IEEE machine learning Projects
- Natural Language Processing and Computer Vision for the Detection of Potentially Illicit Messages on Twitter and Linked Websites
- Evaluation of Subjective Answers Using Machine Learning
- ML approach for musical therapy using facial expressions
- Dynamic traffic management system based on IoT and image processing
- A CNN-based Framework for Comparison of Contactless to Contact-based Fingerprints
- Air Pollution Prediction System for Smart City using Data Mining Technique: A Survey
- Phishing Web Sites Features Classification Based on Extreme Learning Machine.
- Machine Learning for Web Vulnerability Detection: The Case of Cross-Site Request Forgery.
- Image processing based Tracking and Counting Vehicles
- SMS Spam Detection using Machine Learning Approach.
- Rainfall Prediction using Regression Model.
- Human Activity Recognition using Open Cv & Python
- Customer Segmentation using K-Means Algorithm
- Crime analysis and prediction using data mining techniques
- Raspberry pi based hand gesture recognition & voice conversion system for dumb people.
- django framework for CV analysis and personality prediction using big 5 traits
- Driver Drowsiness Monitoring System using Visual Behavior and Machine Learning
- Assistive Device for Blind, Deaf and Dumb People using Raspberry-pi
- Classification of Cancerous Profiles using Machine Learning
- Voice based Email for Blind.
- Disease Prediction by Machine Learning Over Big Data From Healthcare Communities
- Object Detection, convert object name to text and text to speech.
- Raspberry Pi based unauthorized car parking system by capturing the number plate of a vehicle using open cv technique thereby alarming the police & vehicle owner.
- Application of Data Mining Methods in Diabetes Prediction
- An Approach to Maintain Attendance using Image Processing Techniques
- Sentiment Analysis of Top Colleges
- Detection of fake online reviews using semi-supervised and supervised learning
- Prediction and Diagnosis of Heart Disease Patients using Data Mining Technique
- A Deep Learning Approach for Face Detection using YOLO
- Analysis of Chronic Kidney Disease Dataset by Applying Machine Learning Methods
- Random Forest for Credit Card Fraud Detection.
- Smart Voting System through Facial Recognition
List of Machine Learning Project ideas
1. Natural Language Processing and Computer Vision for the Detection of Potentially Illicit Messages on Twitter and Linked Websites
The global problem of human trafficking causes millions of victims to lose their dignity. At the moment, this crime is spread online through social networks using covert messages that promote these illegal businesses. Given the limited resources available to law enforcement in this case, automatically identifying texts that may be associated with this incident and may also be used as clues is essential. In the current study, we search tweets for tweets that potentially promote these illegal services and exploit children. We do this using natural language processing. Since the photographs and URLs retrieved from suspicious communications were processed and categorised by gender and age group, it is possible to locate images of children under the age of 14. Here's how it was done.
First, real-time mining is carried out on tweets containing relevant hash tags for minors. Prior to determining if a tweet is suspicious or not, background noise and typos are removed from it. Additionally, the geometrical features of the body and face are chosen using Haar models. SupportVector Machine (SVM) and Convolutional Neural Network (CNN) can be used to determine a person's gender and age group, even if the face's details are hazy or the torso is not proportional to the head. As a result, the SVM model that just uses torso characteristics outperforms CNN.
2. Evaluation of Subjective Answers Using Machine Learning
Currently, subjective writing is evaluated using negative methods. An important role is assessing the subjective responses. The quality of an appraisal made by a human can vary based on their emotional state. In machine learning, the user's input data determines every outcome. Our proposed strategy makes use of machine learning and NLP to address this issue. Our system performs tasks including tokenizing words and phrases, categorising speech components, chunking, chinking, lemmatizing words, and wordneting in order to assess the subjective reaction. Our proposed method additionally provides the semantic meaning of the context. Two modules make up our system. The first step entails getting the data from the scanned photographs and properly organising it. The second involves text analysis using ML and NLP.
