USING MACHINE LEARNED CLASSIFIERS TO PREDICT FLIGHT DELAYS WITH ERROR CALCULATION
G. Mahesh1, P.Jagadeesh2, G.Navyu3, Sk. Imtiyaj Babu4, M. David Raju5
1Asst. Professor, Krishna Chaitanya Institute of Technology & Sciences , Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
*E-mail: maheshlucky13@gmail.com
Keywords - Logistic Regression, Decision Tree Regression, Bayesian Ridge, Random Forest Regression and Gradient Boosting Regression.
ABSTRACT
A significant issue in the aviation industry is flight delays. The expansion of the aviation industry during the past two decades has increased air traffic, which has delayed flights. Not only do flight delays cost money, but they also have a bad effect on the environment. Airlines that operate commercial flights suffer huge losses as a result of flight delays. In order to minimise or avoid flight delays and cancellations, they thus take all reasonable precautions. In this research, we forecast whether a certain flight's arrival will be delayed or not using machine learning models including Logistic Regression, Decision Tree Regression, Bayesian Ridge, Random Forest Regression, and Gradient Boosting Regression.
DEEP LEARNING APPROACH TO RESTORE OLD IMAGE BACK TO LIFE
Aishwarya Patil¹, Hritik Londhe¹, Shubham Kadave¹, Megha V. Gupta²
¹Research Scholar, ²Vice Principal, Department of Computer Engineering, New Horizon Institute of Technology and Management, University of Mumbai, Thane, India
Keywords - Variational Autoencoder, Latent space, Structured Degradation, Unstructured Degradation, Non-local block, Residual Block.
ABSTRACT
We propose using deep learning to recover historical old pictures that have suffered substantial deterioration. Unlike conventional traditional restoration tasks that can be handled using supervised learning, the deterioration in real world photos is complicated, an the domain gap between synthetic images and real-world old photos causes the network to fail to generalize. As a result, we present a unique novel triplet domain translation network by leveraging real photos along with massive synthetic image pairs. Specifically, we train two variational autoencoders (VAEs) to transform old photos and clean photos into two latent spaces, respectively. Synthetic paired data is used to learn the translation between these two latent spaces areas. Because the domain gap is closed in the compact latent space, this translation generalizes effectively to real photographs. Furthermore, to handle numerous degradations intermingled in a single old photo, we build a global branch with a partial nonlocal block targeting structured defects, like scratches and dust spots and a local branch targeting unstructured defects such as noises and blurriness. In the latent space, two branches are fused, resulting in increased potential to restore old photos that have various defects. In addition, we use another face refinement network to retrieve fine aspects of faces in the old photos, ultimately generating photos with improved perceptual quality. With comprehensive experiments, the proposed pipeline demonstrates superior performance over state-of-the-art methods as well as existing commercial tools in terms of visual quality for old photos restoration.
REALISTIC FACE IMAGE GENERATION BASED ON GAN
Purna Nandiboina¹, Akshata Salian¹, Shweta Akhadmal¹, Megha V. Gupta²
¹Research Scholar, ²Vice Principal, Department of Computer Engineering, New Horizon Institute of Technology and Management, University of Mumbai, Thane, India
Keywords - GAN, CNN, text to face, image generation, face synthesis, legal identity for all.
ABSTRACT
Text to face generation is a sub-domain of text to image synthesis. It has a huge impact on new research areas along with the wide range of applications in the public safety domain. Most of the work for text to face generation until now is based on the partially trained generative adversarial networks, in which the pre-trained text encoder has been used to extract the semantic features of the input sentence. Later, these semantic features have been utilized to train the image decoder. The proposed system will be a fully trained generative adversarial network to generate realistic and natural images. The system will train the text encoder as well as the image decoder at the same time to generate more accurate and efficient results. In addition to the proposed methodology, another contribution is to generate the dataset by the amalgamation of LFW and CelebA datasets. The dataset has also been labeled according to our defined classes. The system will create a model – a discriminator network and a generator network by eliminating the fully connected layer in the traditional network and applying batch normalization and deconvolution operations. The proposed work also presents the details of the similarity between the generated faces and the ground-truth input description sentences.
ETHEREUM BASED SYSTEM FOR COFFEE SUPPLY CHAIN MANAGEMENT
Apurva R. Suryawanshi¹, Siddhesh Tari¹, Nitin Pople¹, Sunil Bobade², Taruna Sharma³
¹Research Scholar, Department of Computer Engineering
²Head of Department Artificial Intelligence and Data Science New Horizon Institute of Technology and Management Thane, India
³Assistant Professor Department of Artificial Intelligence and Data Science New Horizon Institute of Technology and Management Thane, India
Keywords - Blockchain, Smart Contracts, BTC, Decentralized application.
ABSTRACT
Blockchain technology has grown in popularity in recent years, and numerous applications have emerged as a result of this technology. The cryptocurrency Bitcoin is a very well-known Blockchain application. The BTC network has not only found the solution against 51% attack, but it has also made it easier to confirm the legitimacy of transactional records without the use of centralized systems. As a result, any application that uses Blockchain technology as its foundational architecture ensures that the contents of its data are tamper-proof. This paper uses a decentralized Blockchain technology approach- so that all stakeholders in a coffee supply chain do not rely on third-party organizations or authoritative governing bodies for streamlining the whole process of coffee supply chain management. In this paper, we describe a decentralized application on the Ethereum Blockchain for solving, three main issues pertaining to coffee supply chain management: transparency, material traceability, and data accessibility.
DETECTION OF MALICIOUS SOCIAL BOTS USING URL FEATURES
Aarti B. Mastud¹, Svara R. Masurekar¹, Adarshsingh M. Mokashi¹, Aarti Abhyankar²
¹Research Scholar, ²Professor, Department of Computer Engineering, New Horizon Institute of Technology and Management, University of Mumbai, Thane, India
Keywords - Logistic Regression Model, malicious social bots, trust, legitimate, Non Legitimate.
ABSTRACT
Malicious social bots generate fake messages and automate their social relationships either by pretending like a follower or by creating multiple fake accounts with malicious activities. Moreover, malicious social bots post shortened malicious URLs in the message in order to redirect the requests of online social networking participants to some malicious servers. Hence, distinguishing malicious social bots from legitimate users is one of the most important tasks. To detect malicious social bots, extracting URL-based features (such as URL redirection, frequency of shared URLs, and spam content in URL) consumes less amount of time in comparison with social graph-based features (which rely on the social interactions of users). Furthermore, malicious social bots cannot easily manipulate URL redirection chains. In this article, a Logistic Regression algorithm is proposed by integrating a trust computation model with URL-based features for identifying trustworthy participants (users) in the social media network. This will help to determine the trustworthiness of each participant accurately. Experimentation has been performed on two social media network data sets, and the results illustrate that the proposed algorithm achieves improvement in precision, recall, and accuracy compared with existing approaches for MSBD.
