DFuzzy: a deep learning-based fuzzy clustering model for large graphs

Graph clustering is successfully applied in various applications for finding similar patterns. Recently, deep learning- based autoencoder has been used efficiently for detecting disjoint clusters. However, in real-world graphs, vertices may belong to multiple clusters. Thus, it is obligatory to analyze the membership of vertices toward clusters. Furthermore, existing approaches are centralized and are inefficient in handling large graphs. In this paper, a deep learning-based model ‘DFuzzy’ is proposed for finding fuzzy clusters from large graphs in distributed environment. It performs clustering in three phases. In first phase, pre-training is performed by initializing the candidate cluster centers. Then, fine tuning is performed to learn the latent representations by mining the local information and capturing the structure using PageRank. Further, modularity is used to redefine clusters. In last phase, reconstruction error is minimized and final cluster centers are updated. Experiments are performed over real-life graph data, and the performance of DFuzzy is compared with four state-of-the-art clustering algorithms. Results show that DFuzzy scales up linearly to handle large graphs and produces better quality of clusters when compared to state-of-the-art clustering algorithms. It is also observed that deep structures can help in getting better graph representations and provide improved clustering performance.     

Pressure-Dependent Behavior of Defect-Modulated Band Structure in Boron Arsenide

The recent observation of unusually high thermal conductivity exceeding 1000 W m⁻¹ K⁻¹ in single-crystal boron arsenide (BAs) has led to interest in the potential application of this semiconductor for thermal management. Although both the electron/hole high mobilities have been calculated for BAs, there is a lack of experimental investigation of its electronic properties. Here, a photoluminescence (PL) measurement of single-crystal BAs at different temperatures and pressures is reported. The measurements reveal an indirect bandgap and two donor–acceptor pair (DAP) recombination transitions. Based on first-principles calculations and time-of-flight secondary-ion mass spectrometry results, the two DAP transitions are confirmed to originate from Si and C impurities occupying shallow energy levels in the bandgap. High-pressure PL spectra show that the donor level with respect to the conduction band minimum shrinks with increasing pressure, which affects the release of free carriers from defect states. These findings suggest the possibility of strain engineering of the transport properties of BAs for application in electronic devices.     

Big Data Framework for Zero-Day Malware Detection

Malware has already been recognized as one of the most dominant cyber threats on the Internet today. It is growing exponentially in terms of volume, variety, and velocity, and thus overwhelms the traditional approaches used for malware detection and classification. Moreover, with the advent of Internet of Things, there is a huge growth in the volume of digital devices and in such scenario, malicious binaries are bound to grow even faster making it a big data problem. To analyze and detect unknown malware on a large scale, security analysts need to make use of machine learning algorithms along with big data technologies. These technologies help them to deal with current threat landscape consisting of complex and large flux of malicious binaries. This paper proposes the design of a scalable architecture built on the top of Apache Spark which uses its scalable machine learning library (MLlib) for detecting zero-day malware. The proposed platform is tested and evaluated on a dataset comprising of 0.2 million files consisting of 0.05 million clean files and 0.15 million malicious binaries covering a large number of malware families over a period of 7 years starting from 2010.     

An Ontology driven model for detection and classification of cardiac arrhythmias using ECG data

Journal of Intelligent Information Systems - Cardiac arrhythmias are not life-threatening straight away but can cause serious heart-related complications if not medically handled appropriately. An...     

A scalable correlation‐based approach for outlier detection in wireless body sensor networks

Outlier detection is one of the prominent research domain in the field of data mining and big data analytics. Nowadays, most of the data in healthcare centers are remotely monitored and are generated from different wireless sensors. The core objective of outlier detection in this domain is the recognition of the true physiologically anomalous data and the anomalies because of faulty sensors. In real healthcare monitoring scenario, various sensors are related to each other. So, while detecting outliers in wireless body sensor networks (WBSNs), correlation among different sensor nodes is of major concern. Most of the existing outlier detection techniques consider the sensors to be linearly correlated, which may not always be the case in real life applications. The traditional techniques for outlier detection are also not scalable to big data. To address the above issues, in this paper, we propose an approach for outlier detection that is scalable to big data and also handles the nonlinearly correlated attributes efficiently. The proposed approach is implemented on Hadoop map reduce framework for the rapid processing of big data. The evaluation results are validated using the simulated dataset of WBSNs taken from the Physionet library. The results are compared with various existing outlier detection approaches and demonstrated that the proposed approach is more effective in spotting the physiological outliers and sensor anomalies accurately.     

