Abstract This work presents a parallel implementation of JADE: Adaptive Differential Evolution With Optional External Archive, using the Compute Unified Device Architecture (CUDA), in order to reduce the execution run-time of the algorithm. The algorithm was tested using the well-known function Sphere and the execution run time was compared against its sequential version. The results were measured in terms of... This work presents a parallel implementation of JADE: Adaptive Differential Evolution With Optional External Archive, using the Compute Unified Device Architecture (CUDA), in order to reduce the execution run-time of the algorithm. The algorithm was tested using the well-known function Sphere and the execution run time was compared against its sequential version. The results were measured in terms of “Speed-up” and they show that the execution run-time can be reduced significantly by the use of CUDA, this benefit can be observed better when working with large amounts of data. However, not necessarily the population with more data reaches the best performance. Read More Read Less

Abstract Every day, thousands of people travel by train, not escaping the leniency of unforeseen events, the fluidity of the rail network being able to be disrupted by equipment breakdowns. Therefore, predictive maintenance is relevant and necessary to help anticipate these breakdowns and thus act against any mechanical, electrical, or technical constraints or obstacles that could disrupt or prevent the normal... Every day, thousands of people travel by train, not escaping the leniency of unforeseen events, the fluidity of the rail network being able to be disrupted by equipment breakdowns. Therefore, predictive maintenance is relevant and necessary to help anticipate these breakdowns and thus act against any mechanical, electrical, or technical constraints or obstacles that could disrupt or prevent the normal circulation of trains. This process is carried out using artificial intelligence approaches and various machine learning and deep learning models. This article will implement an approach that combines two essential concepts: time series and neural networks. We will start with the univariate analysis of the number of failures per day using a range of machine learning and deep learning algorithms, namely LSTM, BiLSTM, GRU, and SVR. The results show that we manage to minimize the prediction error; for example, with the GRU model, we get an RMSE of 0.487, but with increasing data, we get an RMSE of 0.463. Moreover, the problem encountered is the detection of peaks; the models cannot detect outliers, hence the use of the SVR model, which gives better coordination between the test data and the predicted data, with a gamma value of 0.03. Then, we tested the VAR model and the LSTM with several outputs; the latter gives satisfactory results with an accuracy rate of 92% and an RMSE of 0.006. Finally, we address the problem of classification of the nature of failures. We used several machine learning algorithms, such as SVM, KNN, Random Forest, then tested a method of “Ensemble Learning,” the Vote. In the latter, we combined the three algorithms used previously, which increased the accuracy of the test to 61.73%.  Read More Read Less

Abstract In this paper, a new algorithm called Reliability Ratio Weighted Bit Flipping–Sum Product (RRWBFSP) decoding is proposed for low-density parity-check codes. The new algorithm combines two different algorithms, Sum Product [4] and Reliability Ratio Weighted Bit Flipping [6]. The results of the simulation show that the new algorithm achieves a 0.34 dB performance gain over of the standard Sum- Product decoding... In this paper, a new algorithm called Reliability Ratio Weighted Bit Flipping–Sum Product (RRWBFSP) decoding is proposed for low-density parity-check codes. The new algorithm combines two different algorithms, Sum Product [4] and Reliability Ratio Weighted Bit Flipping [6]. The results of the simulation show that the new algorithm achieves a 0.34 dB performance gain over of the standard Sum- Product decoding algorithms. Furthermore, RRWBFSP algorithm has almost the same computational complexity compared to the standard Sum-Product algorithm. Read More Read Less

Abstract In this paper, we present a comparative study of the object detection accuracy and speed of various state-of-the art models for the road scene context. Commensurate with the model training method, the algorithms can be divided into two types: one-stage models and two-stage models. We focused on the road context in order to detect all occurrences of objects in... In this paper, we present a comparative study of the object detection accuracy and speed of various state-of-the art models for the road scene context. Commensurate with the model training method, the algorithms can be divided into two types: one-stage models and two-stage models. We focused on the road context in order to detect all occurrences of objects in the road such as, car, person, traffic signs, etc. Accordingly, we find that one-stage detectors are stronger in terms of prediction speed, while two-stage models are stronger in terms of accuracy. To train deep neural networks with platform GPU type on a large amount of image data that required time, Because the computational cost of computer vision is very high, so we are focused to transfer learning technique, where a model trained on one task is reused on another related task, gives better results. Read More Read Less

Abstract The semiconductor process cannot avoid defects due to its complex and diverse processes. In particular, wafer can be said to be the core of semiconductor manufacturing because they are directly related to the productivity of semiconductors. Therefore, detecting and classifying defects on wafers can help engineers address the root cause of defects and improve yield. In this paper, we propose... The semiconductor process cannot avoid defects due to its complex and diverse processes. In particular, wafer can be said to be the core of semiconductor manufacturing because they are directly related to the productivity of semiconductors. Therefore, detecting and classifying defects on wafers can help engineers address the root cause of defects and improve yield. In this paper, we propose a convolutional autoencoder using skip connection for wafer map defect classification. First, the encoder and decoder are designed by constructing a convolutional block. And connect the symmetrical blocks with skip connection. Finally, the training data of the classifier is encoded using the weights of the learned encoder. The loss of the model was successfully reduced with skip connection, and improved performance was obtained by reusing the encoder. Read More Read Less