Rdouan Faizi

A new recommender system for e-learning environments

Today recommender systems are widely used not only in e-commerce but in e-learning as well. They are actually made use of in the latter environment to suggest resources and learning materials to learners and, thus, contribute in improving the quality of both teaching and learning. In this paper, we put forward a new recommendation system that provides learners with the most appropriate answers and clues through a request /answer module.

Hybrid approach-based support vector machine for electric load forecasting incorporating feature selection

Forecasting future electricity demand is very important for the electric power industry. In fact, it has been shown in several research works that machine learning methods are useful for electric load forecasting (ELF) since electric load data are nonlinear in relation and complex. On the other hand, it is important to determine the irrelevant factors as a preprocessing step for ELF. Our objective in this paper is to investigate the importance of applying the feature selection approach to remove the irrelevant factors of electric load. To this end, we introduce a hybrid machine learning approach that combines support vector machine (SVM) and particle swarm optimisation (PSO) in both continuous and binary forms. Specifically, the binary hybridisation is used for feature selection and the continuous one is used for ELF. Experimental results demonstrate the feasibility of applying feature selection by SVM and PSO algorithms without decreasing the performance of the forecasting model for ELF.

A novel based Hidden Markov Model approach for controlling the ACS-TSP evaporation parameter

The aim of this paper is to propose a new method capable of dynamically controlling the evaporation parameter in an Ant Colony System (ACS) using a Hidden Markov Model. The purpose is to improve the performance of ACS by controlling the exploration and exploitation in the search space. To this end, two HMM approaches are proposed. The first is a training method that best suits the observed data of the Hidden Markov Model. The second is a method that dynamically controls the adapted parameter by applying several processes. To test our algorithm we used a set of Travelling Salesman Problem (TSP) instances.

Facing the Feature Selection Problem with a Binary PSO-GSA Approach

Feature selection has become the focus of much research in many areas where we can face the problem of big data or complex relationship among features. Metaheuristics have gained much attention in solving many practical problems, including feature selection. Our contribution in this paper is to propose a binary hybrid metaheuristic to minimize a fitness function representing a trade-off between the classification error of selecting the feature subset and the corresponding number of features. This algorithm combines particle swarm optimization (PSO) and gravitational search algorithm (GSA). Also, a mutation operator is integrated to enhance population diversity. Experimental results on ten benchmark dataset show that our proposed hybrid method for feature selection can achieve high performance when comparing with other metaheuristic algorithms and well-known feature selection approaches.

Prediction of Supplier Performance: A Novel Dea-Anfis Based Approach

The focus of this paper is on investigating the feasibility of using ANFIS combined with DEA for suppliers post-evaluation. The proposed framework aims at modeling performance measurement, and forecasting of a selected hospitals drug suppliers. Even though it is broadly employed as a benchmarking tool to evaluate DMUs efficiency, DEA can hardly be used to predict the performance of unseen DMUs. For this reason, ANFIS model has been integrated to DEA due to its nonlinear mapping, strong generalization capabilities and pattern prediction functionalities. DEA based BCC model is used to evaluate the efficiency scores of a set of suppliers, then ANFIS intervenes to learn DEA patterns and to forecast the performance of new suppliers. The results of this research highlight the prediction power of the proposed model in a new scope. They present it as an efficient benchmarking tool and a promising decision support system applied at the operational level.

Artificial Neural Network Based Approach for Blood Demand Forecasting: Fez Transfusion Blood Center Case Study

Blood demand and supply management are considered one of the major components of a healthcare supply chain, since blood is a vital element in preserving patients life. However, forecasting it faces several challenges including frequent shortages, and possible expiration caused by demand uncertainty of hospitals. This uncertainty is mainly due to high variability in the number of emergency cases. Thereupon, this investigation presents a real case study of forecasting monthly demand of three blood components, using Artificial Neural Networks (ANNs). The demand of the three blood components (red blood cells (RBC), plasma (CP) and platelets (PFC)) and other observations are obtained from a central transfusion blood center and a University Hospital. Experiments are carried out using three networks to forecast each blood component separately. Last, the presented model is compared with ARIMA to evaluate its performance in prediction. The results of this study depict that ANN models overcomes ARIMA models in demand forecasting. Thus high ANN models can be considered as a promising approach in forecasting monthly blood demand.

An automatic ethical-based recommender system for e-commerce

Nowadays, recommender systems are considered as one of the basic pillars of e-commerce as they help users to take decisions easily. These systems involve a multitude of techniques ranging from hybrid filtering mechanisms to techniques derived from statistics or artificial intelligence. In the present paper, we put forward an improved recommender system that supports ethics in an automatic way without the user intervention.

Topic and sentiment model applied to the colloquial Arabic: a case study of Maghrebi Arabic

Recently, the multiplication of communication and sharing platforms such as social networks, personal blogs, forums, etc., has facilitated the expression of views and opinions about products, personalities, and public policy. However, gathering these points of view is a complex task that requires resolution of many problems in different disciplines, especially issues related to our language. Among the research areas, topic modeling and sentiment analysis stimulates interest and curiosity of the scientific community. Lately, the current economic, geo-political and geostrategic trends have made researchers specifically more interested in Arabic language, except that the majority of these studies focus on the classical Arabic; nevertheless it is a language of the elites which is different from what is mainly used on the Web. Our paper focuses on Maghrebi colloquial Arabic since the little research that exists in this area is limited to East colloquial Arabic. On a corpus extracted from different Facebook pages we implemented a supervised approach to extract the sentiments, and an unsupervised approach to extract topic, then we proposed a new, semi-supervised, approach in the Arabic language that combines the topic and the sentiment in a single model, in order to join each topic to a specific sentiment.

Hidden Markov Model Classifier for the Adaptive ACS-TSP Pheromone Parameters

The Hidden Markov Models (HMM) are a powerful statistical techniques for modeling complex sequences of data. In this paper a Hidden Markov Model classifier is a special kind of these models that aims to find the posterior probability of each state given a sequence of observations and predicts the state with the highest probability. The purpose of this work is to enhance the performance of Ant Colony System algorithm applied to the Travelling Salesman Problem (ACS-TSP) by varying dynamically both local and global pheromone decay parameters based on the Hidden Markov Model algorithm, using two indicators: Diversity and Iteration that reflect the state of research space in a given moment. The proposed method was tested on several TSP benchmark instances, which compared with the basic ACS, the combination of Fuzzy Logic Controller (FLC) and ACS to prove the efficiency of its performance.

An Enhanced MSER Based Method for Detecting Text in License Plates.

In this paper, we propose a novel method for detecting license plates (LP) in images. The proposed algorithm is an extension of Maximally Stable Extremal Regions (MSER) for extracting candidate text region of LP. The approach is more robust to edge and more powerful thanks to its stability, and robustness against the changes of scale and illumination. We propose a novel method based on a bilateral filter as well as an adaptive dynamic threshold so as to improve the MSER results. Besides, we consider the outer tangent of circles intersection for filtering the region with the same orientation, and finally a character classifier based on geometrical and statistical constraints of character to eliminate false detection. Thus, our proposal consists of three steps namely, image preprocessing, candidate license plate character detection, and finally filtering and grouping to eliminate false detection.