3. ML approach for musical therapy using facial expressions
The majority of people believe that music is expressive and that its expressiveness can be easily linked to people's emotions, and that almost all types of human emotions have a direct relationship with the specific music genre. Looking at someone's facial expressions usually reveals whether they are joyful, sad, furious, terrified, depressed, or tender. Music can affect a person's emotions, which may also have an effect on their mood and health. Music therapy is one of the initial treatments for several psychological conditions. An intelligent system that arranges a music library according to the genres that each song represents
The combination of musical therapy and facial emotion detection then creates a well-suited music playlist for the patients depending on their facial emotions. The image is run via the patients' skills for emotion and facial identification. The best playlist for this emotion is then recommended along with the songs that go with it. resulted in patients feeling calm and relaxed.
4. Dynamic traffic management system based on IoT and image processing
The rise in the number of cars has made traffic management a serious problem. The manual traffic system is inefficient. This study blends image processing and the Internet of Things to create an adaptive traffic control system (IoT). The recommended system has the ability to analyse real-time data using image processing. The numerous lanes are continuously observed by cameras. examining the information gathered from different channels. Finding and counting the number of vehicles in each lane is done via image processing. The count is sent from each lane to the central processing unit. Once the algorithm based on vehicle count has computed the waiting time for each lane, the signal lights will be set.
This method reduces the typical waiting time while increasing the effectiveness of the traffic flow. It is also useful in emergency situations and reduces CO2 emissions, making the device an Internet of Things-based adaptive traffic management system (IoT).
5. A CNN-based Framework for Comparison of Contactless to Contact-based Fingerprints
The success of new contactless 2D fingerprint technologies, which allow more sanitary and deformation-free acquisition of fingerprint features, depends on accurate comparison of contactless 2D fingerprint photos with contact-based fingerprints. Convolutional neural networks (CNN) have demonstrated impressive ability in the recognition of biometrics. However, utilising CNN-based methods to match fingerprint pictures has hardly ever been attempted. In order to accurately compare contactless and contact-based fingerprint images, this research develops a CNN-based framework.
Utilizing fingerprint minutiae, the appropriate ridge map, and a specific region of the ridge map, our framework first trains a multi-Siamese CNN. Using a distance-aware loss function, this network generates deep fingerprint representations. For more precise cross comparison, deep fingerprint representations produced in such a multi-Siamese network are concatenated. On two publicly accessible datasets containing contactless 2D fingerprints and the corresponding contact-based fingerprints, the proposed approach for cross-fingerprint comparison is examined. Our tests, which are given in this study, consistently surpass those of the literature's contactless to contact-based fingerprint comparison methods as well as numerous widely used deep learning architectures.
6. Air Pollution Prediction System for Smart City using Data Mining Technique: A Survey
One of the biggest risks associated with environmental degradation is air pollution. since every living thing requires fresh, high-quality air at all times. Without such air, no living thing can survive. But our air is becoming more and more polluted as a result of cars, farming, manufacturing, industries, mining, and the combustion of fossil fuels. These actions released particulate matter, carbon monoxide, sulphur dioxide, nitrogen dioxide, and nitrogen dioxide into the air, which is bad for all living things.
Numerous health problems are brought on by the air we breathe every moment. Therefore, we require an effective system that can forecast such pollutants and contribute to a better environment. It prompts us to look for cutting-edge methods of forecasting air pollution. As a result, we are employing data mining to forecast air pollution in our smart city. We use the multivariate multistep Time Series data mining technique with the random forest algorithm in our model. In order to forecast air pollution, our method uses both historical and current data. This methodology makes decisions more trustworthy and precise for environmental protection agencies in smart cities while reducing complexity and improving efficacy and practicability.
7. Phishing Web Sites Features Classification Based on Extreme Learning Machine.
One of the most frequent and hazardous cybercrime assaults is phishing. The information utilised by people and companies to execute transactions is the target of these assaults. Phishing websites use a variety of indicators in both their text and information that is based on web browsers. The goal of this work is to classify 30 variables, including data from phishing websites, using Extreme Learning Machines (ELM) on a database at UC Irvine. ELM had the highest accuracy of 95.34% when compared to other machine learning techniques used for results evaluation, including Support Vector Machine (SVM) and Naive Bayes (NB).
