ABSTRACT “users can be problematic for techniques that are available at ‘using identity hiding tools such as anonymising proxies or This work presents computational models to address the VPN traffic. A model to detect usage of virtual with a Machine learning algorithms that was trained snsmission Control Protocol (TCP) header of captured | that the presented models are capable of classifying (originating directly from a user’s own device) or 1 hiding features of VPNs) with high degrees ofpersonality Analysis Using Machine Learning 8. CONCLUSION cost effective software system that can revolutionise human intelligence. This system can help various organisations build efficient human resource bodies by recruiting and this system is significantly more time efficient. for improvement. Many more classification algorithms of fide much better functionalities. Thus, a system that in the report can be built as proposed and also bepersonality Analysis Using Machine Learning 8. CONCLUSION cost effective software system that can revolutionise human intelligence. This system can help various organisations build efficient human resource bodies by recruiting and ght jobs. As compared to the traditional job recruitment , this system is significantly more time efficient. I for improvement. Many more classification algorithms of much better functionalities. Thus, a system that in the report can be built as proposed and also beABSTRACT ed to important advances in society. One of the most exciting applications s been in psychological science. It has been used for the development of powerfully predict human behavior and personality traits. The goal of software capable of analyzing the personality of applicants and ‘human resource recruitment process to comprehend its effectiveness. candidate with a questionnaire. The questions will be personality I be designed in such a way that the answers provided will reflect the platform in which the candidates can take the questionnaire will be d and used for further analysis. Machine learning algorithms that | words to determine personality traits are applied to arrive at the the applicant. This project will be based on The Big Five Factor Model, which is the most widely accepted personality By analyzing the correlation between the words in response is high, low, or in between in each one of the five qualities n of where the applicant falls on the spectrum of the five general disposition and potentially his/her future behavior. dicate aptitude for a particular kind of job. Thuscomes from agriculture which stands as the backbone now-a-days the crops are being infected by multiple h in turn damages the entire crop fields if they are and bacteria and also by worms and insects that with the help of emerging technologies like for disease detection. The proposed model (SRCNN) to improve the quality of the ‘that helps to detect the crop disease. When and illness of crops is also identified inTHROUGH FACE RECOGNITION 10, CONCLUSION ng voting system has many defects such as lengthy pro ‘security level but now we can say that our approach using three level of security cess, time \ existing system. Since, we are rs can be easily identified ‘The facial authentication the fraud voters, so we can avoid the bogus votes | cast their voting from anywhere By Jogin to our s every operation is performed through internet [goverament. Voters’ location is not important but tralized repository 30, data js accessible at any voting system provides updated result st each resources. The database needs to beupdeted aos may be enrolled and those who are deadTHROUGH FACE RECOGNITION ABSTRACT ‘category of biometric software which works by matching the facial the implementation of various algorithms in the field of secure ir levels of verification which were used for the voters in our u pasion. second is for the voter card number, third is a mobile number of the voter and the fourth level 1s algorithms for facial recognition. In this project, we these algorithms, namely: Eigen face, Fisher Face & democratic countries like India. Vote z means to choose , is called voting. In this project, we propose a system India and whose age is above 18 years they can vote the day of Election.§. CONCLUSION the -] tag CNN-Based frameworks LeNet-5 and ResNet-152 for the Arabic Numerals. In this, we have done the work on handwritten pe g with the datasets we have used and the feature we have considered for work, we have created a dataset on our own and then applied the CNN- jet-5 and ResNet-152. We have trained the model up to 100 epochs and the accuracy obtained for both frameworks. ResNet-152 gave a good - dataset with an accuracy of 96%-99%. Then we have tested the inputs from the random people and got good results with moreter Recognition of Hand-written Telugu Alphabets using CNN uw Alph ? the CNN-Based frameworks LeNet-5 and ResNet-152 for the ¢ Numerals. In this, we have done the work on handwritten atasets we have used and the feature we have considered for we created a dataset on our own and then applied the CNN- et-152. We have trained the model up to 100 epochs and ‘obtained for both frameworks. ResNet- 152 gave a good Fith an accuracy of 96%-99%. Then we have tested the s from the random people and got good results with moreABSTRACT sed Language in the Arab World, with Language is the broadest and most u: like e. There are various representations in the Arabic Number system ‘Arabic, Eastern Arabic, and Western Arabic, which are look-a-like. nition or Optical Character Recognition has gained popularity in on the importance of Optical Character Recognition (OCR) and cs such as Lenet-5, ResNet-101, and ResNet-152 available today t with more than 50,000 training and 7000 testing images of Arabic bers written by various age groups and named it LBREADB. were trained with LBREADB and tested on the rement criteria been made among