Machine Learning Applications to Smart City PDF

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/331346839 Machine learning applications to smart city Article · February 2019 DOI: 10.19101/TIPCV.2018.412004 CITATIONS...

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/331346839 Machine learning applications to smart city Article · February 2019 DOI: 10.19101/TIPCV.2018.412004 CITATIONS READS 19 9,362 1 author: Badri Mohapatra AISSMS Institute of Information Technology 22 PUBLICATIONS 109 CITATIONS SEE PROFILE Some of the authors of this publication are also working on these related projects: Microcontroller based dual axis solar tracking system View project All content following this page was uploaded by Badri Mohapatra on 26 February 2019. The user has requested enhancement of the downloaded file.  ACCENTS Transactions on Image Processing and Computer Vision, Vol 5(14) ISSN (Online): 2455-4707 Review Article http://dx.doi.org/10.19101/TIPCV.2018.412004 Machine learning applications to smart city Badri Narayan Mohapatra1*and Prangya Prava Panda2 Assistant Professor, Department of Instrumentation & Control, AISSMS IOIT, Pune, India1 B.Tech Student, Department of Computer Science, The Techno School, Bhubaneswar, India2 Received: 21-December-2018; Revised: 23-February-2019; Accepted: 25-February-2019 ©2019 Badri Narayan Mohapatra and Prangya Prava Panda. This is an open access article distributed under the Creative Commons Attribution (CC BY) License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Abstract The basic need of human is increasing as they interact with different devices and also, they provide many feedbacks. Many smart devices generate high data and that can be retrieved and reviewed by humans. Applications are not fixed as it increases day to day life. Based on these data generated by different smart devices and smart city applications machine learning approach is the best adaptive solution. Rapid development in software, hardware with high speed internet connection provides large data to this physical world. The key contribution of this paper is a machine learning application survey towards smart city. Keywords Smart city, Machine learning, Machine learning algorithm, Smart city application. 1.Introduction The way we interact with the world, there is beautiful Driving style and driving condition have a big impact machine learning technologies can have the on vehicle fuel emission as well as consumption. prediction the data driven decision with ability to Critical knowledge about road type and related to monitor and manage advanced devices smoothly. congested traffic will get help by implementing machine learning algorithm. Machine learning 2.Smart city will help the power for control of hybrid vehicle. Smart refers to demand of more services to increase To reduce delays in traffic and to reduce emissions or complexity with increase population. Smart means it we can say for optimization through the prediction should handle the problem intelligently with through looking past and present scenario by improved service like energy, public health, observing the complete set of combination then transportation, traffic management and many more. machine learning will approach the best optional set Smart city uses internet of things for fast for multiple traffic controller. Machine learning communication and easy pass of information. Smart like linear regression and neural network helps in means to enhance starting from education, estimation of energy which will be helpful in smart entertainment, business, city wastage, utilities all to city. increase the quality of living. It provides an integrated approach for successful critical planning Cities that are supported by an extensive digital goals. Now a day‟s machine learning has become infrastructure of sensors, databases and intelligent remarkable and used in many applications like applications accurate identification of criminal medical diagnosis and also used in computer vision. activity easily possible in smart cities using deep It is also a promising technique to enhance society learning. Machine learning also help by providing and also useful to find or identify malicious in cyber information to the flight management system also world. Figure 1 shows the applications of machine helps to pilot for taking some risk decision. learning. Modern life will change by using machine learning  Machine learning application is more towards algorithm based on historical approach. smart city.  