8. Machine Learning for Web Vulnerability Detection: The Case of Cross-Site Request Forgery.
In this paper, we offer a way for utilising machine learning (ML) to identify vulnerabilities in online applications. Because of their diversity and the extensive use of bespoke programming techniques, web applications are particularly difficult to examine. Thus, ML is highly beneficial for web application security because it can use manually labelled data to incorporate automatic analytic tools with a human understanding of the semantics of online applications. In the creation of Mitch, the first machine learning (ML) solution for the identification of Cross-Site Request Forgery (CSRF) vulnerabilities, we applied our methodology. Mitch helped us find 35 brand-new CSRFs.
9. Image processing based Tracking and Counting Vehicles
In this research work, we explore the vehicle detection technique that can be used for traffic surveillance systems. This system works with the integration of CCTV cameras for detecting the cars. Initial step will always be car object detection. Haar Cascades are used for detection of car in the footage. Viola Jones Algorithm is used in training these cascade classifiers. We modify it to find unique objects in the video, by tracking each car in a selected region of interest. This is one of the fastest methods to correctly identify, track and count a car object with accuracy up to 78 percent
10. SMS Spam Detection using Machine Learning Approach.
Short Message Service (SMS) has become a multi-billion dollar industry in recent years as the use of mobile phones has expanded in popularity. The cost of messaging services has also decreased, which has led to an increase in the amount of spam that is delivered to mobile devices. Up to 30% of SMS messages in some regions of Asia were spam in 2012. The established email filtering algorithms may perform poorly in their classification due to the absence of true databases for SMS spams, short message lengths and features, and their informal language.
In this study, real SMS spam databases from the UCI Machine Learning repository are used. Following feature extraction and preprocessing, various machine learning algorithms are used to the databases. The optimal algorithm for text message spam filtering is then introduced after comparison of the outcomes. The best classifier in this study reduces the total error rate of the best model in the original research mentioning this dataset by more than half, according to the final simulation results using 10-fold cross validation.
11. Rainfall Prediction using Regression Model.
Weather forecasting is one of the applications where artificial intelligence is employed extensively. Predicting rainfall is one of the most commonly used research areas when it comes to weather forecasting because it causes significant property damage and loss of life. A heavy downpour can have a wide range of effects on society and daily life, from agriculture to emergency preparedness. The very complex mathematical tools utilised in earlier, widely used rainfall prediction models were insufficient to achieve a higher categorization rate.
In this study, we suggest brand-new, cutting-edge approaches for applying linear regression analysis to forecast monthly rainfall. Predictions of when it will rain are made using quantitative data about the atmosphere's state right now. Many machine learning algorithms can learn intricate mappings between inputs and outputs using only a small number of samples. Because the atmosphere is always changing, it is challenging to anticipate rainfall with any degree of accuracy. Utilizing the variation in conditions from previous years is necessary to forecast the likelihood of rainfall in the future. We've suggested using linear regressions with a variety of characteristics, including temperature, humidity, and wind.
Since the suggested model tends to anticipate rainfall based on the historical data for a certain geographic area, this prediction should show to be considerably more accurate. Comparing the model's performance to conventional rainfall forecast systems, it performs with more accuracy.
12 Human Activity Recognition using Open Cv & Python
Because it has so many important and cutting-edge applications, such as automated surveillance, automated vehicles, language interpretation, and human computer interfaces, human activity recognition has emerged as a field of study of great interest (HCI). A thorough and in-depth investigation has recently been conducted, and this field has advanced. The proposed system is designed to be a system that can be utilised for monitoring and surveillance purposes. This research shows a portion of a more recent technology that recognises human activity and interaction on human skeletal positions for video surveillance utilising a single stationary camera to record video data. Inefficiently and expensively, the standard surveillance camera system requires people to watch the cameras around-the-clock.