INTERACTIVE HOLOGRAM USING FINGER GESTURES
Yadnesh Chowkekar1, Athava Deshmukh2, Sakshi Polekar3, Yogita chavan4
Department of Computer Engineering, New Horizon Institute of Technology and Management, Thane, India
Keywords - hologram, gestures, quadrangular pyramid, 3D projection, LCD display.
ABSTRACT
In this work, we show how to use the quadrangular pyramid hologram and parabolic system (for 3D holographic object reconstruction) and the flick gesture module (as a finger action) to make a 3D holographic projection. Not only can this device reconstruct and project a 3D hologram item in mid-air, but it also allows users to interact with it by providing certain finger gestures. The three basic processes are the reconstruction of a 3D object, projection of a 3D hologram object in mid-air, and interactive manipulation of a 3D hologram object. A quadrangular pyramid hologram with an LCD display is used for the first step, whereas the other hardware components help in processing the gestures and applying the changes in the hologram.
DIABETES PREDICTION USING MACHINE LEARNING
Anuja Mhase, Sankalp Suryawanshi, Tushar Vaishnav, Prof. Laxmikant Malphedwar
Dept of Computer Engineering, DY Patil School of Engineering Academy, Savitribai Phule Pune University, Pune, Maharashtra, India.
Keywords - Machine Learning, Diabetes, Decision tree, K nearest neighbour, Logistic Regression, Support vector Machine, Accuracy.
ABSTRACT
Diabetes is a chronic disease with the potential to cause a worldwide health care crisis. According to International Diabetes Federation 382 million people are living with diabetes across the whole world. By 2035, this will be doubled as 592 million. Diabetes is a disease caused due to the increase level of blood glucose. This high blood glucose produces the symptoms of frequent urination, increased thirst, and increased hunger. Diabetes is a one of the leading cause of blindness, kidney failure, amputations, heart failure and stroke. When we eat, our body turns food into sugars, or glucose. At that point, our pancreas is supposed to release insulin. Insulin serves as a key to open our cells, to allow the glucose to enter and allow us to use the glucose for energy. But with diabetes, this system does not work. Type 1 and type 2 diabetes are the most common forms of the disease, but there are also other kinds, such as gestational diabetes, which occurs during pregnancy, as well as other forms. Machine learning is an emerging scientific field in data science dealing with the ways in which machines learn from experience. The aim of this project is to develop a system which can perform early prediction of diabetes for a patient with a higher accuracy by combining the results of different machine learning techniques. The algorithms like K nearest neighbour, Logistic Regression, Random Forest, Support vector machine and Decision tree are used. The accuracy of the model using each of the algorithms is calculated. Then the one with a good accuracy is taken as the model for predicting the diabetes.
OPENCV BASED IMAGE PROCESSED ATTENDANCE SYSTEM
Sushant Borkar, Shrutik Ghavate, Sudhanshu Pawar, Prof. Vinod Bharat
Dept of Computer Engineering, DY Patil School of Engineering Academy, Savitribai Phule Pune University, Pune, Maharashtra, India.
Keywords - OpenCV, Face Recognition, AI, Data Set.
ABSTRACT
Face recognition is among the most productive image processing applications and has a pivotal role in the technical field. Recognition of the human face is an active issue for authentication purposes specifically in the context of attendance of students. Attendance system using face recognition is a procedure of recognizing students by using face biostatistics based on the high definition monitoring and other computer technologies. The development of this system is aimed to accomplish digitization of the traditional system of taking attendance by calling names and maintaining pen-paper records. Present strategies for taking attendance are tedious and time-consuming. Attendance records can be easily manipulated by manual recording. The traditional process of making attendance and present biometric systems are vulnerable to proxies. This project is therefore proposed to tackle all these problems. The proposed system makes the use of Open CV and face recognition libraries. After face recognition attendance reports will be generated and stored in excel format. The system is tested under various conditions like illumination, head movements, the variation of distance between the student and cameras. After vigorous testing overall complexity and accuracy are calculated. The Proposed system proved to be an efficient and robust device for taking attendance in a classroom without any time consumption and manual work. The system developed is cost-efficient and needs less installation.
DRIVER DISTRACTION DETECTION USING PYTHON & OPENCV
Mrs. Kavita Sawant, Pratiksha Meshram, Seema Mahadev Jadhav, Samruddhi Renakale, Gayatri Nilesh Yadav
Department of Computer Engineering, Bharati Vidyapeeth College of Engineering for Women, Pune, Maharashtra.
Keywords - Drowsiness Detection, Eye-blink, ML, python, etc.
ABSTRACT
Driver fatigue is one of the major causes of accidents in the world. Detecting the drowsiness of the driver is one of the surest ways of measuring driver fatigue. In this project we aim to develop a prototype drowsiness detection system. This system works by monitoring the eyes of the driver and sounding an alarm when he/she is drowsy. The system so designed is a non intrusive real-time monitoring system. The priority is on improving the safety of the driver without being obtrusive. In this project the eye blink of the driver is detected. If the drivers eyes remain closed for more than a certain period of time, the driver is said to be drowsy and an alarm is sounded. The programming for this is done in OpenCV using the Haarcascade library for the detection of facial features.
DESIGNING DISEASE PREDICTION MODEL USING MACHINE LEARNING APPROACH
Prof. N.R. Jain, Harshal Chaudhari, Lalit Jadhav and Vaishnavi Patole
Department of Information Technology, PDEA’s COE, Manjari, Hadapsar Maharashtra, Pune 412307
Keywords - CNN, KNN and Machine learning, Disease Prediction.
ABSTRACT
Nowadays, human creatures confront different afflictions since of the natural circumstance and their dwelling conduct. So the expectation of affliction at an in progressed degree will got to be a crucial assignment. But the right forecast on the thought of signs will ended up as well difficult for specialists. The exact forecast of ailment is the greatest difficult errand. To triumph over this inconvenience, data mining performs a crucial work to are anticipating the sickness. Therapeutic mechanical know-how contains a enormous amount of data increment concurring to year. Due to the increased amount of data increment withinside the clinical and healthcare range the proper assessment of clinical data has been cashing in on early influenced individual care. With the help of ailment data, data mining uncovers covered up test truths in a expansive amount of clinical data. We proposed in vogue ailment expectation fundamentally based completely at the signs of the influenced individual. For the affliction expectation, we utilize K-Nearest Neighbor (KNN) and Convolutional neural organize (CNN) contraption examining set of rules for the proper forecast of ailment. For affliction forecast required affliction signs dataset. In this in vogue ailment forecast, the dwelling conduct of somebody and checkup facts do not disregard for the right expectation. The precision of elegant ailment forecast through way of implies of the utilize of CNN is 84.5% that's additional than the KNN set of rules. And the time and the memory prerequisite also are additional in KNN than CNN. After in vogue ailment forecast, this device is prepared t provide the peril related to a elegant affliction that's a diminish peril of in vogue ailment or higher.
