B290 spray dryer as a humidifier attached to the main spray dryer decreased the particle glass-transition temperature (Tg), while allowing the particle temperature (Tp) to reach higher values by using an insulated drying chamber. Differential scanning calorimetry (DSC) and moisture sorption analysis were used to measure the degree of crystallinity for the spray-dried lactose powders. The results showed that higher Tp-Tg, as a result of applying humid air, improved the process yield from 21 ± 4 to 26 ± 2% and crystallinity of the particles by decreasing the latent heat of crystallization from 43 ± 1 to 30 ± 11 J/g and the sorption peak height from 7.3 ± 0.7% to 6 ± 0.7%.

Keywords: Lactose, crystallization, spray drying, humid air.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 3439

Authors: Dias de J. C. Lilian, Lobato M. F. Fábio, de Santana L. Ádamo

Abstract:

The occurrence of missing values in database is a serious problem for Data Mining tasks, responsible for degrading data quality and accuracy of analyses. In this context, the area has shown a lack of standardization for experiments to treat missing values, introducing difficulties to the evaluation process among different researches due to the absence in the use of common parameters. This paper proposes a testbed intended to facilitate the experiments implementation and provide unbiased parameters using available datasets and suited performance metrics in order to optimize the evaluation and comparison between the state of art missing values treatments.

Keywords: Data imputation, data mining, missing values treatment, testbed.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1513

Authors: A. I. Augie, M. Saleh, A. A. Gado

Abstract:

In this research, Self-Potential (SP) method was employed to locate anomalous electrical conductivity located in Goronyo area and also to determine the condition of the embankment of the dam. SP data were plotted against distance along with the profile and spacing of electrode using surfer software (version 12). High and low zones of SP values were identified along the right and left abutments of the dam reservoir. The regions with high SP values were described to be high tendency of fluid flow associate with wet sandy soil. These zones have the SP values ranging from 200 mV and above. High SP values were due to the high moisture content that may lead to the seepage of water leaking through this zone. The zones with high SP values occupied Profiles S1, S2, S3, S4 and S5 indicating the presence of potential seepage paths within the subsurface of the embankment. These regions of seepage were identified as weak zones and potential pathways through which water could be lost from the dam reservoir. The SP values for the regions range from 250 m to 400 m (S1), 306 m to 400 m (S2), 192 m to 400 m (S3), 48 m to 200 m (S4) and 7 m to 170 m (S5) with their corresponding maximum depths of 30 m, 28 m, 28 m, 30 m and 26 m respectively. However, zones of low SP values in the overburden were observed which shows the presence of intact regions, which may be due to the compactness and dryness around the dam. The weak zones were considered as geological features (such as fractures, joints, and faults) that have undermined the integrity of the dam structure, which has led to the abnormal seepage.

Keywords: Self-potential, subsurface, seepage, condition and dam.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 690

Authors: Vanchai Laemlaksakul

Abstract:

This research was to evaluate a technical feasibility of making single-layer experimental particleboard panels from bamboo waste (Dendrocalamus asper Backer) by converting bamboo into strips, which are used to make laminated bamboo furniture. Variable factors were density (600, 700 and 800 kg/m3) and temperature of condition (25, 40 and 55 °C). The experimental panels were tested for their physical and mechanical properties including modulus of elasticity (MOE), modulus of rupture (MOR), internal bonding strength (IB), screw holding strength (SH) and thickness swelling values according to the procedures defined by Japanese Industrial Standard (JIS). The test result of mechanical properties showed that the MOR, MOE and IB values were not in the set criteria, except the MOR values at the density of 700 kg/m3 at 25 °C and at the density of 800 kg/m3 at 25 and 40 °C, the IB values at the density of 600 kg/m3, at 40 °C, and at the density of 800 kg/m3 at 55 °C. The SH values had the test result according to the set standard, except with the density of 600 kg/m3, at 40 and 55 °C. Conclusively, a valuable renewable biomass, bamboo waste could be used to manufacture boards.

