Chương 1: Tỉnh hình nghiên cứu hiện nay, Chương 2: Cơ sở lÿ Huyết, Chương 3: Mô hình đê xuất, Chương 4: Dánh giá hiệu quả mô hình. Ngày tháng năm 2021 Giáo viên hưứng dẫn Tác giá luận văn CHỦ TỊCH HỘI ĐỒNG Content Confent. List of Vigures. List of Tables.
List of Equations. 1 12 sting solutions and problems - | 1.3 Goals and approaches 1.4 Structure of thesis Chapter 1. Related works Chapter2, Theoretical Background 2.2 Deep learnirys averview 3 2.3 Long short-term memory - - - co 2.4 Encoder-Decoder model - - 10 2.1 The importance of features.7 Reszarch methods - 28 Chapter 3. Proposed Forecasting Framework (OFFGED).1 Overview: - - 19 Acknowledgements First of all, I would like to deeply thank my family, especially my parents - who have worked hard to raise me.
My parents have always been with me and created the best conditions for me to have all the necessities needed for my studies Parents are the spiritual fulcrum, helping me to have a springboard to overcome difficulties and challenges. I would like to express my gratitude to my advisors, Dr. Nguyen Phi Le for supporting my studies and rescarch on this subject. She is very kindhearted and supportive person, who has guided me from the first day T worked with her Moreover, I would like to thank Dr.
Nguyen Thanh Hung, who has spent his precious time supporting, giving me advice and along with Dr. Nguyen Phi Le, giving me opportunities to work in many amazing projects. My sincere thanks also go to all the people in the ICN laboratory of the BK. T have a wonderful time working with Lalented and special peers.
T learned a Jot from them and they always spread positive energy [or me Finally. I would like to thank my friends wha have always stood by me, shared joys and sorrows, and always supported and helped me all the time. Abstract ‘The concentration of fine particulate matter (PM2.5), which represents inhalable particles with diameters of 2.5 micrometers and smaller, is a vital air quality index. Such particles can penetrate deep into the human lungs and severely affect human health.
This paper studies accurate PM2.5 prediction, which can potentially contribute to reducing or avoiding the negative consequences. Our approach’s novelty is to utilize the genetic algorithm (GA) and an encoder-decoder (E-D} model for PM2. ‘Ihe GA benefits feature selection and remove outliers to enhance the prediction accuracy. ‘The encoder-decoder model with long short-term memory (LS‘I'M), which relaxes the restrictions between the input and output of the model, can be used to effectively predict the PM2.
We evaluate the proposed model on air quality datasets from Hanoi and ‘aiwan. ‘The evaluation results show that our model achieves excellent performance. By merely using the H-L) model, we can obtain more accurate (up to 53.7%) predictions than those of previous works. Moreover, the GA im our model has the advantage of obtaining the optimal feature combination for predicting the PM2.
By combining the GA-based feature selection algorithm and the E-D model, our proposed approach further umproves the accuracy by at least 13. List of Figures Figure 1. An example of an artificial neural network, - -. Structure of RNN - - 1 Figure 3.
Structure of the LSTM unit 9 Figure 4. The basic strueturz of thz encoder-deooder modil. An example of feature extraction. An cxamplc of feature sclcetien.
An example of feature construction. The basic structure of Genetic Algorithm. Overview of the proposed model 19 Figure 10. Encoding a feature combination (the white and gray cells represent the selected feature encoded by 1 and 0, respectively).
Mustration of the GA’s crossovor and mutation operations. Structure of the LSTM-based encoder-decoder model. Notation of the proposed GA-based training mechanism. training strategy — fixed, shuffling = (true, false) u Figure 15.
Training strategy— fixed, shuffling = (false, false). Training strategy — fixed, shuffling = (true, true). Training stralogy — fixed, shuffling = (false, true). 26 Figure 18, Impact of the number of generations.
