Computer Systems Engineering

The availability of a standard and phonetically rich speech corpus provides a common platform for comparing the performance of different speech recognition approaches and therefore is the first step for the research in a language. This work presents the development of a large multilingual speech corpus of Pashto, Urdu and English. Recordings have been made from a total of 194 speakers in the three languages, covering diverse dialects, age groups, genders and professions. Pashto and Urdu both native and non-native speakers have been considered while for English, all the speakers were non-native. The corpus comprises of three categories of phonetically rich spoken data in each language, that is, short questions regarding speaker's personal information, read speech and spontaneous speech from the domain of tourism. Although the corpus is developed primarily for research on Automatic Spoken Language Identification purpose, nevertheless, it can also be used for research on other topics such as Automatic Speech Recognition, Accent Recognition, Automatic Speaker Identification and the study of effects of non-nativeness on Language and Speaker Identification.

A training algorithm that maximizes the margin between the training patterns and the decision boundary is presented. The technique is applicable to a wide variety of classifiaction functions, including Perceptions, polynomials, and Radial Basis Functions. The effective number of parameters is adjusted automatically to match the complexity of the problem. The solution is expressed as a linear combination of supporting patterns. These are the subset of training patterns that are closest to the decision boundary. Bounds on the generalization performance based on the leave-one-out method and the VC-dimension are given. Experimental results on optical character recognition problems demonstrate the good generalization obtained when compared with other learning algorithms.

In recent years, there has been an exponential growth in the number of complex documents and texts that require a deeper understanding of machine learning methods to be able to accurately classify texts in many applications. Many machine learning approaches have achieved surpassing results in natural language processing. The success of these learning algorithms relies on their capacity to understand complex models and non-linear relationships within data. However, finding suitable structures, architectures, and techniques for text classification is a challenge for researchers. In this paper, a brief overview of text classification algorithms is discussed. This overview covers different text feature extractions, dimensionality reduction methods, existing algorithms and techniques, and evaluations methods. Finally, the limitations of each technique and their application in real-world problems are discussed.

In recent years, with the rapid development of Internet Technology, text data is growing rapidly every day. Users need to filter out the information they need from a large amount of text. Therefore, automatic text classification technology can help users find information. In order to address problems, such as ignoring contextual semantic links and different vocabulary importance in traditional text classification techniques, this paper presents a vector representation of feature words based on the deep learning tool Word2vec, and the weight of the feature words is calculated by the improved TF-IDF algorithm. By multiplying the weight of the word and the word vector, the vector representation of the word is realized. Finally, each text is represented by accumulating all the word vectors. Thus, text classification is carried out.

In recent years, it has become possible to perform text classification with high accuracy by using convolutional neural networks (CNNs). Zhang et al. decomposed words into characters and classified texts using a CNN with relatively deep layers to obtain excellent classification results. However, it is often difficult to prepare a sufficient number of labeled samples for solving real-world text-classification problems. One method for handling this problem is transfer learning, which uses a network tuned for an arbitrary task as the initial network for a target task. While transfer learning is known to be effective for image recognition, for tasks in natural language processing, such as document classification, it has not yet been shown for what types of data and to what extent transfer learning is effective. In this paper, we first introduce a character-level CNN adopting the structure of a residual network to construct a network with deeper layers for Japanese text classification. We then demonstrate that we can improve classification accuracy by performing transfer learning between two particular datasets. Additionally, we propose an approach to evaluate the effectiveness of transfer learning and use it to evaluate our model.

Random forests are a combination of tree predictors such that each tree depends on the values of a random vector sampled independently and with the same distribution for all trees in the forest. The generalization error for forests converges a.s. to a limit as the number of trees in the forest becomes large. The generalization error of a forest of tree classifiers depends on the strength of the individual trees in the forest and the correlation between them. Using a random selection of features to split each node yields error rates that compare favorably to Adaboost (Y. Freund & R. Schapire, Machine Learning: Proceedings of the Thirteenth International conference, * * * , 148-156), but are more robust with respect to noise. Internal estimates monitor error, strength, and correlation and these are used to show the response to increasing the number of features used in the splitting. Internal estimates are also used to measure variable importance. These ideas are also applicable to regression.

Face detection is an important contributing factor to make computer vision applicable to problems of everyday life. Of the techniques used for face detection, two are more efficacious i.e. skin detection and violajones based face detection. These two techniques have limitations of their own. One of the most restricting factors to these techniques is the color cluster of the image. Face detection algorithms are trained on a set of images. Those images are sure to be occupying a particular color cluster of the color plane because it is not possible for the set to contain encompass the full color space. If an image occupying a color space other than that on which the algorithms is trained is an input to the algorithm, the result is not optimal then. This paper explores the effect of different color clusters on the performance of both techniques. Images of color clusters other than that on which the algorithms are trained are given to the algorithms and the results are analyzed to decide the optimality of the techniques in face of images on which the algorithms are not trained.

Viola-Jones algorithm and skin detection are the two widely used approaches for face detection. This paper aims at showing, experimentally, the trade-offs associated with each of the algorithm. The parameters used in this work are the ones encountered in real images. These include the illumination, color-space representation, face orientation, image sharpness and image complexity. Experimentally, the peculiarities associated with each algorithm are shown. Based on these experimental results, the situations are commented upon where these algorithms will be efficacious. This paper moreover aims at to provide an objective comparison of the two algorithms rather than going into the subjective details of peculiarities of any individual implementation.

In this paper, we propose a new scale-free social networks (SNs) evolution model that is based on homophily combined with preferential attachments. Our model enables the SN researchers to generate SN synthetic data for the evaluation of multi-facet SN models that are dependent on users’ attributes and similarities. Homophily is one of the key factors for interactive relationship formation in SN. The synthetic graph generated by our model is scale-invariant and has symmetric relationships. The model is dynamic and sustainable to changes in input parameters, such as number of nodes and nodes’ attributes, by conserving its structural properties. Simulation and evaluation of models for large-scale SN applications need large datasets. One way to get SN data is to generate synthetic data by using SN evolution models. Various SN evolution models are proposed to approximate the real-life SN graphs in previous research. These models are based on SN structural properties such as preferential attachment. The data generated by these models is suitable to evaluate SN models that are structure dependent but not suitable to evaluate models which depend on the SN users’ attributes and similarities. In our proposed model, users’ attributes and similarities are utilized to synthesize SN graphs. We evaluated the resultant synthetic graph by analyzing its structural properties. In addition, we validated our model by comparing its measures with the publicly available real-life SN datasets and previous SN evolution models. Simulation results show our resultant graph to be a close representation of real-life SN graphs with users’ attributes.

We proposed an application and data variations-independent, generic social media Textual Variations Handler (TVH) to deal with a wide range of noise in textual data generated in various social media (SM) applications for enhanced text analysis. The aim is to build an effective hybrid normalization technique that ensures the use of useful information of the noisy text in its intended form instead of filtering them out to analyze SM text better. The proposed TVH performs context-aware text normalization based on intended meaning to avoid the wrong word substitution. We integrate the TVH with state-of-the-art (SOTA) deep-learning-based text analysis methods to enhance their performance for noisy SM text data. The proposed scheme shows promising improvement in the text analysis of informal SM text in terms of precision, recall, accuracy, and F1-score in simulation.