Air Quality Prediction Using Classification Techniques
1 Sumathi K, 2aishwarya K, 3gayathri M, 4gayathri R, 5menaha A
1,2,3,4,5
Department of Computer Science and Engineering, MKumarasamy College of Engineering, Thalavapalayam, Karur, Tamil Nadu, India -639113.
1[email protected], 2[email protected],3[email protected],
4[email protected],5[email protected]
ABSTRACT
In several industry applications and metro areas nowadays, air internal control and restoration is one of the most significant practices. The quality of the air is harmed by various types of pollution caused by transportation, power, and petroleum usage, among other things. The evidence of noxious particles is posing significant consequence to the improving the overall quality of life in green infrastructure. It is important to integrate reliable environmental monitoring models that collect data on pollutant levels and provide an analysis of environmental pollution in each area compared to air pollution. Air pollution harms both the atmosphere and human health, resulting in global warming, acid rain, heart disease, lung cancer, and skin cancer among the population. More analysis is needed to assess raw information in view of identify pollutants. ML algorithms and deep learning ensure that potential AQI can be predicted and effective steps can be taken. As a result, air quality estimation and prediction has become a major research subject. We will look at various ML [26] and DL algorithms from which can identify air quality in this analysis.
Keywords
Air quality index, Machine Learning, Deep Learning techniques, Air pollutants, Quality prediction.
INTRODUCTION
The statistics indicate that people are increasingly heading to towns. The United Nations (UN) reports, the Metropolitan People in 2018 would be around 55.3%, and it is expected to reach 68 percent by 2050. Urbanization brings with it a slew of issues relating to transportation, health care, air quality, and other facets of life. To address these issues, the smart city concept was created, which is a technology that combines information and communication (ICT) that residents as well as evolving assets to promote long-term growth and improved quality of life. There are many definitions to explain smart city, prefer 'A smart community is a metropolis where six key aspects include the intelligent economy, intelligent transportation, advanced environment, advanced citizenry, advanced life and procedural knowledge‟ or 'The use of sophisticated communication technology in order to make the smarter and more efficient city infrastructure and services such as city government, education, healthcare, real estate, public safety, transport and utilities.' Using intelligent city services, we can access a large number of stats to learn more about
the present situation to get a better view of the situation throughout the town. A key feature of urban mobility[25] is the accessibility of data provided by different sensors.
Air is amongst the most valuable ecological supporting the growth and survival of lifeforms.
Air is essential for the survival of all types of life, including plants and animals. Consequently, every living organism needs high-quality, toxic-free air for survival. Because of the trend toward industrialization and a temporal long-term exposure interaction with fine mean a thing of particulates and acute rises through mortality, such as lung cancer and cardiopulmonary disease, air pollution has been the topic of many recent environmental studies. Throughout the years, traditional methods have been used to assess air quality. These approaches include manual data collection and analysis. Many researchers have suggested different solutions for predicting air pollution using mathematical or deep learning models, ranging from environmental to data science. At different times of the day, the parameters measured by meteorological stations (temperature, wind, humidity, and insolation) cloud form various set of potential features.
The aim of this paper is to review papers on estimation pollution of the air in digital machine learning will help cities algorithm, compare methodologies used by different authors, and get a general idea of applied approaches.
Deep learning techniques have begun to be actively established in this location or environment, and more surveys and assumptions were carried out conducted, that is dependent on the significance in the detention center. Combining all of it the data will aid in determining the trend and identifying the technologies used in the analysis survey setting, which then will guide us into the research consideration.
Fig 1: Air quality prediction RELATED WORK
RUCHI RAUTRI, et.al,.[1] implemented the most interesting and toughest task is predicting air pollution making use of data mining. The majority of the systems in those days are intended to aid in the storage of air pollution data, analysis of that particular data and providing result in the form of statistics. Most of the systems make use of decision support systems, but their capabilities are severely limited. They will respond to basic questions such as "What is the highest range of the air pollution" and "Where are the most pollutants?" However, they are unable to answer complex queries such as “Predict subsequent month air pollutants count” or
“Given me, tomorrow‟s air pollution data”. This method employs prediction techniques.