Consequently, this research report will offer the essential inspiration for properly identifying human behaviour in real-time (future work). This study employs visual processing techniques to identify common activities including walking, running, sitting, and standing.
13. Customer Segmentation using K-Means Algorithm
In the world we live in, enormous amounts of data are gathered every day. A crucial requirement is the analysis of such data. The business plan needs to be in line with the current circumstances in the modern period of innovation, when there is intense competition to be better than everyone. Since there are so many potential clients nowadays who are unsure of what to buy and what not to buy, company today is based on creative concepts. The businesses operating also lack the ability to identify the target potential clients.
This is where machine learning enters the picture; different algorithms are used to find the data's hidden patterns in order to make better decisions. Using the clustering technique, the customer segmentation process determines which consumer segment to target. The partitioning technique K-means is employed in this study's clustering algorithm to divide the customers into groups based on shared traits. The elbow approach is employed to identify the best clusters.
14. Crime analysis and prediction using data mining techniques
A methodical approach to detecting and evaluating patterns and trends in crime is crime analysis and prediction. Our algorithm can identify places with a high likelihood of crime and also depict crime-prone zones. due to the widespread adoption of computerised technologies. The major goal is to place more emphasis on crime factors than on the causes of crime incidents. We can obtain previously undiscovered important information from an unstructured data set by employing the notion of data mining. To create a data mining process that can aid in quicker crime solving, we can take a combined approach from computer science and criminal justice. The crime statistics can also be used to forecast criminals. The previous few years have seen a rise in criminal activity and illicit events.We provide a system that can analyse, spot, and forecast different crime probabilities in a specific area.
15. Raspberry pi based hand gesture recognition & voice conversion system for dumb people.
For dumb people to communicate with normal people, hand gestures are crucial. It is quite challenging for silent persons to communicate with non-mute people. because hand language is not taught to the general public. It might be extremely challenging to communicate during an emergency. To get around this, a prototype called Speaking Mouth for Dumb People is proposed. They will find it quite useful for communicating their ideas to others. The combination of hardware and software is the foundation of this system. Dumb people believe that every movement has a purpose. Therefore, software handles the part where human hand gestures are recognised and converted to human hearing voice. Voice conversion hence aids the dumb.
16. django framework for CV analysis and personality prediction using big 5 traits
A person's unique personality can be described as a collection of traits. A key area of psychology is the study of personality. There are several traditional methods for determining someone's personality, but they either require too much manual labour or can't be used in real time. This study attempts to address these issues by measuring the Big-Five personality traits using a series of questions. The user is asked to respond to a short list of questions, and based on their responses, a logistic regression model is used to forecast the user's personality.
17. Driver Drowsiness Monitoring System using Visual Behavior and Machine Learning
One of the leading factors in traffic accidents and fatalities is drunk driving. As a result, identifying driver weariness and related signs is a current research topic. The majority of traditional techniques are either based on vehicles, behaviours, or physiological principles. Some systems involve expensive sensors and data handling, while others are invasive and distract the driver. As a result, a real-time, low-cost system for detecting driver drowsiness is devised in this work.
The created system uses a webcam to record the video, and image processing methods are used to identify the driver's face in each frame. Facial landmarks on the detected face are identified, and tiredness is then determined using created adaptive thresholding based on the eye aspect ratio, mouth opening ratio, and nose length ratio values. There have also been offline implementations of machine learning algorithms. Support vector machine-based classification has been successful with a sensitivity of 95.58% and a specificity of 100%.
18. Assistive Device for Blind, Deaf and Dumb People using Raspberry-pi
It's challenging to provide solutions for those with visual, hearing, and vocal impairments with just one assistive technology. Many current studies, but not all of them, concentrate on finding solutions to one or more of the problems listed above. The work is focused on developing a novel method that helps the visually impaired by enabling them to hear what is portrayed as text. This is accomplished using a method that takes an image using a camera and converts it into speech signals.