Keywords: Particleboard, Urea Formaldehyde Resin, BambooWaste

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 5646

Authors: F. Kadioglu, M. Z. Polat, A. R. Gunay

Abstract:

Correct measurement of a structural damping value is an important issue for the reliable design of the components exposed to vibratory and noise conditions. As far as a vibrating beam technique is concerned, the specimens under the test somehow are interacted with measuring and exciting devices, and also with boundary conditions of the test set-up. The aim of this study is to propose a vibrating beam method that offers a non-contact dynamic measurement of solid beam specimens. To evaluate the possible effects of the clamped portion of the specimens with clamped-free ends on the dynamic values (damping and the elastic modulus), the same measuring devices were used, and the results were compared to those with the free-free ends. First, the governing equations of beam specimens related to the free-free and clamped-free boundary conditions were expressed to be able to find their natural frequencies, flexural modulus and damping values. To get a clear idea of the sensitivity of the boundary conditions to the damping values at low, medium and high levels, representative materials were subjected to the tests. The results show that the specimens with low damping values are especially sensitive to the boundary conditions and that the most reliable structural damping values are obtained for the specimens with free-free ends. For the damping values at the low levels, a deviation of about 368% was obtained between the specimens with free-free and clamped-free ends, yet, for those having high inherent damping values, comparable results were obtained. It was obvious that the set-up with clamped-free boundary conditions was not able to produce correct/reliable damping values for the specimens with low inherent damping. 

Keywords: Boundary conditions, damping, dynamic values, non-contact measuring systems, vibrating beam technique.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 468

Authors: Xiaocong Liu, Huazhen Wang, Ting He, Xiaozheng Li, Weihan Zhang, Jian Chen

Abstract:

The utilization of electronic medical record (EMR) data to establish the disease diagnosis model has become an important research content of biomedical informatics. Deep learning can automatically extract features from the massive data, which brings about breakthroughs in the study of EMR data. The challenge is that deep learning lacks semantic knowledge, which leads to impracticability in medical science. This research proposes a method of incorporating lexical-semantic knowledge from abundant entities into a convolutional neural network (CNN) framework for pediatric disease diagnosis. Firstly, medical terms are vectorized into Lexical Semantic Vectors (LSV), which are concatenated with the embedded word vectors of word2vec to enrich the feature representation. Secondly, the semantic distribution of medical terms serves as Semantic Decision Guide (SDG) for the optimization of deep learning models. The study evaluates the performance of LSV-SDG-CNN model on four kinds of Chinese EMR datasets. Additionally, CNN, LSV-CNN, and SDG-CNN are designed as baseline models for comparison. The experimental results show that LSV-SDG-CNN model outperforms baseline models on four kinds of Chinese EMR datasets. The best configuration of the model yielded an F1 score of 86.20%. The results clearly demonstrate that CNN has been effectively guided and optimized by lexical-semantic knowledge, and LSV-SDG-CNN model improves the disease classification accuracy with a clear margin.

Keywords: lexical semantics, feature representation, semantic decision, convolutional neural network, electronic medical record

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 594

Authors: Loai AbdAllah, Ilan Shimshoni

Abstract:

Missing values in data are common in real world applications. Since the performance of many data mining algorithms depend critically on it being given a good metric over the input space, we decided in this paper to define a distance function for unlabeled datasets with missing values. We use the Bhattacharyya distance, which measures the similarity of two probability distributions, to define our new distance function. According to this distance, the distance between two points without missing attributes values is simply the Mahalanobis distance. When on the other hand there is a missing value of one of the coordinates, the distance is computed according to the distribution of the missing coordinate. Our distance is general and can be used as part of any algorithm that computes the distance between data points. Because its performance depends strongly on the chosen distance measure, we opted for the k nearest neighbor classifier to evaluate its ability to accurately reflect object similarity. We experimented on standard numerical datasets from the UCI repository from different fields. On these datasets we simulated missing values and compared the performance of the kNN classifier using our distance to other three basic methods. Our  experiments show that kNN using our distance function outperforms the kNN using other methods. Moreover, the runtime performance of our method is only slightly higher than the other methods.