Ensembled machine learning framework for drug sensitivity prediction

Drug sensitivity prediction is one of the critical tasks involved in drug designing and discovery. Recently several online databases and consortiums have contributed to providing open access to pharmacogenomic data. These databases have helped in developing computational approaches for drug sensitivity prediction. Cancer is a complex disease involving the heterogeneous behaviour of same tumour‐type patients towards the same kind of drug therapy. Several methods have been proposed in the literature to predict drug sensitivity. However, these methods are not efficient enough to predict drug sensitivity. The present study has proposed an ensemble learning framework for drug‐response prediction using a modified rotation forest. The proposed framework is further compared with three state‐of‐the‐art algorithms and two baseline methods using Genomics of Drug Sensitivity in Cancer (GDSC) and Cancer Cell Line Encyclopedia (CCLE) drug screens. The authors have also predicted missing drug response values in the data set using the proposed approach. The proposed approach outperforms other counterparts even though gene mutation data is not incorporated while designing the approach. An average mean square error of 3.14 and 0.404 is achieved using GDSC and CCLE drug screens, respectively. The obtained results show that the proposed framework has considerable potential to improve anti‐cancer drug response prediction.Inspec keywords: medical computing, molecular biophysics, genomics, genetics, learning (artificial intelligence), patient treatment, drugs, cellular biophysics, cancer, biology computing, tumours, diseasesOther keywords: ensembled machine learning framework, drug sensitivity prediction, drug therapy, ensemble learning framework, drug‐response prediction, Cancer Cell Line Encyclopedia drug screens, drug response values, CCLE drug screens, anti‐cancer drug response prediction     

Stacking-Order-Driven Optical Properties and Carrier Dynamics in ReS2

Two distinct stacking orders in ReS₂ are identified without ambiguity and their influence on vibrational, optical properties and carrier dynamics are investigated. With atomic resolution scanning transmission electron microscopy (STEM), two stacking orders are determined as AA stacking with negligible displacement across layers, and AB stacking with about a one-unit cell displacement along the a axis. First-principles calculations confirm that these two stacking orders correspond to two local energy minima. Raman spectra inform a consistent difference of modes I & III, about 13 cm⁻¹ for AA stacking, and 20 cm⁻¹ for AB stacking, making a simple tool for determining the stacking orders in ReS₂. Polarized photoluminescence (PL) reveals that AB stacking possesses blueshifted PL peak positions, and broader peak widths, compared with AA stacking, indicating stronger interlayer interaction. Transient transmission measured with femtosecond pump–probe spectroscopy suggests exciton dynamics being more anisotropic in AB stacking, where excited state absorption related to Exc. III mode disappears when probe polarization aligns perpendicular to b axis. The findings underscore the stacking-order driven optical properties and carrier dynamics of ReS₂, mediate many seemingly contradictory results in the literature, and open up an opportunity to engineer electronic devices with new functionalities by manipulating the stacking order.     

An intuitive general rank-based correlation coefficient

Correlation analysis is an effective mechanism for studying patterns in data and making predictions. Many interesting discoveries have been made by formulating correlations in seemingly unrelated data. We propose an algorithm to quantify the theory of correlations and to give an intuitive, more accurate correlation coefficient. We propose a predictive metric to calculate correlations between paired values, known as the general rank-based correlation coefficient. It fulfills the five basic criteria of a predictive metric: independence from sample size, value between ?1 and 1, measuring the degree of monotonicity, insensitivity to outliers, and intuitive demonstration. Furthermore, the metric has been validated by performing experiments using a real-time dataset and random number simulations. Mathematical derivations of the proposed equations have also been provided. We have compared it to Spearman’s rank correlation coefficient. The comparison results show that the proposed metric fares better than the existing metric on all the predictive metric criteria.     

AutoTrustRec: Recommender System with Social Trust and Deep Learning using AutoEncoder

Deep learning is the most active research topic amongst data scientists and analysts these days. It is because deep learning has provided very high accuracy in various domains such as speech recognition, image processing and natural language processing. Researchers are actively working to deploy deep learning on information retrieval. Due to large-scale data generated by social media and sensor networks, it is quite difficult to train unstructured and highly complex data. Recommender system is intelligent information filtering technique which assists the user to find topic of interest within complex overloaded information. In this paper, our motive is to improve recommendation accuracy for large-scale heterogeneous complex data by integrating deep learning architecture. In our proposed approach ratings, direct and indirect trust values are fed in neural network using shared layer in autoencoder. Comprehensive experiment analysis on three public datasets proves that RMSE and MAE are improved significantly by using our proposed approach.