SECURE E WALLET ARCHITECTURE USING BCT
Siddhant Tyagi, Rutuja Pawar, Mayuri Uttarwar, Riddhi Padalkar, Prof. Mrs. Pradnya Kasture
Dept of Computer Engineering, RMDSSOE, Savitribai Phule Pune University, Pune, Maharashtra, India
Keywords - digital economy, cashless, SHA256, AES, digital India, java, jsp, servlet, etc.
ABSTRACT
A cashless society is one in which financial transactions are conducted without the use of real money, such as banknotes or coins, but rather through the exchange of digital information (usually an electronic representation of money) between the parties involved. Since the birth of human civilization, cashless societies have existed, based on barter and other types of exchange, and cashless transactions are now possible with the usage of digital currencies like bitcoin. However, this article focuses on the term "cashless society" in the sense of a transition to a society in which cash is replaced by its digital equivalent—in other words, legal tender (money) exists, is recorded, and is only transferred in electronic digital form—as well as the implications of such a society. Such a concept has gotten a lot of attention, especially because digital methods of recording, managing, and exchanging money are becoming increasingly popular in commerce, investment, and everyday life in many parts of the world, and transactions that would have been done with cash in the past are now frequently done electronically. Non-electronic payment methods are now subject to transaction and transaction amount limits in some countries. We'll look at how block chain technology can be applied to the digital economy to help India become more digital in this post.
IMPLEMENTATION OF SECURE ENCRYPTION FRAMEWORK USING AES
Akshay Bhagwan Bahadure, Rahul Dilpak, Darshan Satpute, Prof. Rama Barwal
Dept of Computer Engineering, Savitribai Phule Pune University, Pune, Maharastra, India.
Keywords - Security, framework AES, java, jsp, cloud, etc.
ABSTRACT
In this project we propose the data on cloud computing is encrypted due to security concern or the factor of third party digging into it. As the consequent to this, the search over encrypted data becomes a complex task. The traditional approaches like searching in plain ext cannot be apply over encrypted data. So the searchable encryption techniques are being used. In searchable encryption techniques the order of relevance must be consider as the concern because when it is large amount of data it becomes complex as relevant documents are more in number. We have discussed the Re-encryption technique. The expected result is to be that cloud server cannot penetrate in actual user data and provide the search on encrypted data will be performed and results will appear in order of relevance score. Even though with good security of Re-encryption the cloud can get the information of the plain text if differential attack occurred on the cipher text by calculating the differences between the cipher text.
APPLICATION OF MD5 FOR DE DUPLICATION IN CLOUD ENVIRONMENT
Prof. Smita Gumaste, Vipul Kargaonkar, Indranil oza, Shubham Kosaiker, Yash Khade
Dept of Computer Engineering, JSCOE, Savitribai Phule Pune University, Pune, Maharashtra, India.
Keywords - Data deduplication, cloud, AES, MD5, Java, JSP & Servlet, etc.
ABSTRACT
Computing resources are given as a utility on demand to consumers over the Internet, and cloud computing plays a significant role in the commercial domain today. Cloud storage is one of the services offered by cloud computing that has grown in popularity. Customers benefit the most from cloud storage since they can cut their expenditures on purchasing and maintaining storage equipment by simply paying for the amount of storage they need, which can be scaled up and down on demand. With cloud computing's expanding data size, a reduction in data quantities could assist providers in lowering the costs of running huge storage systems and conserving energy. As a result, data deduplication techniques have been implemented in cloud storage to improve storage efficiency. Because of the dynamic nature of data in cloud storage, data utilization in the cloud fluctuates over time. For example, some data chunks may be accessed often one time but not the next. Some datasets may be viewed or updated frequently by several users at the same time, while others may require a high amount of redundancy for stability. As a result, it's critical that cloud storage provide this dynamic functionality. Current techniques, on the other hand, are primarily focused on a static scheme, which limits their full applicability in the dynamic nature of data in cloud storage. We propose a dynamic deduplication strategy for cloud storage in this research, with the goal of increasing storage economy while maintaining redundancy for fault tolerance.
SIGNATURE BASED AUTHENTIATION USING DECISION TREE AND SUPPORT VECTOR MACHINE
Yerininti Venkata Narayana¹, Chintakrindi Harika², Katra Keerthana², ArlaTirumala²
¹Asst.Professor, Information Technology Department, Vasireddy Venkatadri Institute of Technology, Guntur, Andhra Pradesh, India
²B. Tech Students, Department of Information Technology, Vasireddy Venkatadri Institute of Technology, Guntur, Andhra Pradesh, India
Keywords - Machine Learning, Signature verification, Support Vector Machine, K-means, Decision Tree.
ABSTRACT
Every person has his/her own unique signature that is used mainly for the purposes of personal identification and verification of important documents or legal transactions. There are two kinds of signature verification: static and dynamic. Static(off-line) verification is the process of verifying an electronic or document signature after it has been made, while dynamic(on-line) verification takes place as a person creates his/her signature on a digital tablet or a similar device. Offline signature verification is not efficient and slow for a large number of documents. To overcome the drawbacks of offline signature verification, we have seen a growth in online biometric personal verification such as fingerprints, eye scan etc. In this project we created MACHINE LEARNING models like Support Vector Machine and K means using python for offline signature and after training and validating, the accuracy of testing.
HANDWRITTEN CHARACTER RECOGNITION USING CNN
Telaprolu Gayathri1, Ponnaganti Swarna2, Vaddimukkala Srisai Charitha3, Y. Venkata Narayana4
1,2,3B. Tech Students, IT, Vasireddy Venkatadri Institute of Technology, Guntur, Andhra Pradesh, India
4Asst. Professor, IT, Vasireddy Venkatadri Institute of Technology, Guntur, Andhra Pradesh, India
Keywords - Handwritten Character Recognition, Convolutional Neural Network, Feature extraction, TensorFlow.
ABSTRACT
Machine Learning comes under the field of computer science and technology that provides computers the ability to learn things without being programmed explicitly. Machine learning is used where it is not feasible to program and design algorithms with better performance; like a number of applications include filtering of emails, detection of malicious insiders working towards a data breach, optical character recognition (OCR) learning to rank, face recognition and computer vision. It became important to recognize handwriting as we start moving towards automated world. Deep learning is showing unbelievable results in the domain of visual and speech recognition. But still we lack in matching the accuracy of human vision and advancements will be continuing until we cross that limit. TensorFlow is one of the Google's open source machine learning as well as deep learning framework, which is convenient to build the standard deep learning model. Convolutional neural network is a unique model of deep learning; the advantage of using CNN is its powerful feature extraction capabilities of convolutional blocks. Based on the TensorFlow platform, a convolutional neural network model with five-convolution layers has been created. The proposed system has been trained on samples of large collection of IAM database images and tested on sample images from user defines data set and in this experiment we found the highest recognition results.
NETWORK INTRUSION DETECTION SYSTEM USING MACHINE LEARNING
Prof. Vrushali V. Kondhalkar, Swapnil Shende, Abhishek Patwari, Pranav Ovhal
Dept. of Computer Engineering, JSCOE, Pune, Maharashtra, India
Keywords - NIDS, Machine Learning, KDD Dataset.