Keywords: Missing values, Distance metric, Bhattacharyya distance.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2751

Authors: Sarah Beaver, Renee Bryce

Abstract:

Audiograms detect hearing impairment, but missing values pose problems. This work explores imputations in an attempt to improve accuracy. This work implements Linear Regression, Lasso, Linear Support Vector Regression, Bayesian Ridge, K Nearest Neighbors (KNN), and Random Forest machine learning techniques to impute audiogram frequencies ranging from 125 Hz to 8000 Hz. The data contain patients who had or were candidates for cochlear implants. Accuracy is compared across two different Nested Cross-Validation k values. Over 4000 audiograms were used from 800 unique patients. Additionally, training on data combines and compares left and right ear audiograms versus single ear side audiograms. The accuracy achieved using Root Mean Square Error (RMSE) values for the best models for Random Forest ranges from 4.74 to 6.37. The R2 values for the best models for Random Forest ranges from .91 to .96. The accuracy achieved using RMSE values for the best models for KNN ranges from 5.00 to 7.72. The R2 values for the best models for KNN ranges from .89 to .95. The best imputation models received R2 between .89 to .96 and RMSE values less than 8dB. We also show that the accuracy of classification predictive models performed better with our imputation models versus constant imputations by a two percent increase.

Keywords: Machine Learning, audiograms, data imputations, single imputations.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 160

Authors: Orkide Donma, Mustafa M. Donma

Abstract:

Basal metabolic rate (BMR) is widely used and an accepted measure of energy expenditure. Its principal determinant is body mass. However, this parameter is also correlated with a variety of other factors. The objective of this study is to measure BMR and compare it with the values obtained from predictive equations in adults classified according to their body mass index (BMI) values. 276 adults were included into the scope of this study. Their age, height and weight values were recorded. Five groups were designed based on their BMI values. First group (n = 85) was composed of individuals with BMI values varying between 18.5 and 24.9 kg/m². Those with BMI values varying from 25.0 to 29.9 kg/m² constituted Group 2 (n = 90). Individuals with 30.0-34.9 kg/m², 35.0-39.9 kg/m², > 40.0 kg/m² were included in Group 3 (n = 53), 4 (n = 28) and 5 (n = 20), respectively. The most commonly used equations to be compared with the measured BMR values were selected. For this purpose, the values were calculated by the use of four equations to predict BMR values, by name, introduced by Food and Agriculture Organization (FAO)/World Health Organization (WHO)/United Nations University (UNU), Harris and Benedict, Owen and Mifflin. Descriptive statistics, ANOVA, post-Hoc Tukey and Pearson’s correlation tests were performed by a statistical program designed for Windows (SPSS, version 16.0). p values smaller than 0.05 were accepted as statistically significant. Mean ± SD of groups 1, 2, 3, 4 and 5 for measured BMR in kcal were 1440.3 ± 210.0, 1618.8 ± 268.6, 1741.1 ± 345.2, 1853.1 ± 351.2 and 2028.0 ± 412.1, respectively. Upon evaluation of the comparison of means among groups, differences were highly significant between Group 1 and each of the remaining four groups. The values were increasing from Group 2 to Group 5. However, differences between Group 2 and Group 3, Group 3 and Group 4, Group 4 and Group 5 were not statistically significant. These insignificances were lost in predictive equations proposed by Harris and Benedict, FAO/WHO/UNU and Owen. For Mifflin, the insignificance was limited only to Group 4 and Group 5. Upon evaluation of the correlations of measured BMR and the estimated values computed from prediction equations, the lowest correlations between measured BMR and estimated BMR values were observed among the individuals within normal BMI range. The highest correlations were detected in individuals with BMI values varying between 30.0 and 34.9 kg/m². Correlations between measured BMR values and BMR values calculated by FAO/WHO/UNU as well as Owen were the same and the highest. In all groups, the highest correlations were observed between BMR values calculated from Mifflin and Harris and Benedict equations using age as an additional parameter. In conclusion, the unique resemblance of the FAO/WHO/UNU and Owen equations were pointed out. However, mean values obtained from FAO/WHO/UNU were much closer to the measured BMR values. Besides, the highest correlations were found between BMR calculated from FAO/WHO/UNU and measured BMR. These findings suggested that FAO/WHO/UNU was the most reliable equation, which may be used in conditions when the measured BMR values are not available.