BRAHIM K.K., et al., [2] developed a new framework of DL for analyze data from IoT[38,41] digital cities. We are proposing paper back method based on LSTM networks for
predicting hereafter air quality estimates in a digital city. The proposed model's result of the estimation is positive, indicating that it can be applied to other digital city complications in prediction as well.
PING-WEI SOH, et.al.,[3] goals to forecast air pleasant for as much as forty- eight the usage of a mixture of a couple of neural networks[28], which includes SNN, CNN, and LST period reminiscence in find temporal of spatial connections. Terrain extraction effect in air quality, proposed copy for prediction takes into account various meteorological data getting from the preceding couple of hours as well elevation space data. Observation from various locations areincluded in the model, which were derived from relationship between nearest areas and among comparable areas in domain of temporal. Observations with Taiwan and Beijing data units display suggested version higher overall outperforms contemporary trendy methods.
MENGLIRUAN, et.al,.[4] implemented an environment air reliability assessment and prediction copy focused in data mining. To assess environment air reliability, PCA is used to determine the principal component ratings. To design the environment air reliability model, data mining exponential smoothing method is used to process the input data. The results show that the designed method is comparable the real measurement range in surrounding quality of air and method has a better accuracy and better efficiency, which can be used as a benchmark for improving environmental quality.
XIUWENYI, et.al., [5] developed a DNN(Deep Neural Network) based method (entitled Deep air), it contains a spatial transformation aspect as well as a deep distributed network. Based on air pollutants spatial correlations, the previous part converts spatially sparse air quality prediction data into a coherent input that can re-create pollutant sources based on air pollutants spatial correlations. We deployed Deep Air in Air Pollution Prediction system, providing prediction of quality of air for more than three hundred cities of china in every hour. The result of the experiment is based on the 3-year and from 9-cities of china demonstrate benefit of Deep Air method which consists of ten base line. The percentage of short term[20], long term, sudden changes predictions are2.4, 12.2, 63.2 respectively.
GAGANJOT KAUR KANG, et.al.,[6] investigated air quality forecasting making use of different big-data and machine learning (ML) based techniques. This paper summaries published research findings on prediction of air quality using AI, decision trees, ML, deep learning, and other technologies. Furthermore, it gives the solution to future research paper. It is suggested that it takes into account both spatial and temporal correlations. To extract inherent air quality features using a SAE (Stacked Auto-Encoder) method is used, which is trained greedily layer- wise.
YING ZHANG, et.al.,[7] resolved the issues of analyzing high-dimensional large-scale data. It is solved using LightGBM model. We could boost prediction accuracy by exploring the data forecasting[23,29,35,40] function and available data. we are compared more predicted[24]
data after testing data in the dataset by comparing original data. The original data getting from Beijing which contains the 35 air quality monitoring stations. The sliding mechanism used to multi-dimensional secular property to increase the training data features in the dataset due to a lack of data.
DANIEL SCHURHOLZ, et.al.,[8] proposed a prediction model which uses context-aware data processing principles for combining an effective algorithm (Long Short-Term Memory and Deep Neural Network) which is used to predict the pollution of air and it gets data from pollution sources and the user's health profile. Outcomes acquired with My Air Quality Index -MyAQI show high accuracy levels are obtained (90% - 96%) and the usage of this model is very high to the user.
DITSUHIISKANDARYAN, et al.,[9] goal is to review articles on prediction of pollution level in air digital cities which is implemented by using ML algorithms, to compare methodologies used by different authors, a general idea about applied methods. Machine learning techniques have begun to be actively established in this area, several research and their result had conducted, and this is monitored based on the field's importance. The combination of all data will aid us in detecting trends and identifying developments in the experimental field, which will help us in our future experimental studies.
The main goal of this models is to review studies on prediction [17,22,27] of air pollution using ML algorithms which get the data from the sensor in the digital cities. Most of the important papers and articles were chosen using the most common databases.