the results based on the measut which were shown at the ending. iu-Arabic Numbers, Mashrigi Arabic, Optical Character 01, ResNet-152lividuals together who share similar fs communities together. As a result of the mindset that people ‘a business. Nowadays, a person may go a day without ‘have a friend, yet that same person cannot go a day without into several genres. Using machine fearing techniques, with a categorized list of music, The structure of music and umans are explained in relation to musical genre classification. data from audio signals rather than data from several shensive machine learning strategy for automatically ive is to create a machine learning system that outperforms existing je genres. In this study, we built a number of classification models GTZAN dataset, We compared and documented the results of all offunction will ead in the path to an audio fie, ancl time seris'(typesaeray) corresponding to Tate that was used to process the auclio. n in Telugu language by different speakers. 4.3 shows the graphs of each input signal X-axis represents the Sample and are plotted over few samples of input ‘because the actual data is very large in ‘and there will be a lot of noise in the audio Ihe summarized which it can be transformed features from the audio signal and using ‘than directly considering raw ‘and reduces the data which helps to d of learning. This process fs called that can be processed while signals of digits Zero to Nineract tthe digit spoken from a particular utter-An Efficient Deep Learni proach To Classify A Kidney With A Tumor Or A Stone Using CT Images 2.2. PROPOSED SYSTEM AND ADVANTAGES The main motive of this proposed model is to Provide a better system to predict kidneys with tumors or normal stones by using CT images. The existing system can classify the infectious stones frommnon-infectious stones. As, itis difficult to classify tumor stones from normal stones by sceing CT images with human naked eye, apply deep-learning approach to extract features from the images and can able to classify them. we proposed a model to classify them with best accuracy by using CNN(DenseNet -121) which would be helpful for Hospital management and Radiomic analysis. So that it would be helpful for the patients for treatment. Inorder to rectify the problem with the existing system, we proposed a model to detect the tumors based upon the real-time kidney CT scan images with CNN (DenseNet!21) which would be helpful for Hospital Management, Radiology Departments, Diagnosis, So that immediate action I can be taken and potentially save thousands of lives and Property. he saniathe Nett ofaeepane ofthe sytem byte wet Ts includes the system efficiently. The user must not be threatened bythe be increased so that he/she is able 10 make some ify tumor siones from normal stones by seeing approach to extract features from the the tumors which are present in kidney we exiting which will result in lower accuracy -with the existing system, we proposed = ‘CT-scan images of kidney with CNN pital Management, Radiology Department, different medical procedures.ABSTRACTg an individual's profile based on People could potentially be vith similar personalities, We could be made using a yghts, Our research mightner the relationship, tweets, and comments in the ul survey on this regular study operation. We | standards for Twitter usage using the same and connected to forecast a user's examining the social and linguistic ights. We got maximum accuracy ofABSTRACT Social media has captured the interest of today's youth and available on the internet. Researchers can better investigate and Je: i behavior, interests, and personality traits thanks to the activity of social media users. ‘We're conducting our research on Twitter, a popular social media network with « lary user base. The user data can be retrieved and subsequently analyzed based on the language, status profiles, or preferences provided by other Twitter users to dete the person's nature. The user's remark, tweet, social behaviour, and language habits Twitter are used as data for analysis to better understand the various types of texts. The ‘user's remark, tweet, social behaviour, and language habits on Twitte for analysis to better understand the various types of texts. This res / by processing the data and the user's actual predicted personality using, © Learning techniques. We can determine people's emotions and climinate ts by recognising these good and negative tweets. We divided our two parts: train and test. The train data is used to train the model, and the ‘emotions in tweets by comparing the train and test a. The using SVM, Nave Bayes, and LDA algorithms. We use tweepy (and use Random Forest and Extra Gradient Boosting techniques arumour is true or not. We projected the best outcome with 90% algorithm for personality prediction and 99.61 percent Gradient Boosting for rumour prediction based on the e usedCHAPTER-9 19 epidemic will cause a massive international and government agencies around the world are will affect a disproportionately large amount of the Tesearch, an associate degree ML-based prediction east the global threat of COVID-19 irruption. The ining day-by-day historical data and provides ichine learning methods. Overall, we believe this state of affairs Area, and that they can be AS a result, the study forecasts may be of actions and making decisions to manage telling will be one of our primary areas of10 enhance the choice creating on the long run course 's have a fast result on each field of life or the thods are being deployed to handle prediction that using these methods for this pandemic of ifically, 3 commonplace prediction models, like (GBM), Light Gradient Boosting Machine ms are utilized in this study to forecast 3 styles of predictions are created by every nfected cases, the quantity of deaths, and xt ten days. The results created by the these ways for this situation of theABSTRACT ML) primarily based prediction Mechanism have proven their il areas to enhance the choice creating on the long run course advancements have a fast result on each field of life or the tion methods are being deployed to handle prediction as created that using these methods for this pandemic of Specifically, 3 commonplace prediction models, like achine (GBM), Light Gradient Boosting Machine orest mechanisms are utilized in this study to forecast VID-19. 