Machine learning algorithm helps to smart city mission. *Author for correspondence 1 Badri Narayan Mohapatra and Prangya Prava Panda Also, the air quality can be calculated by different concentration and percentage range. To train an effective model, a huge amount of data from past decade entitled as AIRNet which will predict air quality by machine learning approach. The approach of machine learning is to predict the generation of solar power from the past, whether the database. The reflection of data from both researchers and practitioners has emerged important to business intelligence and all this kind of emerging research through machine learning effect the characteristic of the business analytics. Transparent machine learning will require depth knowledge for empowering beyond human computation. Identifying bubbles in a substrate Figure 1 Applications of machine learning and repeatable and qualify able from the dust through machine learning. The complex relationship 3.Machine learning applications between surface, ground and climate water can easily Cell profile learning analyst function can easily do implemented in machine learning algorithm which by machine and more suitable for image base will help agriculture region. measurement. Patient based deep brain close loop simulator can be created using machine learning. Risk of figures and early detection of myrtle rust can Personal sensing for translating row sensor and status be possible by machine learning. Machine related to human behaviour and mental health learning has a greater impact of nitrogen estimation, according to the use of computers, social media & lead this large and makes a great attention in smart phones, machine learning can know easily the agriculture. Machine learning makes high impact status of the mental requirements. Early on production in agricultural. Huge information detection also possible for sarcoma patients. from the different multispeed camera will predict to Analysis & study of diabetes through machine estimate corn crops. Linear regression of learning will be helpful to doctors easily. Health machine learning help to agriculture. based large electronic resends as well as large valuable information with machine learning give a Home automation by a speech and device very good remarkable to the health science by reorganization and control by machine learning very examining diabetes. Machine learning, helpful number of needs of the home [30, 31]. Wide prediction about solar radiation and solar output verities operational data from home as well as live power for grid operation in grid point management system are helpful for optimization and also in saving. The machine learning approach will be very energy more. Suicide and emotion concept helpful for realization of comfort with the addition of identify through machine learning. By use of energy saving. Boosted regression tree, random large amounts of data set with the use of smart home forest tree and boosted tree is the machine learning sensors, machine learning predicts human activity approach which will produce the potential of most [34, 35]. Machine learning approach predicts human value able ground level fresh water. activity by smartly considering human-computer interaction and pattern reorganization methods [36, Machine learning which could significantly improve 37]. Figure 2 shows the applications based on smart the best monitoring the air quality by calibrating real city. time pollutant data sensor package. 2  ACCENTS Transactions on Image Processing and Computer Vision, Vol 5(14) Road Traffic Air Predicting atrial fibrillation Imaging experiments Volitional control for deep brain Machine Health Understanding mental health Learning Early detection of hepatocellular Carcinoma Diabetes research Early prediction of asthma Solar radiation forecasting Environment Energy management (Occupant) Figure 2 Applications based on smart city Ground water mapping 4.Machine learning algorithm There is different type of clustering methods like K-means algorithm is the simplest partitioning hierarchical, parametric and density based. K-means, method for clustering analysis and widely used in mixture models comes under parametric clustering. data mining applications. One class support vector machine (OCSVM) E   kK1 xCk d 2 (x, mk ) (1) algorithm maps input data into a high dimensional Optimal partition achieved via minimising the sum of feature space and iteratively finds the maximal squared distance to its “representative object” in each margin in the hyper plane which best separates the cluster for example Euclidean distance. training data from the origin. The machine learning N algorithm can classify as unsupervised, supervised d 2 (x, m k )   (xn m kn ) 2 (2) and reinforcement learning. Modifications and n1 optimization of analytical data very easily through The process of grouping a set of objects into classes machine learning algorithm. Linear regression based of similar objects is known as clustering. Equation 1 on multiplied with some constant. Breaking into and 2 used for k-means algorithm. smallest data set with recursion until all data set used. LASSO regression means „Least absolute shrinkage Machine learning investigates the mechanisms by and selection operators. Figure 3 shows the which knowledge is acquired through experience. classification of supervised learning. 