A New Method for Sentiment Analysis Using Contextual Auto-Encoders

Sentiment analysis, a hot research topic, presents new challenges for understanding users’ opinions and judgments expressed online. They aim to classify the subjective texts by assigning them a polarity label. In this paper, we introduce a novel machine learning framework using auto-encoders network to predict the sentiment polarity label at the word level and the sentence level. Inspired by the dimensionality reduction and the feature extraction capabilities of the auto-encoders, we propose a new model for distributed word vector representation “PMI-SA” using as input pointwise-mutual-information “PMI” word vectors. The resulted continuous word vectors are combined to represent a sentence. An unsupervised sentence embedding method, called Contextual Recursive Auto-Encoders “CoRAE”, is also developed for learning sentence representation. Indeed, CoRAE follows the basic idea of the recursive auto-encoders to deeply compose the vectors of words constituting the sentence, but without relying on any syntactic parse tree. The CoRAE model consists in combining recursively each word with its context words (neighbors’ words: previous and next) by considering the word order. A support vector machine classifier with fine-tuning technique is also used to show that our deep compositional representation model CoRAE improves significantly the accuracy of sentiment analysis task. Experimental results demonstrate that CoRAE remarkably outperforms several competitive baseline methods on two databases, namely, Sanders twitter corpus and Facebook comments corpus. The CoRAE model achieves an efficiency of 83.28% with the Facebook dataset and 97.57% with the Sanders dataset.     

On Bayesian EWMA control charts under different loss functions

The motivation for this study is to analyze Bayesian exponentially weighted moving average (EWMA) control chart under 3 loss functions namely, SELF, LLF, and PLF. Informative priors (normal and mixture of normal) and non‐informative priors (Uniform and Jefferys) are considered for the analysis. The performance of Bayesian EWMA control chart using posterior and posterior predictive distribution scheme has been evaluated using average run length (ARL) and standard deviation run length (SDRL) as performance measures. Monte Carlo simulations are used to compute the performance measures for different values of smoothing constant. An illustrative example is also presented for practical considerations of Bayesian EWMA control chart.     

The electrical characteristics of metal–semiconductor hetero-structures with graphene oxide and perylenetetracarboxylic dianhydride interface

Journal of Materials Science: Materials in Electronics - In this study, we investigated the electrical properties of Al/p-Si, Al/GO/p-Si, Al/PTCDA/p-Si and Al/GO-PTCDA/p-Si structures were examined...     

Spin Valve Effect of 2D‐Materials Based Magnetic Junctions

The magnetotransport properties of spin valve structure are highly influenced by the type of intervening layer inserted between the ferromagnetic electrodes. In this scenario, spin filtering effect at the interfaces plays a crucial role in determining the magnetoresistance (MR) of such magnetic structures, which can be enhanced by using a suitable intervening layer. Here, the authors investigate the spin filtering effect of the two‐dimensional layers such as hexagonal boron nitride (hBN), graphene (Gr), and Gr‐hBN hybrid system for modifying the magnetotransport characteristics of the vertical spin valve architectures (Ni/hBN/Ni, Ni/Gr/Ni, and Ni/Gr‐hBN/Ni). Compared to graphene, hBN incorporated magnetic junction reveals higher MR and spin polarizations (P) suggesting better spin filtering at the interfaces. The MR for hBN incorporated junction is calculated to be ≈0.83%, while that of graphene junction it is estimated to be ≈0.16%. Similar contrast is observed in the ‘P’ of ferromagnets (FMs) for the two junctions, that is, ≈6.4% for hBN based magnetic junction and ≈2.8% for graphene device. However, for Gr‐hBN device, the signal not only get inverts, but it also suggests efficient spin filtering mechanism at the FM interfaces. Their results can be useful to comprehend the origin of spin filtering and the choice of non‐magnetic spacer for magnetotransport characteristics.

     

Cooperative advertising in a supply chain with retail competition

This research assesses the effects of cooperative (coop) advertising in a channel with competing retailers considering both advertising and pricing as decision variables. We develop a game-theoretic model and provide equilibrium solutions for two games. In Game 1, the manufacturer and the retailers do not use cooperative advertising (status quo); and in Game 2, coop advertising is implemented. We also obtain optimal solutions for the case where the channel is coordinated. Contrary to the results provided for one-manufacturer, one-retailer channels, we find that coop advertising may not be profitable for the retailers or for the channel, especially when the market is characterised by low levels of price competition and high advertising competition between retailers. Although it benefits the manufacturer, the total effect of cooperative advertising on the channel profit might be negative under such conditions. The results also show that coop advertising stimulates retailers’ spending but may result in lower advertising expenditures than for a fully coordinated channel. Finally, when coop advertising benefits the entire channel, it does not fully achieve results from vertical integration.     

Correction to a novel framework for rapid diagnosis of COVID-19 on computed tomography scans

Pattern Analysis and Applications - A correction to this paper has been published: https://doi.org/10.1007/s10044-021-00969-x     

A novel solution of deep learning for sleep apnea detection: enhancement of SC and elimination of GVICS

Multimedia Tools and Applications - Convolutional neural network (CNN) classification has not achieved a medically-satisfied level of accuracy in sleep apnea detection due to the negative effect of...     

Automatic detection of Moroccan coastal upwelling zones using sea surface temperature images

An efficient unsupervised method is developed for automatic segmentation of the area covered by upwelling waters in the coastal ocean of Morocco using the Sea Surface Temperature (SST) satellite images. The proposed approach first uses the two popular unsupervised clustering techniques, k-means and fuzzy c-means (FCM), to provide different possible classifications to each SST image. Then several cluster validity indices are combined in order to determine the optimal number of clusters, followed by a cluster fusion scheme, which merges consecutive clusters to produce a first segmentation of upwelling area. The region-growing algorithm is then used to filter noisy residuals and to extract the final upwelling region. The performance of our algorithm is compared to a popular algorithm used to detect upwelling regions and is validated by an oceanographer over a database of 92 SST images covering each week of the years 2006 and 2007. The results show that our proposed method outperforms the latter algorithm, in terms of segmentation accuracy and computational efficiency.