3 styles of predictions are created by every yf new infected cases, the quantity of deaths, and ‘the next ten days. The results created by the m to use these ways for this situation of theABSTRACT \ 16.4 prominent component of online social networks (OSN’s) Me their security to post, remark and label the photograph for In this paper, the issue is to take care and concentrate the an individual not 10 possess himself (called short co- Protection holes of a photograph, we will plan a of posting every ii jual in the photograph and ‘For this reason, we have created a new setting where t ‘on their interest which includes the options like ‘agreement way to convey a private preparing erior to other methodologies as far as we have cy Our methodology is done as proof on android ; management we aimed to match the achieved d in the accurate manner which increases ratiod to give solutions for many problems in knowledge and data pplication of machine learning is the prediction of an outcome machine learns pattems from the existing dataset, and then in order to predict the outcome. Classification is a nique that is commonly used for prediction. Some ‘with satisfactory accuracy, whereas others exhibit a limited investigates a method termed Logistic Regression sing the accuracy of weak algorithms. Other classifiers ting 1, SGD Classifier: Existing 2. Implementation with lung cancer disease dataset. A comparative analytical v the regression technique can be applied for improving ‘The focus of this method is not only on zation algorithms, but also on the implementation of how its utility to predict disease at an early stage.to . inet Lada, Black Pest and Leaf present in an image which is given as 1 Lat, the scope of this project by adding some more classes too This 10 wet id eine {YPE of the insect present on their crop. So that they can use iene oe ‘them from the field if they are harmful to the crop. For this Tadha, Black pest and normal leaf from the field of cotton and ahs of our project by adding more data sets and different insects ne Elanning to collect more data sets. We will include some other hewitt Proiect will detect some more classes too. This project can es Continuously in the field and detect the insects & will get an alert message whenever this moving camera fies‘music player js i oF cn Es more accuracy and less computation Reece time to capture the image and select Mee, could give physically challenged people a stress and also act as a boon for the music Songs manually is eradicated and the songs that soothe Based Music Player can be installed in cars and it can ‘based on that particular emotion easily. This Emotion further to identify more emotions of the user accurately andABSTRACT ‘The human face also plays rtof'a human body is the human face itself n of an individual’s current emotional state. Music can ually segregating the list of songs and creating playlists according be very exhaustive, time consuming, boring and labor intensive cas well as saving humans from @ lot of boring hardship a Il as developed. However these algorithms have their slow, less accurate and some of them even require can obviously add up to the overall cost. Through this tion which would automate the process of playlist ‘as time required to segregate the list of songs. And time in obtaining the results, improvingnsembled Activation LSTM Nework ; to balance their financial status by earning their money apart is may sometimes deviate the routines and lead to the death ilies. So, the proposed system developed stacked LSTM oth volumes and share values per day. The proposed modelABSTRACT ‘prediction is that the scene of trying to complete the long-run value of company stock. swe take the price. By using the neural network, we develop a model. Within the neural ‘neural network that remembers each and each information through time. tial commodity forthe survival of human beings, irrespective of their social economics money, one of the easy money eaming is spending their money on shares. Every le invested in shares may get profit or loss. So, Stock prediction is gaining more forecast the value based on time series and recommend a good share to the Ive the stock prediction problem using artificial networks which deals only with set but the major decision element of any stock prediction system is previous es this issue by designing LSTM network by using ensemble activation architecture i.e. standard values for all the parameters which is not e proposed system identifies the hyper-parameter values for each n these values and ensemble activators to suit for generalized stocktowards MRS with CF Based On Machine Learning CONCLUSION 10 avoid the use of content-based filtering, the Item-based CF filtering “Obtaining better results. KNN collaborative recommendation system is Iby employing Movielens dataset containing 28M rating for are compared and found that the proposed system is more reliable ‘when the proposed methodology is applied to different larger increase which proves that our system is both accurate and filtering is more convenient than user-based The main aim on algorithm and to provide better results. The research » fulfill the aim of the project. In the future, more features actor, genre, casting details, ete) to make ‘The content-based filtering and collaborative ‘and improve the performance as hybridABSTRACT systems come much