Keywords: Adult, basal metabolic rate, FAO/WHO/UNU, obesity, prediction equations.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1010

Authors: Funmilola O. Omoya

Abstract:

Muscid flies are known to be vectors of disease agents and species that annoy humans and domesticated animals. An example of these flies is Musca domestica (house fly) whose adult and immature stages occur in a variety of filthy organic substances including household garbage and animal manures. They contribute to microbial contamination of foods. It is therefore imperative to control these flies as a result of their role in Public health. The second and third instars of Musca domestica (Linn) were infected with varying cell loads of Bacillus subtilis in vitro for a period of 48 hours to evaluate its larvicidal activities. Mortality of the larvae increased with incubation period after treatment with the varying cell loads. Investigation revealed that the second instars larvae were more susceptible to treatment than the third instars treatments. Values obtained from the third instar group were significantly different (P<0.05) from those obtained from the second instars group in all the treatments. Lethal concentration (LC₅₀) at 24 hours for 2^nd instars was 2.35 while LC₅₀ at 48 hours was 4.31.This study revealed that Bacillus subtilis possess good larvicidal potential for use in the control of Musca domestica in poultry farms.

Keywords: Bacillus subtilis, larvicidal activities, Musca domestica, poultry farms.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2237

Authors: Andreea Muntean, Cornelia Măgureanu

Abstract:

This article presents the analysis of experimental values regarding cracking pattern, specific strains and deformability for reinforced high strength concrete beams. The beams have the concrete class C80/95 and a longitudinal reinforcement ratio of 2.01%, respectively 3.39%. The elements were subjected to flexure under static short-term and long-term loading. The experimental values are compared with calculation values using the design relationships according to Eurocode 2.

Keywords: High strength concrete, beams, flexure.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1593

Authors: Z. Kurtaslan, H. Hakan Okay, E. Can Dönmez, I. Kuçukdoğan

Abstract:

Music education aims to bring up individuals most appropriately and to advanced levels as a balanced whole physically, cognitively, affectively, and kinesthetically while making a major contribution to the physical and spiritual development of the individual. The most crucial aim of music education, an influential education medium per se, is to make music be loved; yet, among its educational aims are concepts such as affinity, friendship, goodness, philanthropy, responsibility, and respect all extremely crucial bringing up individuals as a balanced whole. One of the most essential assets of the music education is the training of making music together, solidifying musical knowledge and enabling the acquisition of cooperation. This habit requires internalization of values like responsibility, patience, cooperativeness, respect, self-control, friendship, and fairness. If musicians lack these values, the ensemble will become after some certain time a cacophony. In this qualitative research, the attitudes of music teacher candidates in orchestra/chamber music classes will be examined in terms of values.