ANTOINE ALLÉON, et al.,[10] concentrated on predicting the air pollution to given place and concentrated on air quality temporal variability. Air quality prediction isbuildby given monitoring station utilizing preceding measurements as well as weather forecasts. The past data are used to train the models, and the majority of recent papers that use this method use neural networks[30,37], specifically LSTM architectures. The four pollutants[11,15] expected by the engine are not always measured at each monitoring station. The experiments discussed in this paper were carried in Europe and the US, which are the two regions where we have obtained most accurate values.
S.NO TITLE TECHNIQUES FINDINGS
1 Index Using
Artificial Neural Network
Recognition of Future Air Quality
Multilayer Perceptron
We present the prediction techniques that were used to provide next day and next month air pollution counts which is used to avoid problems. The prediction of air pollution using Data mining which is the most fascinating and challenging tasks, and we present the prediction
techniques[18,19] that were used to provide next day and next month air pollution counts in order to avoid problems.
2 Air Quality Prediction -Deep Learning Model
LSTM[43] The main goal of the model focused on LSTM- (Long Short Term Memory) network method for predicting air quality range in a smart city.
3 Air Quality
Prediction Model (Most Relevant Spatial-Temporal Relations) using Adaptive Deep Learning (DL)
Convolutional Neural
Network-(CNN)
The main goal is to forecast the quality of air for up to forty eight hours and a combination of various neural
network[14,36] methods which includes ANN[12,31,33], LSTM, CNN [32]to implement seal-temporal connections.
4 Air Quality
Evaluation and Prototype-Data Mining Techniques
Principal Component Analysis method
The use of data mining to verify and forecast ambient air quality is proposed.
We may assess environmental air quality by calculating principal component ratings and using principal component analysis.
5 Air Quality
Prediction-(Deep Distributed Fusion Network)
Deep Neural Network (DNN) algorithm
Proposed DNN method that includes a spatial transformation function it uses a Deep Distributed Fusion Network. A former portion takes into account the spatial associations of air pollutants
6 Air Quality
Prediction- (Big Data & Machine Learning)
Genetic algorithm It is important to introduce Successful air quality monitoring models that collect data on pollutant concentrations and provide an estimation of air pollution in each area are critical. As a result, air quality assessment and forecasting has grown in importance as a research subject.
7 Air Quality
Prediction-
Predictive Data Feature Exploration
Data fusion model Use the LiRhtGBM model to solve the problem of analysing high-resolution of the large amount of data and as one of the data sources used to predict the air
quality.
8 Air Quality
Prediction-Artificial Intelligence
Long Short-Term Memory Deep Neural Network
Present a novel history prediction model that incorporates data from both surrounding pollution sources with an efficient air pollution prediction algorithm
9 Air Quality Prediction-Machine Learning
(ML)Algorithms
Neural Network (NN)[44,45,46,47]
In thesmart cities, Us paper review studies on with the help of data measured by the sensor and with the algorithm of machine learning (ML) we were able to find the quality of air. The most important papers were chosen by making use of the common databases and also with filtration in the database.
10 Large-Scale Air Quality Forecasting- Convolutional LSTM Network
Convolutional LSTM
Given the close ties between weather and air pollution, develop air quality forecasts.
PROPOSED FRAMEWORK USING DEEP LEARNING METHODS
Artificial Intelligence includes deep learning as a component. In machine learning (ML)[48], a program learns from data and knowledge using various computer algorithms. The computer does not need explicit programming. These will develop and modify the algorithm on their own.
Governments in populated and developing countries view air control as a major challenge.
Air pollution is caused by traffic and weather conditions, the burning of fossil fuels, and industrial waste such as power plant emissions. Weather forecasting has traditionally been done by simulating the atmosphere as a fluid. It is taken a sample of the present state of the atmosphere. The potential state of the atmosphere is calculated by solving thermodynamic and fluid dynamics numerical equations. However, when calculating the initial conditions of the atmosphere, the conventional system of differential equations that governs the physical model can become unstable due to disturbances and uncertainties.