handier in these situations. This and performance of a regular filtering technique. Although n 4 recommendation system, collaborative filtering is the at from the users, rechecks his/her history/past behavior of friends, In this project, to prove the effectiveness, KNN algorithms iinly focus on enhancing the accuracy of results as compared larity using k-nearest neighbor with the help the same time removing the drawbacks of content-based ance is preferred, cosine similarity is used as the accuracy of remain almost the same.ABSTRACT ‘year, as well as considerable financial losses. Road accidents are luding the road surface, weather conditions, lighting conditions, can be better understood and forecast, it may be possible to take ese determinants. This project intends to create models that can can be utilized to help an accident data analysis and classify d on historical data, this study provides a predictive model forthe sign language So,that thay are not People, but they often find it difficult toonal data exchange will alsoon system where it recognize the emotions in tweets. can see that many people use social media where they et in a good way and some of them in a bullying carried out on tweets to analyze public opinion society. In this project, by using machine by classifying tweets as positive and ‘we can identify people emotions where we our dataset into train and test dataset, where ta with the test data, the model recognizes the ithms we classify the text based on twitter into anger, sadness, joy. Based on the performance 82% Fl score.* ABSTRACT {introduced among the people, itis important to know about people's is where sentiment analysis comes into play. Model that’s built for $ 0 predict the sentiment of the content. The audios that we have taset for the system and our proposed model mainly overcomes the involves converting of other language audio reviews to English extraction are applied on the text before training the model. The data based on testsize parameter. Initially, we have carried ‘Supervised machine leaming algorithms i.e., SVM, RandomForest, ral network based deep learning algorithm LSTM used with and stratified sampling is done and we have chosen of the accuracy results produced by system when SVM, RandomForest, Multinomial NB, LSTM with 80.02%, 77.08%, 76.49%, 72.36%, 68.24% 67%, 81.67%, 74.17%, 60.84% on JVD dataset best accuracy, we have selected it as the algorithm for‘18 a challenging application. It has been widely used in cars, robotics, security tracking, and guiding visually ¢ understanding have been improved through deep different network architectures, but with the same nplex images. A large aerial image dataset called MS s. Raspberry Pi can be used as an external hardware pera and sent to Raspberry Pi which trains the data h accuracy and high speed i.e. produces the O algorithm. When it detects the objects it e. So it becomes easy for the user to understand d for autonomous cars, robotics, security, and detect objects around them.forged images created ‘image with the white part isa black and whites a major crop has a huge impact on people's health as $0 produces jobs and many small industries depend directly g is affected by various diseases and pets which can result seience and technology a number of research projects e yields. According to the plant nutrient system, leaf en (N), phosphorus (P), and potassium (K) of data augmentation to increase the size DenseNet-121 which is one of the DenseNet ation. It is then evaluated with Adam, RMSprop, £,98.85%,93.10% of accuracy. This entire ould be used in order to increase the yield of1. INTRODUCTION 'S APP is an Advanced and Informative System helping the users to ting things happening around their Locality. This System helps ‘up to date about the happenings around their locality or in their Information Communication and Technology. We can’t think a From morning to night, we need help of the technology. This . Most of the works depends on web. For this reason, a app by internet at low cost and we can find our expectable p. At present information is one the most valuable resource ‘project so that we can aware the people. There are 2 another for the user. The user is allowed to view related to any city. Admin will monitor if any abuse and will take some action. The Front end used is is shown first and so on, the user can also refresh ‘news will be closed. The user can add aABSTRACT +t actoss the globe. In order to reduce the spread of this ‘been implemented, such as wearing a facemask that to avoid exposure to the virus in a public setting. In this the growth of COVID-19 by facial recognition and ‘or not. To achieve this, we will use the "Viola Jones ‘rapid detection of objects running at a rate of 15 real-time object detection. This algorithm can Viola Jones is trained on a dataset that consists of we create our own dataset by capturing images , and use it as the dataset. Performance of this time inputs and then detect if the person is R‘a healthy diet. Seeds are classified based is method necessitates the extraction of a n, which is inefficient in detection. Deep identification in recent years. For seed on pea, and groundnuts in ourABSTRACT ‘world's most pressing concerns, as they result in a large number of ich year, as well as considerable financial losses. Road accidents are including the road surface, weather conditions, lighting conditions, ts can be better understood and forecast, it may be possible to take the damages. Using accident data from Andhra Pradesh, India, ese determinants. This project intends to create models that can can be utilized to help an accident data analysis and classify d on historical data, this study provides a predictive model for algorithms will be used to create this model. To tackle the collection, the SMOTE with ENN (Edited Nearest ‘Adaboosting algorithm techniques for feature selection @