ABSTRACT
“Network Intrusion Detection System Based on Machine Learning Algorithms” is a software that monitors network of computers for malicious activities that are aimed at stealing sensitive confidential information or corrupting /hacking network protocols. Techniques used in Today’s NIDS are not able to deal with the Dynamic Complex types of security Cyber Attacks on Computer Networks. Performance of an intrusion Detection is mainly depending on accuracy. Accuracy for Intrusion detection must able to decrease false alarms and to increase the detection rate of alarms. To improve the performance, different techniques have been used in recent works. Analyzing huge network traffic data is the main work of intrusion detection system. A well-organized classification methodology is required to overcome this problem. NSL-KDD knowledge discovery data set is used, their accuracy and misclassification rate get calculated.
EXOPLANETS DETECTION USING MACHINE LEARNING
Josiah Joseph, Dhiraj Thakare and Arathi Kamble
Department of Computer Engineering, New Horizon Institute of Technology and Management Thane, India
Keywords - Exoplanets, Kepler, Random Forest, Transportation, Light Therapy & Solar Flux.
ABSTRACT
Kepler's planetary candidate catalog, the first based catalog of all, processed similarly by Kepler's data set. This is the first automated catalog, which uses robots to test robots to accurately measure all the periodic signals received by the Q1 – Q17 Data Release 24 Kepler pipeline. Although we prioritize the same tests over the exact accuracy of each item, we find that our robotic tests are completely the same, and in most cases, the human testing procedures used by previous catalogs. This catalog is the first to use an artificial transport injection to evaluate the effectiveness of our testing procedures and to measure potential bias, which is important in accurately calculating planetary occurrence rates. In terms of the Kepler Object of Interest catalog, we select 1478 new KOIs, 402 of which are classified as PCs. Also, 237 KOIs ranked as FP points in Kepler's previous catalogs changed to PC and 118 PCs changed to FP status. This brings the total number of known KOIs to 8826 and PCs to 4696. Comparing the Q1 – Q17 DR24 KOI catalog with previous KOI catalogs, as well as the corresponding Kepler catalogs, we find a positive agreement between them. We highlight both new PCs that can be rocky and possibly in the host space, many of which revolve around the solar system. This work represents the milestone in the history of the solar system. The transport system, one of the few methods used to detect exoplanets, detects a periodic decrease (or immersion) in starlight that reflects the movement of the planet in front of the star as seen in the observer's point of view. This approach far surpasses all other exoplanet acquisitions in terms of the number of planets discovered and, as a result, represents our best choice for discovering planetary planets with our current technological capabilities. In addition, this method allows us to determine the composition of the planet's atmosphere by examining the spectrum of stars with a high resolution of light from a star passing through the upper atmosphere of the planet. A summary of the concept is provided along with some of the results presented for its use.
DETECTION OF CYBERBULLYING ON TWITTER USING MACHINE LEARNING
D. Sharath Chander¹ and Dr. B. Indira Reddy²
¹MCA Student, Dept. of MCA, Chaitanya Bharathi Institute of Technology (A),
²Assistant Professor, Dept. of MCA, Chaitanya Bharathi Institute of Technology (A), Gandipet, Hyderabad – 500 075, India
Keywords - Cyberbullying, Hate speech, Machine learning, Feature extraction, Twitter.
ABSTRACT
As a symptom of progressively well-known web-based entertainment, cyberbullying has arisen as a significant issue burdening youngsters, youths, and youthful grown-ups. AI strategies make programmed recognition of tormenting messages in virtual entertainment conceivable, and this could assist with building a solid and safe online entertainment climate. Cyberbullying is a significant issue experienced on the web that influences teens and furthermore grown-ups. It has prompted mishappening like self destruction and wretchedness. Guidelines of content via virtual entertainment stages have turned into a developing need. The going with survey uses data from an unmistakable sort of cyberbullying, hate speech tweets from Twitter to foster a model considering the recognizable proof of Cyberbullying utilizing Natural Language Processing and Machine learning.
DATA ANALYSIS BY WEB SCRAPING
S.Soumya¹, D.Shravani¹, V.Shivani¹, L.Swathi¹, G.Ranjith kumar², Dr.R.Jegadeesan³
¹Studentsof IV B.Tech, Department of CSE, Jyothishmathi Institute of Technology &
Science, karimnagr(TS)
²Assistant Professor, Department of CSE, Jyothishmathi Institute of Technology &
Science, karimnagr(TS)
³Associate Professor, Head of CSE department , Jyothishmathi Institute of Technology &
Science, karimnagr(TS)
Keywords - Data Analysis, Web Scrapping, data scrape.
ABSTRACT
The standard information investigation is built on the root and impact relationship, shaped an example minuscule examination, subjective and quantitative examination, the rationality approach of creating extrapolation examination. The Web Scraper’s conniving ethics and procedures are juxtaposed, it explains about the working of how the scraper is premeditated. The technique of it is allocated into three fragments: the web scraper draws the desired links from web, and then the data is extracted to get the data from the source links and finally stowing that data into a csv file. The Python language is implemented for the carrying out. By doing so, linking all these with the moral knowledge of libraries and working know-how, we can have an adequate Scraper in our hand to produce the desired result. Due to an enormous community and library resources for Python and the exquisiteness of coding chic of python language, it is most appropriate one for Scraping desired data from the desired website.
FACE RECOGNITION ATTENDANCE SYSTEM USING RASPBERRY PI
Namrata Bhosale, Pratap Shinde and Vandana Bavkar
Department of Electronics and Telecommunication, Department of Electronics and
Telecommunication, TSSM's Bhivarabai Sawant College of engineering and Research, Narhe, Pune.
Keywords - RFID, Face recognition, detection, accuracy, OpenCV, Raspberry pi.
ABSTRACT
Record your presence using identification strategies such as RFID, iris recognition, and fingerprint recognition. Of all these personal identification strategies, including facial recognition, it's the most natural, fast, and highly efficient, difficult to implement, but a continuous observation to overcome. There are multiple applications in the attendance management system and security system. This paper implements a system that uses facial recognition and recognition techniques to detect the presence of students, industrial workers, etc. in a lecture. The participation period is set and the database is automatically uploaded to the web server over the internet connection. This process takes place without human intervention. A Raspberry Pi with the OpenCV library is installed on the system and a Raspberry Pi camera module is attached for face detection and recognition. The data is stored on a memory card connected to the Raspberry Pi and can be accessed via the internet. The results show that continuous observation improves accuracy and maximizes output.
ARTIFICIAL INTELLIGENCE BASED MODEL FOR HANDWRITTEN RECOGNITION
Dr. Sk. Althaf Hussian Basha1, S. Jayaram Sarma2, M.Anusha3, T.Jyothi Rao4, B. Prasnna Kumar5
1Professor and Head, Department of CSE, KITS, Markapur, Andhra Pradesh, India
2Research Scholar, Dayananda Sagar University, Bangalore, India.
3,4,5Student, Department of CSE, KITS, Markapur, Andhra Pradesh, India
Keywords - Handwritten, Recognition, SVM algorithm, Artificial Intelligence, Multi scripts.