Deep learning is used to foresee the outcome based on historical evidence. Deep learning is an artificial intelligence division (AI) which allows computers to learn without needing to be specifically programmed. The attention of machine learning is on developing applications which can transform when exposed to new knowledge, as well as the fundamentals of machine learning[16], such as the implementation of a simple computer learning algorithm using the .NET system. The use of advanced algorithms is used in the coaching and prediction process. It feeds the training data to an algorithm, which then uses that experience to make predictions based on brand new test data.
The aim is to look into quality of air making use of deep learning. After pre-processing air quality[13,21,42] dataset, supervised deep learning is help to analyze the forecasting of air quality. Furthermore, deep learning methods for weather forecasting will be used. As a result, it is critical that we analyze this data in order to extract any valuable knowledge and purpose. Figure 2 depicts the basic structure of the diagram.
Fig 2: Proposed framework DATASETS ACQUISITION
A data set (or dataset, though many modern dictionaries, such as Merriam-Webster, do not use this spelling) is a broad data gathering. Determined from the data is typically the simple network table contents or statistical data statistics[34], where every table column presents a unique variable, and a specific datum member for each row. For each member of the datum, the datum lists values for all of the variables, such as deep and volume of the object. The data set may contain data through one or so more members depending on no‟s of members. The term datum may refers more loosely that indicates that the information contained in a set of related records that correspond to a specific analyze or event. We may upload air datasets with attributes including CO, Pb, SO2, PM, NO2, O3. We may upload weather datasets such as temperature, humidity, wind speed, and lighting values in this module. These datasets were gathered from the Irvine central location at University of California.
PREPROCESSING
Data Preprocessing is the system of changing unprocessed raw information into a usable format. Some data is inaccurate, incomplete, contradictory and/or missing in certain trends. Data preprocessing is a tried and true method of addressing the existing challenges. Discarded Value changes are critical because if missing values aren't held as they should be, the data insights or overall performance of predictive models may be harmed. Imputation of lacking values from predictive techniques assumes that the person of such lacking observations is not decided at
random and that the variable selected to impute such lacking observations has a few courting with it, in any other case it is able to bring about vague estimates and remodel the dataset from unstructured layout to established layout. Then calculate the missing value evaluation and irrelevant data removal methods.
SIMILARITY MEASUREMENT
Crossroads among both Union and the Jaccard coefficient of similarity make up the index of Jaccard. The Jaccard index equates two sets of values that would determine which are shared and which are unique. It would be the percentage compare of how similar the two different sets of data are, ranging from 0% to 100%. Greater the percentage, the more similar the two populations. Although it is simple to use, small sample dimensions are highly sensitive and can produce incorrect results, particularly with very small samples or datasets with incomplete comments. This module uses common measures to cluster datasets[37]. Grouping the traits or properties using JACCARD resemblance. The common metric between subnets is calculated using the Jaccard distance. The more dissimilar means bigger the “distance”. The greater the difference between the means, the more identical they are. Data is grouped and displayed on a page based on distances.
DEEP LEARNING PROCESS
Deep Learning algorithms have the foreseeable future in the study of automated extraction of complex information descriptions in an automated manner. Deep Learning[39]
computerized can learn and represent data in a layered and hierarchical manner. Through a hierarchical learning process, Deep Learning and Big Data Analytical techniques allows for the elevated retrieval, hybrid abstraction as data representation.
Big Data analysis is a significant advantage of Deep Learning because it can obtain insight from data and is a Big Data Analytics platform when large amount of manually curated information present. Based on attribute threshold checking, we can create the air quality index[15] using the CNN algorithm. We will create rules based on temperature, humidity, and wind speed in this module. The CNN algorithm is used to create the threshold values for each attribute. When the dataset is large and there are many variables to consider, the CNN algorithm is not as simple as it appears in our original example. Pruning is required at this point. Pruning refers to removing branches from our decision tree that do not contribute substantially to our decision-making process.
CONCLUSION
Humans and plants also suffer from the impact of toxins in the air. Since there many different sources of emissions, determining the impact of air pollution on health is extremely difficult, and their individual effects vary greatly. The data has been preprocessed, and it can be further processed using a data mining tool to provide policymakers with appropriate decision support.