Keywords: Education, music, orchestra/chamber music, values.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 872

Authors: Roman Marsalek

Abstract:

In this paper, snow samples containing dust particles from several sampling points around the city of Ostrava were analyzed. The pH values of sampled snow were measured and solid particles analyzed. Particle size, zeta potential and content of selected heavy metals were determined in solid particles. The pH values of most samples lay in the slightly acid region. Mean values of particle size ranged from 290.5 to 620.5 nm. Zeta potential values varied between -5 and -26.5 mV. The following heavy metal concentration ranges were found: copper 0.08-0.75 mg/g, lead 0.05-0.9 mg/g, manganese 0.45-5.9 mg/g and iron 25.7-280.46 mg/g. The highest values of copper and lead were found in the vicinity of busy crossroads, and on the contrary, the highest levels of manganese and iron were detected close to a large steelworks. The proportion between pH values, zeta potentials, particle sizes and heavy metal contents was established. Zeta potential decreased with rising pH values and, simultaneously, heavy metal content in solid particles increased. At the same time, higher metal content corresponded to lower particle size.

Keywords: Dust, snow, zeta potential, particles size distribution, heavy metals.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1983

Authors: Isaac K. E. Ampomah, Seong-Bae Park, Sang-Jo Lee

Abstract:

Over the past decade, there have been promising developments in Natural Language Processing (NLP) with several investigations of approaches focusing on Recognizing Textual Entailment (RTE). These models include models based on lexical similarities, models based on formal reasoning, and most recently deep neural models. In this paper, we present a sentence encoding model that exploits the sentence-to-sentence relation information for RTE. In terms of sentence modeling, Convolutional neural network (CNN) and recurrent neural networks (RNNs) adopt different approaches. RNNs are known to be well suited for sequence modeling, whilst CNN is suited for the extraction of n-gram features through the filters and can learn ranges of relations via the pooling mechanism. We combine the strength of RNN and CNN as stated above to present a unified model for the RTE task. Our model basically combines relation vectors computed from the phrasal representation of each sentence and final encoded sentence representations. Firstly, we pass each sentence through a convolutional layer to extract a sequence of higher-level phrase representation for each sentence from which the first relation vector is computed. Secondly, the phrasal representation of each sentence from the convolutional layer is fed into a Bidirectional Long Short Term Memory (Bi-LSTM) to obtain the final sentence representations from which a second relation vector is computed. The relations vectors are combined and then used in then used in the same fashion as attention mechanism over the Bi-LSTM outputs to yield the final sentence representations for the classification. Experiment on the Stanford Natural Language Inference (SNLI) corpus suggests that this is a promising technique for RTE.

Keywords: Deep neural models, natural language inference, recognizing textual entailment, sentence-to-sentence relation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1454

Authors: Cheng Zeng, George Michailidis, Hitoshi Iyatomi, Leo L Duan

Abstract:

The conditional density characterizes the distribution of a response variable y given other predictor x, and plays a key role in many statistical tasks, including classification and outlier detection. Although there has been abundant work on the problem of Conditional Density Estimation (CDE) for a low-dimensional response in the presence of a high-dimensional predictor, little work has been done for a high-dimensional response such as images. The promising performance of normalizing flow (NF) neural networks in unconditional density estimation acts a motivating starting point. In this work, we extend NF neural networks when external x is present. Specifically, they use the NF to parameterize a one-to-one transform between a high-dimensional y and a latent z that comprises two components [zP , zN]. The zP component is a low-dimensional subvector obtained from the posterior distribution of an elementary predictive model for x, such as logistic/linear regression. The zN component is a high-dimensional independent Gaussian vector, which explains the variations in y not or less related to x. Unlike existing CDE methods, the proposed approach, coined Augmented Posterior CDE (AP-CDE), only requires a simple modification on the common normalizing flow framework, while significantly improving the interpretation of the latent component, since zP represents a supervised dimension reduction. In image analytics applications, AP-CDE shows good separation of x-related variations due to factors such as lighting condition and subject id, from the other random variations. Further, the experiments show that an unconditional NF neural network, based on an unsupervised model of z, such as Gaussian mixture, fails to generate interpretable results.