ABSTRACT
An all-encompassing method for improved handwriting recognition is put forward in this project. By speeding up the process of turning documents into letters, handwriting recognition algorithms can lessen the effort. The thesis uses the multi-script handwritten font family, which includes the Latin, MNIST handwritten alphabet series on prescription, and the Bangla font. Genetic algorithms and artificial intelligence tools were used in the creation and development of this stage. This method was created to deliver correct results in the recognition of the Bangla set, which is 54.05 percent, Latin, which is 98.58 percent, and MNIST, which is 98.58 percent.
ADVANCED IOT BASED REAL-TIME EARTHQUAKE ETECTOR
Pratap Shinde, Vandana Bavkar and Abhishek Kumar
Department of Electronics and telecommunications- TSSM’s Bhivarabai Sawant College of engineering and Research, Narhe, Pune.
Keywords - THINGSPEAK, IoT Platform,sensor, microcontroller, rectifier.
ABSTRACT
The early seismic warning system detects the first tremor of a larger earthquake and triggers the warning system before the most severe tremor. The proposed global warning system uses a network of digital seismographs distributed throughout the state to alert densely populated areas up to 1 minute in advance (depending on the location of the epicenter). Warnings will give businesses, residents, and authorities time to prepare. The purpose of the study is to focus on sensor data to determine if an earthquake will occur. Finally, experimental results are provided showing that the system supports the expected performance of the sensor data. A possible extension of this approach could be the implementation of a wireless sensor network using Thing-peak for data acquisition.
A SECURE AUTHENTICATED KEY MANAGEMENT PROTOCOL FOR CLOUD COMPUTING ENVIRONMENTS-DESIGN
M.GnanaVardhan1, P.Hari Chandana2, P.Meghana3, N.V.Rajashree4, Sd.Ashraf5
1Associate Professor, 2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
Keywords - MAKA, Schnorr, Multi-server.
ABSTRACT
Due to the dependability and performance of cloud computing technologies developing, many services have migrated to the cloud platform. Due to its ability to simplify service access and protect communication privacy on public networks, three-factor Mutual Authentication and Key Agreement (MAKA) protocols for multi-server architectures are attracting a lot of attention. But a lot of the three-factor MAKA protocols that are now in use either lack a formal security proof, making them open to multiple attacks, or have high computation and transmission costs. Furthermore, most three-factor MAKA protocols don't have a dynamic revocation mechanism, making it challenging for dishonest users to have their access rapidly removed. To address these issues, we provide a tried-and-true dynamic, adjustable, three-factor MAKA protocol. This protocol provides a simple random oracle verification and manages users dynamically using Schnorr signatures. According to a security assessment, our protocol can handle a variety of needs when there are several servers involved. Performance analysis demonstrates that the recommended approach is perfect for smart devices with constrained computing power. The effectiveness of the protocol is seen throughout the whole simulation run.
MACHINE LEARNING IN SMART PRODUCTION SYSTEMS WITH SCALABLE ANALYTICS PLATFORM
Mr. Sk. Alimoon1, K.Karuna2, V.Prakash3, P. Lakshmi Latha4, S. Bala chandrudu5
1Associate Professor, 2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
Keywords - Big Data, Machine Learning, Industry 4.0, Industrial Automation.
ABSTRACT
The manufacturing sector faces significant challenges in meeting the consumer's shifting needs. Therefore, manufacturing procedures must be efficient, rarely interrupted, and resource-efficient. To do this, massive amounts of data generated by industrial machines must be managed and assessed using contemporary technology. Because the big data era in the manufacturing sector is still in its infancy, there is a need for a reference architecture that incorporates big data and machine learning technologies and is compliant with Industrie 4.0 standards and specifications. In this article, the requirements for developing a scalable analytics platform for industrial data are established using the Industries 4.0 standards and literature. Based on these requirements, a reference large data architecture for business machine learning applications is proposed, and it is compared to similar publications. Finally, the parallel processing of an industrial PCA model in the Lab Big Data at the Smart Factory OWL has been used to evaluate the performance and scalability of the suggested architecture. The results show that the proposed structure is linearly scaleable, adaptable to machine learning use cases, and would improve the industrial automation processes of the production systems.
CLOUD COMPUTING ENHANCE DEPLOYMENT MODEL: AN OVERVIEW
Asst. Prof. Puja Vishwakarma
Department of computer science and application SDIMT, Haridwar
Keywords - Cloud Computing, Deployment model and services, public cloud, private cloud, hybrid cloud, community cloud, hierarchical cloud and atmos cloud.
ABSTRACT
Cloud Computing now a day’s become very important for any organization. As cloud computing provide such environment that any user any organization can use Services of cloud computing or resources such as storage, database, application server or e-mail services. these type of resources or services can be accessed by cloud computing deployment models like public cloud, private cloud, community cloud, hybrid cloud, multiple cloud, hierarchical cloud and atmos cloud. This paper describes the different types of cloud services and deployment models. It also describes comparative study using charts i.e. how many organization prefer which type of deployment model. Cloud computing is one of the hottest trends. Most technological solutions are now on cloud. Due to its exceptional benefits, it has magnetized the IT leaders and entrepreneurs at all levels.
BIOMETRIC-BASED SECURE ACCESS MECHANISM FOR CLOUD SERVICES: DESIGNING A SECURE AND EFFICIENT MECHANISM
P. Pullaiah1, T.Haritha2, K.Jayanth3, D.Akhilandeswari4, SK. Abdul Mohasin5
1Asst. Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Bio-metric, Cloud server, Real-or-Random, informal security analysis.
ABSTRACT
In our data-driven culture, there is an exponential growth in the need for distant data storage and compute services, necessitating the requirement for safe access to such data and services. In order to enable safe access to a distant (cloud) server, we build a new biometric-based authentication system in this article. In the suggested method, we treat a user's biometric information as a secure credential. From the user's biometric information, we then create a unique identity that is utilized to produce the user's private key. Additionally, we offer a practical method for creating a session key for secure message transmission between two conversing participants utilizing two biometric templates. In other words, the user's private key does not need to be stored anywhere, and the session key is produced secretly. The proposed approach can withstand several well-known attacks against (passive/active) adversaries, according to a thorough formal security analysis using the Real-Or-Random (ROR) model, an informal (non-mathematical) security analysis, and formal security verification using the widely-accepted Automated Validation of Internet Security Protocols and Applications (AVISPA) tool. Finally, thorough tests and a comparison show how effective and practical the suggested strategy is.
MACHINE LEARNING TECHNIQUES FOR CROP YIELD PREDICTION
E. Rajesh1, M. Asha jyothi2, M. Mohan Manikanta3, K. Mounika4, D.Lakshmi Devi5
1Asst. Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Agriculture, Artificial Neural Network, Convolution Neural Network, Crop Yield Prediction, Machine Learning Method.