This project demonstrates how data mining methods can be useful in the fields of environmental monitoring and natural resource research. We were used preprocessed data to find Air Quality Index. Data mining[50] tool has been used for modelling, forecasting, and decision-making support. Convolutional Neural Networks[32,49] were accustomed to analyze the data of the
Artificial Neural Network[12] model in Data Mining techniques. This study suggests that a deep learning algorithm can increase quality prediction performance.
REFERENCES
[1] Ruchi Raturi, Dr. J.R. Prasad, “Recognition of Future Air Quality Index Using Artificial Neural Network” International Research Journal of Engineering and Technology (IRJET), 2018 [2] Murugesan, M., Thilagamani, S. ,” Efficient anomaly detection in surveillance videos based on multi layer perception recurrent neural network”, Journal of Microprocessors and
Microsystems, Volume 79, Issue November 2020,
https://doi.org/10.1016/j.micpro.2020.103303
[3] Soh, Ping-Wei, Jia-Wei Chang, and Jen-Wei Huang. "Adaptive deep learning-based air quality prediction model using the most relevant spatial-temporal relations." Ieee Access 6 (2018): 38186-38199.
[4] Thilagamani, S., Nandhakumar, C. .” Implementing green revolution for organic plant forming using KNN-classification technique”, International Journal of Advanced Science and Technology, Volume 29 , Isuue 7S, pp. 1707–1712
[5] Yi, Xiuwen, et al. "Deep distributed fusion network for air quality prediction." Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.
2018.
[6] Thilagamani, S., Shanti, N.,” Gaussian and gabor filter approach for object segmentation”, Journal of Computing and Information Science in Engineering, 2014, 14(2), 021006, https://doi.org/10.1115/1.4026458
[7] Zhang, Ying, et al. "A predictive data feature exploration-based air quality prediction approach." IEEE Access 7 (2019): 30732-30743.
[8] Rhagini, A., Thilagamani, S. ,”Women defence system for detecting interpersonal crimes”, International Journal of Advanced Science and Technology, 2020, Volume 29,Issue7S, pp. 1669–
1675
[9] Iskandaryan, Ditsuhi, Francisco Ramos, and Sergio Trilles. "Air quality prediction in smart cities using machine learning technologies based on sensor data: A review." Applied Sciences 10.7 (2020): 2401.
[10] K.Deepa, S.Thilagamani, “Segmentation Techniques for Overlapped Latent Fingerprint Matching”, International Journal of Innovative Technology and Exploring Engineering (IJITEE), ISSN: 2278-3075, Volume-8 Issue-12, October 2019. DOI: 10.35940/ijitee.L2863.1081219 [11] L. Wu and H. Zhao, „„Using FGM(1,1) model to predict the number of the lightly polluted day in Jing-Jin-Ji region of China,‟‟ Atmos. Pollut. Res., vol. 10, no. 2, pp. 552–555, Mar. 2019.
[12] Deepa, K., Kokila, M., Nandhini, A., Pavethra, A., Umadevi, M. “Rainfall prediction using CNN”, International Journal of Advanced Science and Technology, 2020, 29(7 Special Issue), pp. 1623–1627. http://sersc.org/journals/index.php/IJAST/article/view/10849.
[13] B. Liu, S. Yan, J. Li, G. Qu, Y. Li, J. Lang, and R. Gu, „„A sequence-to sequence air quality predictor based on the n-step recurrent prediction,‟‟ IEEE Access, vol. 7, pp. 43331–43345, 2019.
[14] Deepa. K , LekhaSree. R , Renuga Devi. B , Sadhana. V , Virgin Jenifer. S ,“Cervical Cancer Classification”, International Journal of Emerging Trends in Engineering Research, 2020, 8(3), pp. 804–807 https://doi.org/10.30534/ijeter/2020/32832020
[15] Q. Wu and H. Lin, „„A novel optimal-hybrid model for daily air quality index prediction considering air pollutant factors,‟‟ Sci. Total Environ., vol. 683, pp. 808–821, Sep. 2019.