The paper offers a method for those who have hearing impairments to see or read material that is in audio form using a speech to text conversion approach. It also offers a method for those who have vocal impairments to represent their voice using a text to voice conversion technique. To fit into a single distinct system, all three of these solutions were modified. With the help of the Raspberry Pi, all of these actions are coordinated.
The Tesseract OCR technique that provides image to text and text to speech is beneficial to persons who are blind (online character recognition). The procedure of an app that enables the deaf to understand what is being said can be aided by the deaf people. Vocally impaired people can use text messaging to communicate such that the other people can hear it through a speaker.
19. Classification of Cancerous Profiles using Machine Learning
For the treatment of cancer, there are numerous choices. The type of cancer, its severity (stage), and, most importantly, its genetic heterogeneity all have an impact on the suggested course of treatment for a given patient. The targeted medication therapies are likely to be ineffective or react differently in such a complicated setting. Understanding malignant profiles is necessary to evaluate anticancer medication response. These malignant profiles contain details that may help identify the underlying causes of the development of the disease. Therefore, it is necessary to study cancer data in order to forecast the best possible course of action. Analysis of these profiles can aid in predicting and identifying possible therapeutic targets. The primary goal of this work is to present a machine learning-based technique for classifying malignant profiles.
20. Voice based Email for Blind.
The programme, as the name implies, will be a web-based tool for visually impaired people that uses interactive voice response (IVR), allowing everyone to manage their email accounts with just their voice in addition to reading, sending, and performing all other necessary functions. The user will respond to voice orders from the system asking them to carry out specific actions. The fundamental advantage of this system is that the user just needs to answer by speaking and clicking a mouse; the keyboard is completely eliminated.
You're probably wondering how a blind person will be able to click the mouse in the proper location on the screen at this point. However, this system will only take action in response to a user's left- or right-click, regardless of where the pointer is on the screen. This gives users the flexibility to click wherever on the screen without regard to where the cursor is located.
21. Disease Prediction by Machine Learning Over Big Data From Healthcare Communities
Accurate medical data analysis aids in early disease identification, patient care, and community services as a result of the expansion of big data in the biomedical and healthcare sectors. However, when the quality of the medical data is lacking, the analysis's accuracy suffers. Additionally, distinct regional diseases in different places have their own features, which could make it harder to forecast when a disease would spread. In this study, we simplify machine learning algorithms for accurate chronic illness outbreak prediction in communities with high disease incidence. We test the updated prediction models using real-world hospital data that was gathered in central China between 2013 and 2015. We employ a latent component model to fill in the blanks in order to get around the problem of incomplete data. We conduct research on a localised, persistent cerebral infarction condition.
Using structured and unstructured hospital data, we suggest a new multimodal disease risk prediction algorithm based on convolutional neural networks (CNNs). To the best of our knowledge, no study has been done in the field of medical big data analytics that specifically addressed both data types. The prediction accuracy of our suggested method achieves 94.8% with a convergence speed that is faster than that of the CNN-based unimodal illness risk prediction algorithm, when compared to many conventional prediction algorithms.
22. Object Detection, convert object name to text and text to speech.
The development of computer vision systems has focused a lot on efficient and precise object detection. Since deep learning techniques have been developed, object detection has become much more accurate. In order to achieve high accuracy and real-time performance, the project seeks to utilise cutting-edge object identification techniques. The dependence on other computer vision techniques for support in many object detection systems, which results in slow and subpar performance, is a significant obstacle. In this project, we take an end-to-end method to solving the object detection problem that is entirely based on deep learning.
The most difficult publicly accessible dataset (PASCAL VOC), on which a yearly object detection challenge is run, is used to train the network. The resulting technology helps applications that need object detection because it is quick and precise. We introduce YOLO, a novel method of object detection. Classifiers have been used in the past to do object detection. As an alternative, we conceptualise object detection as a regression issue to spatially distinct bounding boxes and associated class probabilities. Bounding boxes and class probabilities are directly predicted by a single neural network from complete images in a single assessment. Since the entire detection pipeline consists of a single network, detection performance can be tuned from beginning to end.