3  Badri Narayan Mohapatra and Prangya Prava Panda Supervised Learning Regression Classification Support vector K-nearest neighbour (KNN) Linear regression Logistic regression Decision tree regression Decision tree LASSO regression Naïve Bayes Figure 3 Classification of supervised learning special application of machine learning to solve problems. In supervised learning, machine take set of According to classification KNN is usually used for examples where in unsupervised machine tries to find its simplicity. We can calculate the posterior hidden unlabelled or unstructured data. Different type probability, by Bayes theorem and easy to handle of useful algorithm used for different smart complex parameter estimation. application shown in Table 1. For appropriate output one should use a machine learning algorithm to get In case of unsupervised learning the algorithms desired results. organize data into a group of clusters to make a simple structure description. Reinforced learning is a Table 1 Purpose of machine learning algorithm Machine learning algorithms Purpose Feed forward neural network Smart health Densities based clustering and regression Smart citizen K-means Smart city, Smart home Clustering & anomaly detector Smart traffic One class support vector machine Smart human active control Support vector regression Smart whether Linear regression Smart market analysis Several data mining the intrusion detection can be In future more studies on many other open questions easily possible by supervised, semi-supervised and involving investigations algorithms based on smart unsupervised machine learning. city should be investigated with different collaborative approach. More development and more 5.Conclusion implementation can be possible due to the machine The machine learning algorithm provides services learning algorithm and machine learning algorithms smart city. The services include energy motilities. have the potential to make it successful. Choosing suitable algorithm can give a good result for the important issue. Big data should be required Acknowledgment for accuracy in machine learning algorithm. Machine None. learning has the challenges with variety, volume and velocity increasing with big data management. To Conflicts of interest The authors have no conflicts of interest to declare. reach good and suitable decision as there are different algorithms are there to use to get appropriate output. 4  ACCENTS Transactions on Image Processing and Computer Vision, Vol 5(14) References Finkelstein J. Machine learning approaches to Park J, Chen Z, Kiliaris L, Kuang ML, Masrur MA, personalize early prediction of asthma exacerbations. Phillips AM, et al. Intelligent vehicle power control Annals of the New York Academy of Sciences. 2017; based on machine learning of optimal control 1387(1):153-65. parameters and prediction of road type and traffic Voyant C, Notton G, Kalogirou S, Nivet ML, Paoli C, congestion. IEEE Transactions on Vehicular Motte F, et al. Machine learning methods for solar Technology. 2009; 58(9):4741-56. radiation forecasting: a review. Renewable Energy. Murphey YL, Park J, Chen Z, Kuang ML, Masrur 2017; 105:569-82. MA, Phillips AM. Intelligent hybrid vehicle power Konis K, Annavaram M. The occupant mobile control-Part I: machine learning of optimal vehicle gateway: a participatory sensing and machine-learning power. IEEE Transactions on Vehicular Technology. approach for occupant-aware energy management. 2012; 61(8):3519-30. Building and Environment. 2017; 118:1-3. Baluja S. Continuous selection of optimized traffic Naghibi SA, Pourghasemi HR, Dixon B. GIS-based light schedules: a machine learning approach. In groundwater potential mapping using boosted international conference on intelligent transportation regression tree, classification and regression tree, and systems 2018 (pp. 3205-11). IEEE. random forest machine learning models in Iran. Ertugrul ÖF, Kaya Y. Smart city planning by Environmental monitoring and assessment. 2016; estimating energy efficiency of buildings by extreme 188(1):44. learning machine. In international smart grid congress Zimmerman N, Presto AA, Kumar SP, Gu J, and fair 2016 (pp. 1-5). IEEE. Hauryliuk A, Robinson ES, et al. A machine learning Bikmetov R, Raja MY, Kazi K, Mouftah H, Kantarci calibration model using random forests to improve ME, Rehmani MH. Machine-to-machine sensor performance for lower-cost air quality communications in the smart city-a smart grid monitoring. Atmospheric Measurement Techniques. perspective. In Transportation and Power Grid in 2018; 11(1):291-313. Smart Cities: Communication Networks and Services Liang X, Li S, Zhang S, Huang H, Chen SX. PM2. 5 2017. UK: John Wiley. data reliability, consistency, and air quality assessment Chackravarthy S, Schmitt S, Yang L. Intelligent crime in five Chinese cities. Journal of Geophysical anomaly detection in smart cities using deep learning. Research: Atmospheres. 2016; 121(17). In international conference on collaboration and Zhao S, Yuan X, Xiao D, Zhang J, Li Z. AirNet: a internet computing 2018 (pp. 399-404). IEEE. machine learning dataset for air quality forecasting. Watkins D, Gallardo G, Chau S. Pilot support system: 2018. a machine learning approach. In international Sharma N, Sharma P, Irwin D, Shenoy P. Predicting conference on semantic computing 2018 (pp. 325-8). solar generation from weather forecasts using machine IEEE. learning. In international conference on smart grid Hill NR, Ayoubkhani D, McEwan P, Sugrue DM, communications 2011 (pp. 528-33). IEEE. Farooqui U, Lister S, et al. Predicting atrial fibrillation Chen H, Chiang RH, Storey VC. Business intelligence in primary care using machine learning. Available at and analytics: from big data to big impact. MIS SSRN 3236366. 