ABSTRACT
Agriculture is a sector that has a significant impact on the economy of our nation. Agriculture is the key factor in the development of civilization. India is a predominantly agricultural nation with a crop-based economy. As a result, we may argue that our country's economy can be supported by agriculture. Every crop must be carefully chosen while developing an agricultural project. The choice of crops will be influenced by a variety of factors, including market price, production rate, and government policies. To enhance improvements in our Indian economy, the agriculture sector has to undergo several modifications. Using machine learning techniques that are simple to use in the farming industry, we can enhance agriculture. Along with all the improvements in the tools and technology used in farming, precise and helpful knowledge about many topics is also crucial. The goal of this study is to put the crop selection approach into practice so that it may help farmers and agriculturalists solve a variety of issues. As a result, the Indian economy is enhanced by the highest possible agricultural yield rate.
PREDICTING AGRICULTURAL PRODUCE PRICES WITH CONVOLUTION NEURAL NETWORKS: IMPROVING THE LIVES OF INDEBTED FARMERS WITH DEEP LEARNING
J. Mahalakshmi1, Y. Manisa2, S. Hemanth Kumar3, N.Keerthana4, P.Jhansi Lakshmi5
1Associate Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Neural Networks, PECAD, Deep Learnng, RMSE, Machine learning.
ABSTRACT
Farmer suicides have emerged as a pressing societal issue that governments all around the world are working diligently to address. Most farmers commit themselves because they can't sell their The extensive uncertainty/fluctuation in the market makes it difficult to produce at targeted profit levels, which produce costs as a result of changing market circumstances. This aims to stop farmer suicides, In order to address the issue of product price unpredictability, this study makes a first step. Introducing PECAD, a deep learning method for precise produce price forecasting according to historical price and volume trends. Despite the fact that earlier research has introduced machine learning algorithms for produce price prediction, these algorithms have two drawbacks: I they do not explicitly take into account the spatio-temporal dependence of future prices on past data; as a result, (ii) they rely on classical ML prediction models, which frequently exhibit poor performance when applied to spatio-temporal datasets. Through three key contributions, PECAD tackles these limitations: We collect actual daily prices and (produced) volume data for various crops over a period of 11 years from a website run by the Indian government; (ii) pre-process this raw information using cutting-edge imputation techniques to adjust for missing data entries; and(iii) PECAD suggests a brand-new broad and deep neural network architecture made up of two distinct convolutional neural network models that were trained on price and volume data, respectively. Our simulation findings demonstrate that PECAD surpasses current state-of the-art baseline approaches by obtaining noticeably lower root mean squared error (RMSE) – PECAD produces a coefficient of variance that is around 25% lower than state-of-the-art baselines. To reduce farmer suicides in the Indian state of Jharkhand, we collaborate with a non-profit organization, and PECAD is now being evaluated by for possible implementation.
MACHINE LEARNING BASED TRAFFIC PREDICTION FOR INTELLIGENT TRANSPORTATION SYSTEMS
K. Raj Kiran1, J.Parimala2, M.Parvathi3, N.Jahnavi Devi4, G.Akhil Venkata Rushi5
1Asst. Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Traffic Environment, Deep Learning, Machine Learning, Genetic Algorithms, Soft Computing, Big Data, Image Processing.
ABSTRACT
This paper aims to provide a tool for accurate and timely traffic flow data predictions. Everything that could affect how much traffic is moving along a road is referred to as the "traffic environment," including traffic lights, collisions, demonstrations, and even road works that might cause a delay. If a driver or rider has prior knowledge that is near to accurate about all the aforementioned factors and many more actual situations that might affect traffic, they can make an informed decision. Additionally, it promotes the development of autonomous vehicles. Recent decades have seen a significant increase in traffic data, and big data concepts for transportation are becoming more prevalent.Although several traffic prediction models are used in the current methods for estimating traffic flow, they are still insufficient to deal with real-world scenarios. As a result of this, we started using the traffic data and models to work on the traffic flow forecast problem. It is impossible to accurately predict the traffic flow since the transportation system has access to an insane quantity of data. With the use of machine learning, genetic, soft computing, and deep learning approaches, we aimed to considerably reduce the complexity of the analysis of large data for the transportation system in this work. Additionally, traffic signs are recognised using image processing techniques, which eventually help with the correct training of autonomous vehicles.
CLOUD BASED - DATA STORAGE AND SHARING WITH DUAL ACCESS CONTROL
SK.Yasmin Sulthana1, B.Kavya Sri2, B.Manasa3, D.Namrutha4, S.Chenna Reddy5
1Asst. Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India *Email: yasminbsc6@gmail.com
Keywords - Cloud Based Data Sharing, Access Control, Cloud Storage Server, Index SGX, Attribute-Based Encryption
ABSTRACT
Due to its effective and affordable administration, cloud-based data storage has recently attracted growing interest from academia and business. Since services are delivered via an open network, it is critical for service providers to adopt secure data storage and sharing mechanisms to protect user privacy and the confidentiality of data. The most popular technique for preventing the compromise of sensitive data is encryption. The actual necessity for data management, however, cannot be fully met by merely encrypting data (for instance, using AES). Additionally, a strong access control over download requests must be taken into account to prevent Economic Denial of Sustainability (EDoS) assaults from being performed to prevent users from using the service.In the context of cloud-based storage, we explore dual access control in this article in the sense that we create a control mechanism over both data access and download requests without sacrificing security and effectiveness. In this article, two dual access control systems are developed, each of which is intended for a different planned environment. There is also a presentation of the systems' experimental and security analyses.
DETECTION OF PHISHING EMAILS USING AN IMPROVED RCNN MODEL WITH MULTILEVEL VECTORS AND AN ATTENTION MECHANISM
P. Samson Anosh Babu1, P. Hema Gayathri2, G. Himaja3, J. Jyothi4, A. G M Abhinaya5
1Associate Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Phishing E-Mail, RCNN, Multilevel vector.
ABSTRACT
Phishing emails are one of the significant threats in today's society and have led to significant financial losses. The results of confrontation approaches are currently not particularly good, despite ongoing improvements. Additionally, the quantity of phishing emails has been rapidly increasing in recent years. More efficient phishing detection technology is needed to lessen the threat presented by phishing emails. In this study, we began by looking at the format of emails. Then, we provide an improved Recurrent Convolutional Neural Networks (RCNN) framework with multilevel vectors and an attention mechanism based on a new Fraud email detection model that concurrently models emails at the email header, email content, character level, and word level. To evaluate how well the recommended method works, we use an unbalanced dataset with real ratios of legitimate and phishing emails. As a consequence of this effort, the filter will have a high likelihood of identifying phishing emails and will exclude as few real emails as possible. The trial's findings were favourable.
THE CASE OF CROSS-SITE REQUEST FORGERY AND MACHINE LEARNING FOR WEB VULNERABILITY DETECTION
J.V.Anil Kumar1, P. Lakshmi Usha Sri2, K. Bewala3, G. Kavya4, SK. Abdul Munaf5, M. Amith6
1Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Machine Learning, Mitch, vulnerability, detection techniques.