[16] Santhi, P., Mahalakshmi, G., Classification of magnetic resonance images using eight directions gray level co-occurrence matrix (8dglcm) based feature extraction, International Journal of Engineering and Advanced Technology, 2019, 8(4), pp. 839–846
[17] R. E. Kalman, „„A new approach to linear filtering and prediction problems,‟‟ J. Basic Eng., vol. 82, no. 1, pp. 35–45, Mar. 1960.
[18Santhi, P., Lavanya, S., Prediction of diabetes using neural networks, International Journal of Advanced Science and Technology, 2020, 29(7 Special Issue), pp. 1160–1168.
[19] S. Hua, S. Wang, S. Jin, S. Feng, and B. Wang, „„Wind speed optimisation method of numerical prediction for wind farm based on Kalman filter method,‟‟ J. Eng., vol. 2017, no. 13, pp. 1146–1149, Jan. 2017.
[20] Santhi, P., Priyanka, T.,Smart India agricultural information reterival system, International Journal of Advanced Science and Technology, 2020, 29(7 Special Issue), pp. 1169–1175.
[21] X. Lai, T. Yang, Z. Wang, and P. Chen, „„IoT implementation of Kalman Filter to improve accuracy of air quality monitoring and prediction,‟‟ Appl. Sci., vol. 9, no. 9, p. 1831, May 2019.
[22] G. Galanis, P. Louka, P. Katsafados, I. Pytharoulis, and G. Kallos, „„Applications of Kalman filters based on non-linear functions to numerical weather predictions,‟‟ Ann. Geophys., vol. 24, no. 10, pp. 2451–2460, Jun. 2010.
[23] Vijayakumar, P, Pandiaraja, P, Balamurugan, B & Karuppiah, M 2019, „A Novel Performance enhancing Task Scheduling Algorithm for Cloud based E-Health Environment‟, International Journal of E-Health and Medical Communications , Vol 10,Issue 2,pp 102-117.
[24] S. V. Kumar, „„Traffic flow prediction using Kalman filtering technique,‟‟ Procedia Eng., vol. 187, pp. 582–587, Jan. 2017.
[25] . P. Pandiaraja, N Deepa 2019 ,” A Novel Data Privacy-Preserving Protocol for Multi-data Users by using genetic algorithm” , Journal of Soft Computing , Springer , Volume 23 ,Issue 18, Pages 8539-8553.
[26] M. Mastali, J. Vazquez-Arenas, R. Fraser, M. Fowler, S. Afshar, and M. Stevens, „„Battery state of the charge estimation using Kalman filtering,‟‟ J. Power Sour., vol. 239, pp. 294–307, Oct. 2013.
[27] . N Deepa , P. Pandiaraja, 2020 ,” Hybrid Context Aware Recommendation System for E- Health Care by merkle hash tree from cloud using evolutionary algorithm” , Journal of Soft Computing , Springer , Volume 24 ,Issue 10, Pages 7149–7161.
[28] T. Y. Zhou, B. Lian, S. Yang, Y. Zhang, and Y. Liu, „„Improved GNSS cooperation positioning algorithm for indoor localization,‟‟ Comput., Mater. Continua, vol. 56, no. 2, pp.
225–245, 2018.
[29 N Deepa , P. Pandiaraja, 2020 , “ E health care data privacy preserving efficient file retrieval from the cloud service provider using attribute based file encryption “, Journal of Ambient Intelligence and Humanized Computing , Springer , https://doi.org/10.1007/s12652-020-01911-5 [30] D. Plonis, A. Katkevičius, A. Krukonis, V. Šlegeryte, R. Maskeliunas, and R. Damaševičius,
„„Predicting the frequency characteristics of hybrid meander systems using a feed–forward backpropagation network,‟‟ Electronics, vol. 8, no. 1, p. 85, Jan. 2019.
[31 K Sumathi, P Pandiaraja 2019,” Dynamic alternate buffer switching and congestion control in wireless multimedia sensor networks” , Journal of Peer-to-Peer Networking and Applications , Springer , Volume 13,Issue 6,Pages 2001-2010.