2018. Quarterly. 2012:1165-88. Bray MA, Carpenter AE. Quality control for high- Shih PC. Beyond Human-in-the-loop: empowering throughput imaging experiments using machine end-users with transparent machine learning. In learning in CellProfiler. In high content screening human and machine learning 2018 (pp. 37-54). 2018 (pp. 89-112). Humana Press, New York, NY. Springer, Cham. Houston B, Thompson M, Ko A, Chizeck H. A Xu D, Offner SS. Assessing the performance of a machine-learning approach to volitional control of a machine learning algorithm in identifying bubbles in closed-loop deep brain stimulation system. Journal of dust emission. The Astrophysical Journal. 2017; Neural Engineering. 2018; 16(1):1-9. 851(2). Mohr DC, Zhang M, Schueller SM. Personal sensing: Sahoo S, Russo TA, Elliott J, Foster I. Machine understanding mental health using ubiquitous sensors learning algorithms for modeling groundwater level and machine learning. Annual Review of Clinical changes in agricultural regions of the US. Water Psychology. 2017; 13:23-47. Resources Research. 2017; 53(5):3878-95. Książek W, Abdar M, Acharya UR, Pławiak P. A Heim RH, Wright IJ, Chang HC, Carnegie AJ, Pegg novel machine learning approach for early detection of GS, Lancaster EK, et al. Detecting myrtle rust hepatocellular carcinoma patients. Cognitive Systems (Austropuccinia psidii) on lemon myrtle trees using Research. 2019; 54:116-27. spectral signatures and machine learning. Plant Kavakiotis I, Tsave O, Salifoglou A, Maglaveras N, Pathology. 2018; 67(5):1114-21. Vlahavas I, Chouvarda I. Machine learning and data Chlingaryan A, Sukkarieh S, Whelan B. Machine mining methods in diabetes research. Computational learning approaches for crop yield prediction and and Structural Biotechnology Journal. 2017; 15:104- nitrogen status estimation in precision agriculture: a 16. review. Computers and Electronics in Agriculture. 2018; 151:61-9. 5  Badri Narayan Mohapatra and Prangya Prava Panda Kamilaris A, Kartakoullis A, Prenafeta-Boldú FX. A Scheurer S, Tedesco S, Brown KN, O'Flynn B. review on the practice of big data analysis in Human activity recognition for emergency first agriculture. Computers and Electronics in Agriculture. responders via body-worn inertial sensors. In 2017; 143:23-37. international conference on wearable and implantable Arroyo JA, Gomez-Castaneda C, Ruiz E, De Cote body sensor networks 2017 (pp. 5-8). IEEE. EM, Gavi F, Sucar LE. UAV technology and machine Hassan MM, Uddin MZ, Mohamed A, Almogren A. A learning techniques applied to the yield improvement robust human activity recognition system using in precision agriculture. In humanitarian technology smartphone sensors and deep learning. Future conference 2017 (pp. 137-43). IEEE. Generation Computer Systems. 2018; 81:307-13. Oguntunde PG, Lischeid G, Dietrich O. Relationship Soheily-Khah S, Marteau PF, Béchet N. Intrusion between rice yield and climate variables in southwest detection in network systems through hybrid Nigeria using multiple linear regression and support supervised and unsupervised mining process-a vector machine analysis. International Journal of detailed case study on the ISCX benchmark dataset. Biometeorology. 2018; 62(3):459-69. 2017. Soni N, Dubey M. A review of home automation system with speech recognition and machine learning. Badri Narayan Mohapatra started his International Journal of Advance Research in research on communicationatnthe Computer Science and Management Studies. 2017; Centurion Universityy of Technology 4(5):32-5. and Management, Odisha. He did his Jiang L, Liu DY, Yang B. Smart home research. In B.E. from Berhampur University, proceedings of international conference on machine Odisha and M.Tech from Biju Patnaik learning and cybernetics 2004 (pp. 659-63). IEEE. University of Technology, Rourkela, Barker S, Mishra A, Irwin D, Cecchet E, Shenoy P, Odisha. His research area covers Light Albrecht J. Smart*: an open data set and tools for Propagation, Digital Signal and Image Processing. enabling research in sustainable homes. SustKDD. Email: [email protected] 2012; 111(112). Just MA, Pan L, Cherkassky VL, McMakin DL, Cha Prangya Prava Panda did her C, Nock MK, et al. Machine learning of neural Diploma from Government representations of suicide and emotion concepts Polytechnic, Bhubaneswar, Odisha. identifies suicidal youth. Nature Human Behaviour. Now she is continuing B.Tech in CSE 2017; 1(12):911. Department in The Techno School, Singh D, Merdivan E, Psychoula I, Kropf J, Hanke S, Bhubaneswar, Odisha. Geist M, et al. Human activity recognition using recurrent neural networks. In international cross- domain conference for machine learning and knowledge extraction 2017 (pp. 267-74). Springer, Cham. 6 View publication stats