Comparison of feature selection algorithms - 3 Figure 20. Comparison of GA-based feature selection and using all the features for the Hanoi dafasct. Comparison between models using Hanoi dataset with all features. Comparison between madets using Hanoi dalaset with fealure selected by GA.
MAE of the proposed model with different output lengths. Comparison betw: sr models using Taiwan dalasel with featuras sclcetod by soe AD.2 GA-based feature selectton.3 Encader-Necoder model-based pred 3.4 New training stategy LTS2 Chapter 4.1 Dataset and evaluation settings 27 4.2 Impact of the GA’s number of generations.3 Comparing feature selection algorithms - - - 30 4.4 Comparing prediction modcts - - 38 4.1 Comparing ED-LSTM, AR-BiLSTM, and AC-LSTM 34 4.2 Comparing 1D-LSTM and SI-DNN 3Ð 4.1 Results of OFEGED.2 Results of LTS2 - ¬. References Content Confent. List of Vigures.
List of Tables. List of Equations. 1 12 sting solutions and problems - | 1.3 Goals and approaches 1.4 Structure of thesis Chapter 1. Related works Chapter2, Theoretical Background 2.2 Deep learnirys averview 3 2.3 Long short-term memory - - - co 2.4 Encoder-Decoder model - - 10 2.1 The importance of features.7 Reszarch methods - 28 Chapter 3.
Proposed Forecasting Framework (OFFGED).1 Overview: - - 19 Abstract ‘The concentration of fine particulate matter (PM2.5), which represents inhalable particles with diameters of 2.5 micrometers and smaller, is a vital air quality index. Such particles can penetrate deep into the human lungs and severely affect human health. This paper studies accurate PM2.5 prediction, which can potentially contribute to reducing or avoiding the negative consequences. Our approach’s novelty is to utilize the genetic algorithm (GA) and an encoder-decoder (E-D} model for PM2.
‘Ihe GA benefits feature selection and remove outliers to enhance the prediction accuracy. ‘The encoder-decoder model with long short-term memory (LS‘I'M), which relaxes the restrictions between the input and output of the model, can be used to effectively predict the PM2. We evaluate the proposed model on air quality datasets from Hanoi and ‘aiwan. ‘The evaluation results show that our model achieves excellent performance.
By merely using the H-L) model, we can obtain more accurate (up to 53.7%) predictions than those of previous works. Moreover, the GA im our model has the advantage of obtaining the optimal feature combination for predicting the PM2. By combining the GA-based feature selection algorithm and the E-D model, our proposed approach further umproves the accuracy by at least 13. List of Tables ‘Lable 1.
Details of missing data in the datasets. Hyperparametor tings - - 29 Table 3.STM, AF-Ril.STM, and AC-ILSTM usc all fcalures (Hanoi dalasel).34 Table 4, ED-LSTM, AE-BiLSTM and AC-LSTM use selected features by GA (Hanoi dataset) 35 Table 5, Comparing the MAE of the proposed ED-LSTM model and the ST-DNN model (using the features proposed by [11 )). Hyperparameters of training strategy. - cna AD Table 7.
Training strategy for different cases. Comparing proposed method combining new training strategy with related works. - - ace Table 9, Corclation of features. - en AS iw List of Figures Figure 1.
An example of an artificial neural network, - -. Structure of RNN - - 1 Figure 3. Structure of the LSTM unit 9 Figure 4. The basic strueturz of thz encoder-deooder modil.
An example of feature extraction. An cxamplc of feature sclcetien. An example of feature construction. The basic structure of Genetic Algorithm.
Overview of the proposed model 19 Figure 10. Encoding a feature combination (the white and gray cells represent the selected feature encoded by 1 and 0, respectively). Mustration of the GA’s crossovor and mutation operations. Structure of the LSTM-based encoder-decoder model.
Notation of the proposed GA-based training mechanism. training strategy — fixed, shuffling = (true, false) u Figure 15. Training strategy— fixed, shuffling = (false, false). Training strategy — fixed, shuffling = (true, true).