ABSTRACT
In this paper, we present a method in this research for using machine learning (ML) to identify weaknesses in web applications. Web applications provide a particularly challenging set of analytical issues because of their variety and frequent usage of custom programming approaches. Because it may employ manually labelled data to incorporate automatic analytic tools with a human's understanding of the semantics of online applications, machine learning is therefore very advantageous for web application security. We used our methods to develop Mitch, the first machine learning (ML) tool for detecting Cross-Site Request Forgery (CSRF) vulnerabilities. Mitch assisted us in discovering 35 new CSRFs on 20 important websites and 3 new CSRFs in production applications.
AN APPLICATION OF A DEEP LEARNING SYSTEM FOR AUTOMATED IDENTIFICATION OF UNFORESEEN INCIDENTS IN TUNNELS UNDER POOR CCTV SURVEILLANCE CIRCUMSTANCES
J.V.Anil Kumar1, Y. Jayalakshmi2, J. Bhavana3, J. Dileep Kumar4, CH. Santhi Priya5
1Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Average precision, CCTV, ODTS, CNN.
ABSTRACT
Object Detection and Tracking System (ODTS) will be introduced and applied in this project along with the well-known deep learning network Faster Regional Convolution Neural Network (Faster R-CNN) for Object Detection and Conventional Object Tracking algorithm for automatic detection and analysis of unexpected events on CCTVs in tunnels, which are probable to include (1) Wrong-Way Driving (WWD), (2) Stop, (3) Person out of vehicle in tunnel, (4) Fire. The Bounding Box (BBox) results from Object Detection are obtained by ODTS using a video frame in time as an input. To identify each moving and detected object, a unique ID number is then assigned by comparing the BBox results of the current and previous video frames. This technique makes it feasible to follow a moving item in real time, something that is typically not achievable with other object detection frameworks. A collection of event photos in tunnels was used to train a deep learning model in ODTS, which resulted in Average Precision (AP) values for the target objects Car, Person, and Fire of 0.8479, 0.7161, and 0.9085, respectively. The Tunnel CCTV Accident Detection System was then evaluated using four accident recordings that included each accident, based on a trained deep learning model. As a result, the system has a 10-second detection time for all incidents. The most crucial aspect is that, as the training dataset grows in size, the detection ability of DTS might be automatically improved without any changes to the programme codes.
WEAPON DETECTION USING ARTIFICIAL INTELLIGENCE AND DEEP LEARNING FOR SECURITY APPLICATIONS: IMPLEMENTATION
SK Althaf Hussain Basha1, A Susmitha Reddy2, Ch Vyshnavi3, P Premavathi4, V Venkateswara Reddy5
1Professor & Head, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Weapon, CNN, RCNN, Faster_ RCNN, SSD, Detection.
ABSTRACT
Due to an increase in crime at packed events and unsettling lonely regions, security is always a top issue in all fields. Computer vision is widely used in abnormal detection and monitoring to solve various issues. Due to the increasing need to defend human safety, security, and property, video surveillance systems that can identify and decipher scene and anomaly occurrences are essential for intelligence monitoring. This project uses the SSD and Faster RCNN convolution neural network (CNN) techniques to create automated gun (or) weapon detection. The suggested implementation employs two different datasets. One dataset contained images that were already labelled, and the other contained images that needed to be manually labelled. Both methods produce high accuracy in the results tabulated, but their practical use may depend on the trade-off between time and precision.
RECOGNIZING FACES IN DISGUISE USING TRANSFER LEARNING
P. Samson AnoshBabu1, Y.Rajeswari2, G.Vasudha3, B.Sreenatha Reddy4, G. Rakesh5
1Associate Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Machine Learning, Face Recognition, Convolutional Neural Network, VGG Model.
ABSTRACT
Machine learning uses the technique of face recognition to identify items in a photo or video. Humans can remember other individuals and some items, including animals, plants, living things, and non-living things. This may be accomplished by computers utilizing the Computer Vision field's Machine Learning approach. Additionally, computers are capable of deciphering the faces of individuals in a picture or video. This study suggests putting three well-known Convolutional Neural Network (CNN) Model Architectures to the test to determine which one is best at recognizing a person's face while they are disguised. The "Recognizing Disguised Faces" dataset is used in this study to separate 75 groups of faces, after which it attempts to train and evaluate its models to determine their accuracy in computer recognition. This research is anticipated to advance the machine learning-related algorithm utilised to address the picture classification issue. Utilizing transfer learning in VGG Models significantly improves the experimental outcomes. In this study, face recognition using VGG Models works best when utilizing Image Net weights.
FARMER’S PORTAL: A STUDY OF BLOCK CHAIN TECHNOLOGY
J.Mahalakshmi1, I.Sahithi Thanmai2, S.Venkata Sai Usharani3, G.Sushma4, G.Venkateswarlu5
1Associate Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India Email: mahalakshmi1203@gmail.com
Keywords - Block chain, Digitization, Crypto-currency, Immutability, Public-ledger, ICT, Farmer’s Portal.
ABSTRACT
Block chain is a technique that uses a cryptocurrency to maintain a record of a transaction's confirmation. The record is kept across several computers connected by a peer-to-peer network. The economic system of a nation is defined by contracts, transactions, and the records of those activities. They define limits and provide the assets security. This study emphasises the use of block chain technology with farmer's site that maintains the footage of selling and purchasing information of crops, taking into account the characteristics of block chain such as immutability and keeping the footage of transaction data. Python is a programming language that is integrated with the block chain system in the suggested solution, which would help farmers, vendors, and individuals by maintaining the contract of trade. Block chain technology and the Python programming language are used to create an interface for farmers that stores data on the seller, the buyer, the selling and purchasing of an item, as well as the overall value of the transaction.
DEEP LEARNING FOR BIRD SPECIES IDENTIFICATION
P Pullaiah1, N.Raziya2, A.V.SriLakshmi3, G.V.L.Harshitha4, J.Praveen Kumar5
1Asst. Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Autograph, Caltech, DCNN, grey scale, pixels, Tensor flow.
ABSTRACT
Now a day some bird species are being found rarely and if found classification of bird species prediction is difficult. Naturally, birds present in various scenarios appear indifferent sizes, shapes, colors, and angles from human perspective. Besides, the images present strong variations to identify the bird species more than audio classification. Also, human ability to recognize the birds through the images is more understandable. So this method uses the Caltech UCSD Birds 200 [CUB-200-2011] dataset for training as well as testing purpose. By using deep convolution neural network (DCNN) algorithm an image converted into gray scale for mat to generate autograph by using tensor flow, where the multiple nodes of comparison are generated. These different nodes are compared with the testing data set and score sheet is obtained from it. After analyzing the score sheet it can predicate the required bird species by using highest score.
AN EXAMINATION SYSTEM AUTOMATION USING NATURAL LANGUAGE PROCESSING
A. Amrutavalli1, A.Swathi2, P.Rishitha Jahnavi3, J.Raveendra4, T.Venkata Subbaiah5
1Asst. Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Exam System; SQLite3; Django; Descriptive System; Natural Language Processing, Python; NLTK.
