Vol. 40 · No. 1

High-dimensional with limited-sample size (HDLSS) datasets exhibit two critical problems: (1) Due to the insufficiently small-sample size, there is a lack of enough samples to build classification models. Classification models with a limited-sample may lead to overfitting and produce erroneous or meaningless results. (2) The 'curse of dimensionality' phenomena is often an obstacle to the use of many methods for solving the high-dimensional with limited-sample size problem and reduces classification accuracy. This study proposes an unsupervised framework for high-dimensional limited-sample size data classification using dimension reduction based on variational autoencoder (VAE). First, the deep learning method variational autoencoder is applied to project high-dimensional data onto lower-dimensional space. Then, clustering is applied to the obtained latent-space of VAE to find the data groups and classify input data. The method is validated by comparing the clustering results with actual labels using purity, rand index, and normalized mutual information. Moreover, to evaluate the proposed model strength, we analyzed 14 datasets from the Arizona State University Digital Repository. Also, an empirical comparison of dimensionality reduction techniques shown to conclude their applicability in the high-dimensional with limited-sample size data settings. Experimental results demonstrate that variational autoencoder can achieve more accuracy than traditional dimensionality reduction techniques in high-dimensional with limited-sample-size data analysis.

Smart Grids (SG) have smart meters and advance metering infrasturutre (AMI) which generates huge data. This data can be used for predicting energy consumption using big data analytics. A very limited work has been carried out in the literature which shows the utilization of big data in energy consumption prediction. In this paper, the proposed method is based on Genetic Algorithm – Long Short Term Memory (GA-LSTM). LSTM memorises values over an arbitrary interval that manages time series data very effictively while GA is an evolutionary process that is used for optimization. GA combines with LSTM to process hyperparameters such as hidden layers, epochs, data intervals, batchsize and activation functions. Hence, GA creates a new vector for optimum solution that provides minimum error. These methods provide the best performance when compared with existing benchmarks. Moreover, GA-LSTM is used in a multi-threaded environment which increases the speed of convergence. Here, the multi-core platform is operated for solving one-dimensional GA-based inverse scattering problems. The result shows that GA-LSTM provides better convergence as compared to random approach techniques. For validating the results, Pennsylvania-New Jersey-Maryland Interconnection (PJM) energy consumption data has been used while adopting different performance evaluation metrics.

While there have been several studies related to the effect of term weighting on classification accuracy, relatively few works have been conducted on how term weighting affects the quality of keywords extracted for characterizing a document or a category (i.e., document collection). Moreover, many tasks require more complicated category structure, such as hierarchical and network category structure, rather than a flat category structure. This paper presents a qualitative and quantitative study on how term weighting affects keyword extraction in the hierarchical category structure, in comparison to the flat category structure. A hierarchical structure triggers special characteristic in assigning a set of keywords or tags to represent a document or a document collection, with support of statistics in a hierarchy, including category itself, its parent category, its child categories, and sibling categories. An enhancement of term weighting is proposed particularly in the form of a series of modified TFIDF's, for improving keyword extraction. A text collection of public-hearing opinions is used to evaluate variant TFs and IDFs to identify which types of information in hierarchical category structure are useful. By experiments, we found that the most effective IDF family, namely TF-IDFr, is identity>sibling>child>parent in order. The TF-IDFr outperforms the vanilla version of TFIDF with a centroid-based classifier.

We present a flexible human–robot interaction architecture that incorporates emotions and moods to provide a natural experience for humans. To determine the emotional state of the user, information representing eye gaze and facial expression is combined with other contextual information such as whether the user is asking questions or has been quiet for some time. Subsequently, an appropriate robot behaviour is selected from a multi-path scenario. This architecture can be easily adapted to interactions with non-embodied robots such as avatars on a mobile device or a PC. We present the outcome of evaluating an implementation of our proposed architecture as a whole, and also of its modules for detecting emotions and questions. Results are promising and provide a basis for further development.

Task Scheduling and Resource Allocation (TSRA) is the key focus of cloud computing. This paper utilizes Smart Message Passing Interface based Approach (SMPIA) and the Roulette Wheel selection method in order to determine the best Alternative Virtual Machine (AVM). To do so, the Virtual MPI Bus (VMPIB) is employed for efficient communication among Virtual Machines (VMs) using SMPIA. In this matter, SMPIA is applied on different resource allocation and task scheduling strategies. MakeSpan (MS) was chosen as an optimization factor and solutions with minimum MS value as the best task mapping performance and reduced cloud consumption. The simulation is conducted using MATLAB. The analysis proves that applying SMPIA reduced the Total Execution Time (TET) of resource allocation, maximum MS time, and increase the Resource Utilization (RU), as compared to non-SMPIA for Greedy, Max-Min, Min-Min algorithms. It is observed that SMPIA can outperform non-SMPIA. The effect of SMPIA is more obvious as change in the MS and the number of cloud workloads increase. Furthermore, regarding the TET and MS of the tasks, the SMPIA can significantly reduce the starvation problem as well as the lack of sufficient resources. In addition, this approach improves the system's performance more than the previous methods, what reflects effectiveness of the proposed approach concerning the Message Passing Interface (MPI) communication time in the network virtualization. The mentioned text mining work was prepared concurrently after practical evaluation.

Commercial banks are facing increasingly complex enterprise loan customers and businesses. It is important for banks' enterprise loan business to efficiently assess credit risks. Our study builds an enterprise credit risk assessment model based on the state and constraint of bank and customer, and get empirical researches with RF, SVM and DT algorithms. The results show that our model has excellent performance with accuracy 99

The issue of safe utilization of mobile devices is becoming an increasingly important problem, among others due to the widespread use of such devices to access sensitive data (such as electronic documents or banking data). In our work we analyze the use of biometric techniques in order to secure a mobile device, with particular emphasis on the viability of selected biometric characteristics. For this purpose, we investigate the possibility of applying machine learning models to assess the authenticity of a biometric characteristic. Results of our tests have shown that the most effective method of assessing the viability of a biometric characteristic involves blink and smile detection.

Colored Petri Nets (CP-nets or CPNs) are powerful modeling language for concurrent systems. As for CPNs' model checking, the mainstream method is unfolding that transforms a CPN into an equivalent P/T net. However the equivalent P/T net tends to be too enormous to be handled. As for checking CPN models without unfolding, we present three practical on-the-fly verification methods which are all focused on how to make state space generation more efficient. The first one is a basic one, based on a standard state space generation algorithm, but its efficiency is low. The second one is an overall improvement of the first. The third one sacrifices some applicability for higher efficiency. We implemented the three algorithms and validated great efficiency of latter two algorithms through experimental results.

Video games are complex systems that combine technical and artistic processes. The specification of this type of system is not a trivial task, making it necessary to use diagrams and charts to visually specify sets of requirements. Therefore, the underlying proposal of this work is to present an approach based on the formalism of Petri nets for aiding in the design process of video games. The activities of the game are represented by a specific type of Petri net called WorkFlow net. The definition of a topological map can be represented by state graphs. Using Colored Petri nets, it is possible to define formal communication mechanisms between the model of activity and the model of the map. The simulation of the timed models allows then to produce an estimated time that corresponds to the effective duration a player will need to complete a level of a game. Furthermore, a kind of Soundness property related to gameplay in a game Quest can be verified through state space analysis. For a better understanding of the approach, the video game Silent Hill II is used.