Training stralogy — fixed, shuffling = (false, true). 26 Figure 18, Impact of the number of generations. Comparison of feature selection algorithms - 3 Figure 20. Comparison of GA-based feature selection and using all the features for the Hanoi dafasct.
Comparison between models using Hanoi dataset with all features. Comparison between madets using Hanoi dalaset with fealure selected by GA. MAE of the proposed model with different output lengths. Comparison betw: sr models using Taiwan dalasel with featuras sclcetod by soe AD.
iii Acknowledgements First of all, I would like to deeply thank my family, especially my parents - who have worked hard to raise me. My parents have always been with me and created the best conditions for me to have all the necessities needed for my studies Parents are the spiritual fulcrum, helping me to have a springboard to overcome difficulties and challenges. I would like to express my gratitude to my advisors, Dr. Nguyen Phi Le for supporting my studies and rescarch on this subject.
She is very kindhearted and supportive person, who has guided me from the first day T worked with her Moreover, I would like to thank Dr. Nguyen Thanh Hung, who has spent his precious time supporting, giving me advice and along with Dr. Nguyen Phi Le, giving me opportunities to work in many amazing projects. My sincere thanks also go to all the people in the ICN laboratory of the BK.
T have a wonderful time working with Lalented and special peers. T learned a Jot from them and they always spread positive energy [or me Finally. I would like to thank my friends wha have always stood by me, shared joys and sorrows, and always supported and helped me all the time. List of Tables ‘Lable 1.
Details of missing data in the datasets. Hyperparametor tings - - 29 Table 3.STM, AF-Ril.STM, and AC-ILSTM usc all fcalures (Hanoi dalasel).34 Table 4, ED-LSTM, AE-BiLSTM and AC-LSTM use selected features by GA (Hanoi dataset) 35 Table 5, Comparing the MAE of the proposed ED-LSTM model and the ST-DNN model (using the features proposed by [11 )). Hyperparameters of training strategy. - cna AD Table 7.
Training strategy for different cases. Comparing proposed method combining new training strategy with related works. - - ace Table 9, Corclation of features. - en AS iw List of Tables ‘Lable 1.
Details of missing data in the datasets. Hyperparametor tings - - 29 Table 3.STM, AF-Ril.STM, and AC-ILSTM usc all fcalures (Hanoi dalasel).34 Table 4, ED-LSTM, AE-BiLSTM and AC-LSTM use selected features by GA (Hanoi dataset) 35 Table 5, Comparing the MAE of the proposed ED-LSTM model and the ST-DNN model (using the features proposed by [11 )). Hyperparameters of training strategy. - cna AD Table 7.
Training strategy for different cases. Comparing proposed method combining new training strategy with related works. - - ace Table 9, Corclation of features. - en AS iw Abstract ‘The concentration of fine particulate matter (PM2.5), which represents inhalable particles with diameters of 2.5 micrometers and smaller, is a vital air quality index.
Such particles can penetrate deep into the human lungs and severely affect human health. This paper studies accurate PM2.5 prediction, which can potentially contribute to reducing or avoiding the negative consequences. Our approach’s novelty is to utilize the genetic algorithm (GA) and an encoder-decoder (E-D} model for PM2. ‘Ihe GA benefits feature selection and remove outliers to enhance the prediction accuracy.
‘The encoder-decoder model with long short-term memory (LS‘I'M), which relaxes the restrictions between the input and output of the model, can be used to effectively predict the PM2. We evaluate the proposed model on air quality datasets from Hanoi and ‘aiwan. ‘The evaluation results show that our model achieves excellent performance. By merely using the H-L) model, we can obtain more accurate (up to 53.7%) predictions than those of previous works.
Moreover, the GA im our model has the advantage of obtaining the optimal feature combination for predicting the PM2. By combining the GA-based feature selection algorithm and the E-D model, our proposed approach further umproves the accuracy by at least 13. List of Figures Figure 1.