ABSTRACT
This world has seen a lot many examination portals that are deployed over several servers which are used to conduct online examination for various purposes among which some may include conducting a test for entrance examinations, or Olympiads at national and international level and while some portals are designed to conduct a test for placement purposes.But what we have seen is that mostly all the portals are designed to conduct tests that contain multiple choice questions. Here our aim is not to work on the technology that is already existing, rather some technology that is very rare. Here we talk of the descriptive online examination system. Multiple choice questions are easy to deal as they have a question, a few options and a field in the same question that stores the correct option in the database. While in the case of descriptive questions it is not so. It brings in or uses the concepts of Natural Language Processing or NLP to assign marks to answers. Answers are nothing but strings and the job of the model is to do some operations on the answer string such that it can assign the correct marks to answers written by the examinee. The data is basically collected from a descriptive online examination system. Further, it is analyzed and the designed model assigns accurate marks to the answers for the question. The back-end is written in Python where the web framework usedis Django, the library used for Natural Language Processing includes NLTK and for database purpose, SQLite version 3 is used, while for the front-end HTML version-5, CSS version-3, Bootstrap and JavaScript is used.
BLOCK CHAIN AND AI FOR DATA SECURITY
K. Raj Kiran1,G. Tanuja2, N. Suvarna Kumari3, B. Priyanka4, Sk. Riyaz5
1Asst. Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Data security, Data systems, Artificial intelligence, Cyberspace.
ABSTRACT
Data is the input for various artificial intelligence (AI) algorithms to mine valuable features, yet data in Internet is scattered everywhere and controlled by different stakeholders who cannot believe in each other, and usage of the data in complex cyberspace is difficult to authorize or to validate. As a result, it is very difficult to enable data sharing in cyberspace for the real big data, as well as a real powerful AI. In this paper, we propose the Sec Net, an architecture that can enable secure data storing, computing, and sharing in the large-scale Internet environment, aiming at a more secure cyberspace with real big data and thus enhanced AI with plenty of data source, by integrating three key components: 1) block chain based data sharing with ownership guarantee, which enables trusted data sharing in the large scale environment to form real big data; 2) AI-based secure computing platform to produc more intelligent security rules, which helps to construct a more trusted cyberspace; 3) trusted value-exchange mechanism for purchasing security service, providing a way for participants to gain economic rewards when giving out their data or service, which promotes the data sharing and thus achieves better performance of AI. Moreover, we discuss the typical use scenario of Sec Net as well as its potentially alternative way to deploy, as well as analyse its effectiveness from the aspect of network security and economic revenue.
TEXT CLASSIFICATION USING THE RANDOM FOREST ALGORITHM: AN APPLICATION STUDY
G Mahesh1, A. Ramya Sri2, T. Sai Madhuri3, B. Swathi4, K. Sivananda Reddy5
1Asst. Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Text Classification, Random forest algorithm, vector generation mode, tr-k method.
ABSTRACT
In view of the poor classification effect of traditional random forest algorithms due to the low quality of text feature extraction, a random forest method for text information is proposed. In view of the difficulty in controlling the quality of traditional random forest decision trees, a weighted voting mechanism is proposed to improve the quality of decision trees. This algorithm uses tr-k method based on text feature extraction to improve the quality and diversity of text features, and uses the latest Bert word vector generation model to represent the text. Experimental data in the Python environment show that this method can achieve better results in text classification than IDF based random.
USING CNN AND TRANSFER LEARNING TO RECOGNIZE HUMAN ACTIVITY BASED ON VISION
Sk. Yasmin Sulthana1, S.Ruchika2, M. Varalakshmi3, D.Khadar Vali4, B.Uday Sankar5
1Asst. Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - CNN, Transfer Learning,VGG16 , HAR.
ABSTRACT
With the advent of the Internet of Things(IoT), there have been significant advancements in the area of human activity recognition (HAR)in recent years. HAR is applicable to wider application such as elderly care, anomalous behaviour detection and surveillance system. Several machine learning algorithms have been employed to predict the activities performed by the human in an environment. However, traditional machine learning approaches have been outperformed by feature engineering methods which can select an optimal set of features. On the contrary, it is known that deep learning models such as Convolutional Neural Networks (CNN) can extract feature and reduce the computational cost automatically. In this paper, we use CNN model to predict human activities from Image Dataset model. Specifically, we employ transfer learning to get deep image features and trained machine learning classifiers. Our experimental results showed the accuracy of 96.95%using VGG-16. Our experimental results also confirmed the high performance of VGG-16 as compared to rest of the applied CNN models.
STUDY OF HOUSE PRICING PREDICTION USING PYTHON AND MACHINE LEARNING: IMPLEMENTATION
SK Althaf Hussain Basha1, A.Sucharitha2, N.V.Chandrika Naidu3, P.Naga Mallika4, K.Swetha5
1Professor & Head, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords - Machine learning, Regression Technique, Classification Technique.
ABSTRACT
In this paper provides an overview about how to predict house costs utilizing different regression methods with the assistance of python libraries. The proposed technique considered the more refined aspects used for the calculation of house price and provided the more accurate prediction. It also provides a brief about various graphical and numerical techniques which will be required to predict the price of a house. In this paper contains what and how the house pricing model works with the help of machine learning and which dataset is used in our proposed model.
WITH DATA SHARING AND BLOCK CHAIN ASSISTED COLLABORATIVE SERVICE RECOMMENDATION
Sk. Alimoon1, Yakkali Sindhu2, K. Bhavana3, P. Sateesh Reddy4, S. Sai Swetha5
1Associate Professor, Krishna Chaitanya Institute of Technology & Sciences, Markapur, A.P, India
2,3,4,5Scholar, Krishna Chaitanya Institute of Technology & Sciences, Markapur, India
Keywords -
ABSTRACT
With the speedy growth of cloud computing, many new online services have appeared, placing a great strain on customers to select the services they like. Recommendation algorithms are required in order to suggest online services to users, and several of them have lately been studied. However, the majority of the current recommendation models rely on centralised databases of historical information, which might result in a single point of failure. Most cloud platforms are often hesitant to disclose their own data since it typically contains a lot of sensitive information that might endanger user privacy. Secure data exchange between cloud platforms is required for improved recommendations, which can optimise profitability, in order to address the aforementioned problems. In this paper, we suggest a collaborative service recommendation system that uses block chain technology (BC-SRDS). To encrypt the data, we specifically use the cipher text- policy attribute-based encryption (CP-ABE) method, which secures data secrecy and enables safe data transfer. Then, in order to avoid DoS attacks, DDoS attacks, and single points of failure, we use the block chain to share data. In the meanwhile, the block chain ensures data integrity and tamper-proofing. And in order to suggest the services to consumers, we employ a locality-sensitive hashing technique. Finally, the security analysis demonstrates that BC-SRDS can provide data secrecy, data integrity, and tamper-proof ness. A number of tests demonstrate that BC-SRDS outperforms the current schemes in terms of suggestion accuracy.