23. Raspberry Pi based unauthorized car parking system by capturing the number plate of a vehicle using open cv technique thereby alarming the police & vehicle owner.
The goal of this project is to alert the traffic department when a car enters a no parking area. This is accomplished by using open cv to capture the licence plate of a vehicle. An alert is sent to the vehicle owner for parking in a no parking area, and the police receive the vehicle's licence plate in order to collect the fine.
24. Application of Data Mining Methods in Diabetes Prediction
By providing fresh insight on prevalent issues, data science methodologies have the potential to advance other scientific disciplines. Helping to create predictions based on medical data is one such task. Diabetes, sometimes known as diabetes mellitus or just diabetes, is a condition brought on by elevated blood glucose levels. The diagnosis of diabetes can be made using a variety of conventional techniques based on physical and chemical examinations. The strategies that are heavily reliant on data mining techniques can be used to forecast the risk of high blood pressure. In this study, we investigate the early diabetes prediction using five various data mining techniques, such as SVM and Logistic regression. The experiment's findings offer greater precision than previous methods.
25. An Approach to Maintain Attendance using Image Processing Techniques
These days, more and more research is being done on the development of novel strategies. Face recognition is one of the most popular applications of image processing. To take attendance, numerous cutting-edge devices have been developed. Biometric, thumbprints, access cards, and fingerprints are a few popular ones. The approach suggested in this study uses face detection and face recognition to track attendance using images. Face detection, labelling the observed faces, training a classifier using the labelled dataset, and face recognition are the four phases that make up the suggested method. Both positive and negative photographs were used to build the database.
To identify the faces in a classroom, the entire database has been split into training and testing sets and then processed by a classifier. The last step is to take attendance using a facial recognition technique, which involves providing an input image of a classroom so that faces and IDs may be identified in the image. To prevent missing frames due to rotational difficulties, a video captured for a minute is taken into account.
26. Sentiment Analysis of Top Colleges
Reviews and opinions play a significant role in how clients organise their visits nowadays. They also have an impact on how successful a brand, service, or product is. The same rules apply when choosing the best option out of the many options available. Stakeholders commonly participate in expressing their opinions regarding the strategy and advancement of online networking by using well-known social media, specifically Twitter. Twitter information at the same time can be quite educational; it presents a test for research given its enormous and disorganised nature.
The goal of this report's work is to do sentiment analysis on the best colleges in the nation using tweets from twitter, one of the social media platforms. The data collected by Twitter can be pre-processed using a variety of methods, including machine learning algorithms like KNN(K- closest neighbours) to identify the top college among IITS, NITS, and other institutions. The results were produced in R programming on the educational institutions using Naive Bayes and K-NN algorithms. The accuracy of the outcomes was then evaluated in comparison to KNN and Naive Bayes.
27. Detection of fake online reviews using semi-supervised and supervised learning
The business and commerce of today are greatly influenced by online reviews. The majority of how consumers choose which online things to buy is based on user reviews. As a result, opportunistic people or organisations try to slant product reviews in order to serve their own agendas. This study analyses the effectiveness of both strategies on a dataset of hotel reviews and offers various semi-supervised and supervised text mining models to identify false internet reviews.
28. Prediction and Diagnosis of Heart Disease Patients using Data Mining Techniqu
Our everyday routines are undergoing enormous change in the postmodern era, which has both positive and negative effects on our health. The prevalence of many different diseases has greatly grown as a result of these developments. Particularly, heart disease is now more prevalent than ever. The lives of people are in danger. Variations in blood pressure, blood sugar, pulse rate, and other factors can cause cardiovascular illnesses, which are conditions marked by blocked or restricted blood arteries. It may result in abrupt cardiac arrest, aneurysm, peripheral artery disease, heart attack, and other cardiovascular conditions. Various medical tests and consideration of family medical history can detect or diagnose numerous cardiac diseases.