Keywords: Conditional density estimation, image generation, normalizing flow, supervised dimension reduction.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 166

Authors: Jay M. Joshi, Upena D. Dalal

Abstract:

Multimedia information availability has increased dramatically with the advent of video broadcasting on handheld devices. But with this availability comes problems of maintaining the security of information that is displayed in public. ISMA Encryption and Authentication (ISMACryp) is one of the chosen technologies for service protection in DVB-H (Digital Video Broadcasting- Handheld), the TV system for portable handheld devices. The ISMACryp is encoded with H.264/AVC (advanced video coding), while leaving all structural data as it is. Two modes of ISMACryp are available; the CTR mode (Counter type) and CBC mode (Cipher Block Chaining) mode. Both modes of ISMACryp are based on 128- bit AES algorithm. AES algorithms are more complex and require larger time for execution which is not suitable for real time application like live TV. The proposed system aims to gain a deep understanding of video data security on multimedia technologies and to provide security for real time video applications using selective encryption for H.264/AVC. Five level of security proposed in this paper based on the content of NAL unit in Baseline Constrain profile of H.264/AVC. The selective encryption in different levels provides encryption of intra-prediction mode, residue data, inter-prediction mode or motion vectors only. Experimental results shown in this paper described that fifth level which is ISMACryp provide higher level of security with more encryption time and the one level provide lower level of security by encrypting only motion vectors with lower execution time without compromise on compression and quality of visual content. This encryption scheme with compression process with low cost, and keeps the file format unchanged with some direct operations supported. Simulation was being carried out in Matlab.

Keywords: AES-128, CAVLC, H.264, ISMACryp

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2049

Authors: Kasim Terzic, Bernd Neumann

Abstract:

Scene interpretation systems need to match (often ambiguous) low-level input data to concepts from a high-level ontology. In many domains, these decisions are uncertain and benefit greatly from proper context. This paper demonstrates the use of decision trees for estimating class probabilities for regions described by feature vectors, and shows how context can be introduced in order to improve the matching performance.

Keywords: Classification, Decision Trees, Interpretation, Vision

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1300

Authors: N. Z. Baderisham, H. A. Hamid, N. Osman

Abstract:

The powders of Ba(Ce_1-xZr_x)_0.90Y_0.1O_3-δ (BCZY) with 0.2 ≤ x ≤ 0.6 have been prepared by a modified sol-gel method. Triethylenetetramine (TETA) was employed as chelating agent. Phase formation of calcined powders at 1100^oC and sintered pellets at 1400^oC of BCZY were examined by an X-ray diffractrometer (XRD). XRD results showed the calcined powder and sintered pellet formed a single perovskite phase over the entire range of x values. As the amount of zirconium substitution (x values) increase, the main peaks are shifted to the higher 2theta values which suggest a complete substitution of zirconium into cerium sites. All the obtained calcined powders and sintered pellets possess cubic structure (Pm-3m) at all x values.

Keywords: Structure, phase formation, modified sol-gel, cerate-zirconate electrolye, XRD.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 4231

Authors: Jung-Ho Moon, Tae Kwon Ha

Abstract:

This paper presents a car parking monitoring system using wireless sensor networks. Multiple sensor nodes and a sink node, a gateway, and a server constitute a wireless network for monitoring a parking lot. Each of the sensor nodes is equipped with a 3-axis AMR sensor and deployed in the center of a parking space. Each sensor node reads its sensor values periodically and transmits the data to the sink node if the current and immediate past sensor values show a difference exceeding a threshold value. The sensor nodes and sink node use the 448 MHz band for wireless communication. Since RF transmission only occurs when sensor values show abrupt changes, the number of RF transmission operations is reduced and battery power can be conserved. The data from the sensor nodes reach the server via the sink node and gateway. The server determines which parking spaces are taken by cars based upon the received sensor data and reference values. The reference values are average sensor values measured by each sensor node when the corresponding parking spot is not occupied by a vehicle. Because the decision making is done by the server, the computational burden of the sensor node is relieved, which helps reduce the duty cycle of the sensor node.

