No abstract

No abstract

No abstract

Image understanding and scene classification are keystone tasks in computer vision. The development of technologies and profusion of existing datasets open a wide room for improvement in the image classification and recognition research area. Notwithstanding the optimal performance of exiting machine learning models in image understanding and scene classification, there are still obstacles to overcome. All models are data-dependent that can only classify samples close to the training set. Moreover, these models require large data for training and learning. The first problem is solved by few-shot learning, which achieves optimal performance in object detection and classification but with a lack of eligible attention in the scene classification task. Motivated by these findings, in this paper, we introduce two models for few-shot learning in scene classification. In order to trace the behavior of those models, we also introduce two datasets (MiniSun; MiniPlaces) for image scene classification. Experimental results show that the proposed models outperform the benchmark approaches in respect of classification accuracy.

No abstract

Unsecured networks have recently become widely used for the transmission of confidential images. Consequently, cryptography is crucial for ensuring data confidentiality. Developing a key that is resistant to statistical and differential attacks has always been a challenging objective. In this paper, a novel model is proposed to boost image encryption while maintaining key strength. The proposed model adapts MD5 and SHA-256 hash functions to produce a key. It generates four matrices, X, Y, Z, and W, by using a memristor hyperchaotic system. Arnold's transform was applied to the original image once the key was created. The images were then scrambled using five chaotic maps. The image is then DNA-encoded, diffused using three matrices, and finally DNA-decoded. The proposed model was assessed using twelve performance measures on nine popular images. Compared to previous studies, the results of the proposed model indicate a promising improvement in performance. It achieves a better performance by expanding the key space and increasing its sensitivity.

Summary Cloud users are overwhelmed with great numbers of cloud services. Service recommender systems evaluate the services that provide same functionalities according to the user requirements. A key enabler to accurate recommendation in recommender systems is the appropriate determination of similar users. This paper contributes to the personalized cloud services recommendation area. In specific, we introduce a user‐based similarity measure that integrates relevant similarity aspects: user demographic information, service ratings, and user interest. The proposed similarity measure is used in a hybrid collaborative filtering (CF) approach that leverages the advantages of both model‐ and memory‐based approaches to improve the recommendation process. Experimental evaluation on real‐world services data set shows that the proposed approach outperforms other CF approaches in respect of the prediction accuracy and recommendation time while maintaining better or same coverage.

Summary In this paper, we propose the Software‐as‐a‐Service (SaaS) Recommender (SaaSRec), a personalized reputation‐based QoS‐aware recommender system (RS) for SaaS services. SaaSRec semantically processes user requests in order to find business‐oriented matching services, which are then filtered to satisfy the user QoS requirements and service characteristics. Subsequently, hybrid filtering is utilized to validate the services set on the basis of services metadata, reputation and user interests. Finally, the recommended set of services is ranked using a unique combination of factors: Relevance to user profile, service reputations and service cost. Moreover, we propose a new method for calculating the service reputation from the objective time‐weighted user feedbacks. SaaSRec addresses many challenges faced by the generic RSs: User cold‐start problem, limited content analysis and low performance. In respect of service RSs, SaaSRec tackles the disregarding of relevant factors to services recommendation: Cloud service characteristics, user physical location, service reputation and user interests. Moreover, SaaSRec provides a hybrid justification for the recommended services to increase the user's acceptance. Experimental evaluation against a real‐world services dataset has been carried out, and the results show that the proposed recommendation approach surpasses other collaborative filtering‐based recommendation approaches in respect of both precision and recall. This performance improvement was verified using different matrix density levels and number of recommendations. Copyright © 2016 John Wiley & Sons, Ltd.