This project's goal is to detect various cardiac conditions and take all reasonable preventative measures at an inexpensive cost. For the purpose of predicting heart disorders, we use the "Data mining" technique, in which attributes are input into the classification algorithms SVM, Random forest, KNN, and ANN. The preliminary readings and studies produced using this technology are used to determine if it is possible to diagnose cardiac illnesses accurately and treat them entirely at an early stage.
29. A Deep Learning Approach for Face Detection using YOLO
The term "deep learning" has been popular in recent years and is thought to represent a new phase in machine learning that teaches computers to identify patterns in vast amounts of data. It primarily discusses learning at many levels of representation, which aids in making understanding of data made up of text, music, and visuals. Convolutional neural networks are a form of deep learning that many businesses use to process the objects in a video sequence. The performance of Deep Convolution Neural Networks (CNNs) for object detection, picture classification, and semantic segmentation has been impressive. Combining classification and localisation is the definition of object detection.
One of the most difficult pattern recognition challenges is face detection. Face detection includes a number of face-related applications, such as face verification, facial recognition, face clustering, etc. For detection and recognition, effective training is required. The traditional method's precision in face detection did not produce a successful outcome. This research focuses on enhancing the deep learning model's face detection accuracy. The suggested approach is implemented using the well-known deep learning library YOLO (You only look once). The paper contrasts the efficiency of the efficient face detection method with the conventional method in terms of accuracy. The suggested model employs a deep learning method called the convolutional neural network for detecting face from videos
Our model is trained and tested using the FDDB dataset. On numerous performance parameters, a model is adjusted, and the best suitable values are taken into account. Additionally, the effectiveness of training and the performance of the model on two different GPUs are compared.This study analyses the effectiveness of both strategies on a dataset of hotel reviews and offers various semi-supervised and supervised text mining models to identify false internet reviews.
30. Analysis of Chronic Kidney Disease Dataset by Applying Machine Learning Methods
Chronic kidney illnesses are currently affecting a large number of people worldwide. Several risk factors, including food, the environment, and living conditions, cause many people to get diseases abruptly and without being aware of their condition. Chronic kidney disease diagnosis is typically intrusive, expensive, time-consuming, and frequently dangerous. Because of this, many illnesses are untreated until they are in advanced stages, especially in nations with few resources. As a result, early disease detection strategies are still crucial, especially in developing nations where diseases are frequently discovered in their advanced stages. Finding a solution for the aforementioned issues and overcoming disadvantages became a compelling reason to carry out this investigation. The results of employing clinical characteristics in this investigation wereThis study analyses the effectiveness of both strategies on a dataset of hotel reviews and offers various semi-supervised and supervised text mining models to identify false internet reviews.
31. Random Forest for Credit Card Fraud Detection.
Events involving credit card fraud happen frequently and end up costing a lot of money. Criminals are able to steal credit card information from other people by using technologies like Trojan or phishing. Because it can spot a fraud before a thief uses a stolen card to make a purchase, an efficient fraud detection technology is crucial. One approach is to fully utilise the historical transaction data, including both legitimate and fraudulent ones, to create features based on machine learning that distinguish between legitimate and fraudulent behaviour, and then use those features to determine whether a transaction is fraudulent or not. The behaviour aspects of normal and aberrant transactions are trained using two different types of random forests in this study.
We compare the two random forests, which differ in their underlying classifiers, and examine how well they perform at detecting credit fraud. Our experiments made use of data from a Chinese e-commerce firm.
32. Smart Voting System through Facial Recognition
Software that matches facial traits is referred to as facial recognition software. In the area of safe voting methods, we will research how various algorithms are implemented. For the voters in our proposed system, there were three stages of verification. The first level of verification is UID confirmation, the second level is voter card confirmation, and the third level of confirmation uses various facial recognition techniques. In this work, we will compare these algorithms using the following examples: Fisher Face, Eigenface, and SURF