Keywords: Car parking monitoring, magnetometer, sensor node, wireless sensor network.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 8177

Authors: Xu Zhao

Abstract:

Text categorization techniques are widely used to many Information Retrieval (IR) applications. In this paper, we proposed a simple but efficient method that can automatically find the relationship between any pair of terms and documents, also an indexing matrix is established for text categorization. We call this method Indexing Matrix Categorization Machine (IMCM). Several experiments are conducted to show the efficiency and robust of our algorithm.

Keywords: Text categorization, Sub-space learning, Latent Semantic Space

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1467

Authors: E. Binaeian, A.M. Rashidi, H. Attar

Abstract:

A comparative evaluation of acute toxicity of synthesized nano silvers using two different procedures (biological and chemical reduction methods) and silver ions on bacteria Vibrio fischeri was investigated. The bacterial light inhibition test as a toxicological endpoint was used by applying of a homemade luminometer. To compare the toxicity effects as a quantitative parameter, a nominal effective concentrations (EC) of chemicals and a susceptibility constant (Z-value) of bacteria, after 5 min and 30 min exposure times, were calculated. After 5 and 30 min contact times, the EC50 values of two silver nanoparticles and the EC20 values were about similar. It demonstrates that toxicity of silvers was independent of their procedure. The EC values of nanoparticles were larger than those of the silver ions. The susceptibilities(Z- Values) of V.fischeri (L/mg) to the silver ions were greater than those of the nano silvers. According to the EC and Z values, the toxicity of silvers decreased in the following order: Silver ions >> silver nanoparticles from chemical reduction method ~ silver nanoparticles from biological method.

Keywords: Bioluminescence, Luminometer, silver nano particles, Toxicity, Vibrio fischeri

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 3079

Authors: Ferhat Kadioglu, Ahmet Reha Gunay

Abstract:

Correct measurement of a structural damping value is an important issue for the reliable design of the components exposed to vibratory and noise conditions. As far as a vibrating beam technique is concerned, the specimens under the test somehow are interacted with measuring and exciting devices and also with boundary conditions of the test set-up. The aim of this study is to propose a vibrating beam method that offers a non-contact dynamic measurement of solid beam specimens. To evaluate possible effects of the clamped portion of the specimens with clamped-free ends on the dynamic values (damping and the elastic modulus), the same measuring devices were used, and the results were compared to those with the free-free ends. To get clear idea about the sensitivity of the boundary conditions to the damping values at low, medium and high levels, representative materials were subjected to the tests. The results show that the specimens with low damping values are especially sensitive to the boundary conditions and the most reliable structural damping values are obtained for the specimens with free-free ends. For the damping values at the low levels, a deviation of about 368% was obtained between the specimens with free-free and clamped-free ends, yet, for those having high inherent damping values, comparable results were obtained.

Keywords: Vibrating beam technique, dynamic values, damping, boundary conditions, non-contact measuring systems.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 294

Authors: Takatoshi Ishii, Koji Kimita, Keiichi Muramatsu, Yoshiki Shimomura

Abstract:

To increase the quality of learning, teacher and learner need mutual effort for realization of educational value. For this purpose, we need to manage the co-creational education among teacher and learners. In this research, we try to find a feature of co-creational education. To be more precise, we analyzed learners’ reports by natural language processing, and extract some features that describe the state of the co-creational education.

