Hybrid LSTM Self‐Attention Mechanism Model for Forecasting the Reform of Scientific Research in Morocco

Abstract

Education is the cultivation of people to promote and guarantee the development of society. Education reforms can play a vital role in the development of a country. However, it is crucial to continually monitor the educational model’s performance by forecasting the outcome’s progress. Machine learning-based models are currently a hot topic in improving the forecasting research area. Forecasting models can help to analyse the impact of future outcomes by showing yearly trends. For this study, we developed a hybrid, forecasting time-series model by long short-term memory (LSTM) network and self-attention mechanism (SAM) to monitor Morocco’s educational reform. We analysed six universities’ performance and provided a prediction model to evaluate the best-performing university’s performance after implementing the latest reform, i.e., from 2015–2030. We forecasted the six universities’ research outcomes and tested our proposed methodology’s accuracy against other time-series models. Results show that our model performs better for predicting research outcomes. The percentage increase in university performance after nine years is discussed to help predict the best-performing university. Our proposed algorithm accuracy and performance are better than other algorithms like LSTM and RNN.

Predictive Analysis for Inventory Management of Coconut Warehouse (Case Study: Banio Lahewa)

“…LSTM networks have been used to forecast stock prices, workload in cloud data centre, and power demand. For example, (Fahim et al, 2021) developed a forecasting algorithm called SAM-LSTM, which is a fusion method of self-attention mechanism (SAM) and LSTM, to forecast the reform of scientific research in Morocco. (Qiu et al, 2020) added investor sentiment tendency in model analysis and introduced empirical modal decomposition (EMD) combined with LSTM to obtain more accurate stock forecasts.…”

Target tracking algorithm based on attention mechanism

“…Briefly, Transformer is an architecture for transforming one sequence to another with the help of attention-based encoder and decoder. The attention mechanism can be used to look at the input sequence and decide at each step how important the rest of the sequence is, thus helping to capture global information from the input sequence [7]. Transformer has replaced recurrent neural networks in many sequential tasks (natural language processing, speech processing and computer vision) and is gradually being extended to handle non-sequential problems.…”

A Hybrid Approach for Network Rumor Detection Based on Attention Mechanism and Bidirectional GRU Model in Big Data Environment

“…For the detection of Internet rumors, there has been some research at home and abroad, which can be divided into supervised and unsupervised based methods and deep learning-based detection methods [11,12]. For example, Arp et al in Ref.…”