[32] C. S. Yuan, „„Fingerprint liveness detection from different fingerprint materials using convolutional neural network and principal component analysis,‟‟ Comput., Mater. Continua, vol.
53, no. 3, pp. 357–371, 2017.
[33] Sumathi, K., Naveena, K., Prashanth, P., Revanthkumar, S., Srikeerthanaa, A.P.,E-health based patient surveilling device, International Journal of Emerging Trends in Engineering Research, 2020, 8(3), pp. 792–796
[34] Shankar, A., Pandiaraja, P., Sumathi, K., Stephan, T., Sharma, P. ,” Privacy preserving E- voting cloud system based on ID based encryption ” Journal of Peer-to-Peer Networking and Applications , Springer , https://doi.org/10.1007/s12083-020-00977-4.
[35] Z. Yang, L. Ce, and L. Lian, „„Electricity price forecasting by a hybrid model, combining wavelet transform, ARMA and kernel-based extreme learning machine methods,‟‟ Appl. Energy, vol. 190, pp. 291–305, Mar. 2017.
[36] P. Bento, J. Pombo, M. Calado, and S. Mariano, „„A bat optimized neural network and wavelet transform approach for short-term price forecasting,‟‟ Appl. Energy, vol. 210, pp. 88–97, Jan. 2018.
[37] Y. Zhang, G. Jia, N. Li, and M. Bai, „„A novel adaptive Kalman filter with colored measurement noise,‟‟ IEEE Access, vol. 6, pp. 74569–74578, 2018.
[38] R. Ferrero, F. Gandino, and M. Hemmatpour, „„Estimation of displacement for Internet of Things applications with Kalman filter,‟‟ Electronics, vol. 8, no. 9, p. 985, Sep. 2019.
[39] M. Langkvist, L. Karlsson, and A. Loutfi, "A review of unsupervised feature learning and deep learning for time-series modeling," Pattern Recognition Letters, vol. 42, pp. 11-24, Jun 1 2014.
[40] I. M. Coelho, V. N. Coelho, E. J. d. S. Luz, L. S. Ochi, F. G. Guimarães, and E. Rios, "A GPU deep learning metaheuristic based model for time series forecasting," Applied Energy, 2017.
[41] M. M. Rathore, A. Ahmad, A. Paul, and S. Rho, "Urban planning and building smart cities based on the Internet of Things using Big Data analytics," Computer Networks, vol. 101, pp. 63- 80, 6/4/ 2016.
[42] E. E. Agency, "Air quality in europe," 2016.
[43] B. Hochreiter and J. Schmidhuber, "Long Short Term Memory," Neural Computation, vol.
9, pp. 1735-1780, 1997.
[44] Z. C. Lipton, J. Berkowitz, and C. Elkan, "A Critical Review of Recurrent Neural Networks for Sequence Learning," 2015, 2015.
[45] K. Greff, R. K. Srivastava, J. Koutn´ık, B. R. Steunebrink, and J. Schmidhuber, "Lstm:A Search Space Odyssey," presented at the IEEE Transactions on Neural Networks and Learning Systems, 2016.
[46] D. Bahdanau, K. Cho, and Y. Bengio, "Neural Machine Translation by Jointly Learning to Align and Translate," in International Conference on Learning Representations, 2014.
[47] A. Graves, A.-R. Mohamed, and G. Hinton, "Speech recognition with deep recurrent neural networks," presented at the 2013 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2013.
[48] O. Vinyals and O. V. Le, "A Neural Conversational Model," presented at the Proceedings of the 31 st International Conference on Machine Learning, Lille, France, 2015.
[49] J. Donahue, L. A. Hendricks, S. Guadarrama, M. Rohrbach, S. Venugopalan, K. Saenko, et al., "Long Term Recurrent Convolutional Networks for Visual Recognition and Description,"
Conference on Computer Vision and Pattern Recognition(CVPR), 2015.
[50] J. Han, M. Kamber, and J. Pei, Data Mining: Concepts and Techniques: Morgan Kaufmann Publishers Inc., 2011.