Keywords: Co-creational education, e-portfolios, ICT integration, labeled Latent Dirichlet allocation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1673

Authors: Orkide Donma, Mustafa M. Donma

Abstract:

Chronologic age (CA) of individuals is closely related to obesity and generally affects the magnitude of obesity parameters. On the other hand, close association between basal metabolic rate (BMR) and metabolic age (MA) is also a matter of concern. It is suggested that MA higher than CA is the indicator of the need to improve the metabolic rate. In this study, the aim was to assess some commonly used obesity parameters, such as obesity degree, visceral adiposity, BMR, BMR-to-weight ratio, in several groups with varying differences between MA and CA values. The study comprises adults, whose ages vary between 18 and 79 years. Four groups were constituted. Group 1, 2, 3 and 4 were composed of 55, 33, 76 and 47 adults, respectively. The individuals exhibiting -1, 0 and +1 for their MA-CA values were involved in Group 1, which was considered as the control group. Those, whose MA-CA values varying between -5 and -10 participated in Group 2. Those, whose MAs above their real ages were divided into two groups [Group 3 (MA-CA; from +5 to + 10) and Group 4 (MA-CA; from +11 to + 12)]. Body mass index (BMI) values were calculated. TANITA body composition monitor using bioelectrical impedance analysis technology was used to obtain values for obesity degree, visceral adiposity, BMR and BMR-to-weight ratio. The compiled data were evaluated statistically using a statistical package program; SPSS. Mean ± SD values were determined. Correlation analyses were performed. The statistical significance degree was accepted as p < 0.05. The increase in BMR was positively correlated with obesity degree. MAs and CAs of the groups were 39.9 ± 16.8 vs 39.9 ± 16.7 years for Group 1, 45.0 ± 15.3 vs 51.4 ± 15.7 years for Group 2, 47.2 ± 12.7 vs 40.0 ± 12.7 years for Group 3, and 53.6 ± 14.8 vs 42 ± 14.8 years for Group 4. BMI values of the groups were 24.3 ± 3.6 kg/m², 23.2 ± 1.7 kg/m², 30.3 ± 3.8 kg/m², and 40.1 ± 5.1 kg/m² for Group 1, 2, 3 and 4, respectively. Values obtained for BMR were 1599 ± 328 kcal in Group 1, 1463 ± 198 kcal in Group 2, 1652 ± 350 kcal in Group 3, and 1890 ± 360 kcal in Group 4. A correlation was observed between BMR and MA-CA values in Group 1. No correlation was detected in other groups. On the other hand, statistically significant correlations between MA-CA values and obesity degree, BMI as well as BMR/weight were found in Group 3 and in Group 4. It was concluded that upon consideration of these findings in terms of MA-CA values, BMR-to-weight ratio was found to be much more useful indicator of the severe increase in obesity development than BMR. Also, the lack of associations between MA and BMR as well as BMR-to-weight ratio emphasize the importance of consideration of MA-CA values rather than MA.

Keywords: Basal metabolic rate, chronologic age, metabolic age, obesity degree.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1051

Authors: J.-H. Moon, D.-O Kim, R.-H. Park

Abstract:

This paper presents a video matting method, which extracts the foreground and alpha matte from a video sequence. The objective of video matting is finding the foreground and compositing it with the background that is different from the one in the original image. By finding the motion vectors (MVs) using a sliced block matching algorithm (SBMA), we can extract moving regions from the video sequence under the assumption that the foreground is moving and the background is stationary. In practice, foreground areas are not moving through all frames in an image sequence, thus we accumulate moving regions through the image sequence. The boundaries of moving regions are found by Canny edge detector and the foreground region is separated in each frame of the sequence. Remaining regions are defined as background regions. Extracted backgrounds in each frame are combined and reframed as an integrated single background. Based on the estimated background, we compute the frame difference (FD) of each frame. Regions with the FD larger than the threshold are defined as foreground regions, boundaries of foreground regions are defined as unknown regions and the rest of regions are defined as backgrounds. Segmentation information that classifies an image into foreground, background, and unknown regions is called a trimap. Matting process can extract an alpha matte in the unknown region using pixel information in foreground and background regions, and estimate the values of foreground and background pixels in unknown regions. The proposed video matting approach is adaptive and convenient to extract a foreground automatically and to composite a foreground with a background that is different from the original background.

Keywords: Background estimation, Object segmentation, Blockmatching algorithm, Video matting.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1813