Search

Due to the highest mortality rate in the globe, cancer still poses a severe threat to individuals today. It is because there is insufficient early cancer detection. One of the most deadly diseases for women worldwide and the most frequently detected non-skin cancer in women is breast cancer carcinoma. The probability of a successful treatment is significantly increased by diagnostic testing of this lifethreatening disease. The traditional method of diagnosing cancer highly depends on the technician’s or doctor’s ability in spotting anomalies in the human body. Oncethe breast’s cell tissue starts to divide irregularly and aggressively, breast cancer develops. Bosom malignant tumor can be classified into five stages (0–IV), each of which reflects the severity of the disease within the patient’s body. Mammary gland illness is complicated and has many varieties of clinical effects, which makes it difficult to predict and treat. Additionally, the ability to more accurately predict the onset of a malignant infection will benefit breast cancer patients in planning their future course of treatment under their doctor’s directions. It is particularly challenging to predict breast cancer because of its great heterogeneity and complicated features. This projectexamines the accuracy of multiple CNN models and focuses on the prediction of Breast disease using mammography images.Therefore in the study, we explore various CNN models to identify malignancy in mammography scans, including ResNet- 50, VGG-16, Alex Net, and Google Net. Additionally, we are going to build a special Xception model with 70% accuracy to diagnose breast cancer and compare each model.

One of the most challenging security requirements in a PaaS platform setting is authentication. We provide a new signature-based authenticated key establishment approach for PaaS platform security improvement in this study. It adds a feature to increase confidentiality and non-repudiation by requiring the sender to sign a specific action while sending data from one user to another. It’s signed with the transaction’s private key and a public-key value-based signature generated with the transaction’s public key. Use the public key value-based signature that was generated to authenticate this operation, and the hashing process assures that the operation is immutable. Our proposed solution delivers improved data security, throughput, reaction speed, and a reduction in end-to-end delay and overhead when compared to existing methodologies. The results of the experiments reveal that the proposed system achieves a superior outcome to the existing methods. This project makes use of the MATLAB platform.

This paper aims to compare two recommender systems and each using different selection techniques in Genetic Algorithm (GA) for making user clusters. The recommender systems under consideration are: Information Gain Clustering through Rank based Genetic Algorithm (IGCRGA) and Information Gain Clustering through Roulette Wheel Genetic Algorithm (IGCRWGA) both of which taking influence from IGCN (Information Gain through Clustered neighbors) Algorithm. Unlike k-means, which is used by IGCN, GAsare not dependent on initial configuration for giving global solution. These two recommender systems were separately experimented and then evaluated using Mean Absolute Error (MAE) as evaluation metric by Abdul Hameed et al. Both the techniques showed good results when compared with Popularity, Entropy, IGCN, IGCGA (Information Gain Clustering through Genetic Algorithms), and IGCEGA (Information Gain Clustering through Elitizt Genetic Algorithm). As these two techniques are not compared with each other, we, being the first to do so, experimented these two techniques to find out which works best in giving quality recommendations. As MAE shows false positives the experimentation used Expected Utility (EU) as its second Evaluation metric in addition with MAE. Based on previous work it was expected that IGCRGA would perform better than IGCRWGA but the current work witnessed surprising results which favored IGCRWGA over IGCRGA. Results of EU showed that the performance of RWGA and RGA is purely context based.

Various data mining approaches are now available, which help in handling large static data sets, in spite of limited computational resources. However, these approaches lack in mining high-speed endless streams, as their learning procedure though simple require the entire training process to be repeated for each new arriving information instance. The main challenges while dealing with continuous data streams: they are of sizes many times greater than the available memory, are real-time, and the new instances should be inspected at most once, and predictions must be made. Another issue with continuous real-time data is changing of concepts with time, which is often called concept drift. This paper addresses the above stated problems, and provides a solution by proposing a real-time, scalable, and robust architecture. It is a general-purpose architecture, based on online machine learning, which efficiently logs and mines the stream data in a fault-tolerant manner. It consists of two frameworks: (1) Event aggregation framework, which reliably collects events and messages from multiple sources and ships them to a destination for processing (2) Real-time computation framework, which processes streams online for extraction of information patterns. It guarantees reliable processing of billions of messages per second. Furthermore, it facilitates the evaluation of the stream learning algorithms and offers change detection strategies to detect concept drifts.

Social networking and micro-blogging sites are stores of opinion-bearing content created by human users. Sentiment analysis extracts and measures the sentiment or “attitude” of documents as well as the topics within documents. The attitude may be the person’s judgment (e.g., positive vs. negative) or emotional tone (e.g., objective vs. subjective). In this paper we define a supervised learning solution for sentiment analysis on twitter messages which gives profound accuracy using Naive Bayes Classification. It classifies the tweets either positive or negative. Our dataset consists of tweets containing move reviews retrieved from twitter using certain keywords on a cloud platform. The experiment and analysis has 3 major steps. In the first step, the algorithm used is Naive Bayes which performed Boolean classification on bag of words, resulting in 71 % of accuracy. In the second step, Naive Bayes algorithm is applied on bag of words without stop words which gave 72 % accuracy. In the third step, using the concept of Information gain, high value features are selected using Chi Square, which gave maximum accuracy. The evaluating metrics used in this work are Accuracy, Precision, Recall and F-Measure for both the sentiment classes. The results show that Naive Bayes algorithm with the application of feature selection using the minimum Chi Square value of 3 gave an Accuracy of 89 %, Positive Precision 83.07 %, Positive Recall 97.2 %, Positive F-Measure 89.9 %, Negative Precision 96.6 %, Negative Recall 81.2 % and Negative F-Measure 88.2 %. The significant increase in Positive Recall specifies that the classifier classifies positive words with more probability when compared to the negative words. This paper also describes the preprocessing steps needed in order to achieve high accuracy.

Information shared on social networks can be attributed to various factors such as common interests, innovation in areas of job role, etc. The propagation of information or its reach in a social network is directly related to the influence of the author. Traditional influence scoring algorithms rely only on the number of direct connections of a user or on a ratio of their connections. As proved by recent research, a focus solely on the connection between users skews an understanding of how influence operates and follows. In this paper we differentiate between passive and active information. Information is classified as active if it promotes engagement through conversation and is thus exposed to more members, increasing the reach of information. Using a large amount of social network data we analyze the different components of information generated by a user, its propagation within the network and derive a metric for calculating the user's influence score. This influence score considers both the passive content and the active conversation aspect of the shared information. The score relies on the network metric degree of centrality. The measure of inbound and outbound information for 2 levels in the network, provide insight to the amplification of the reach of the shared information.

Personalization in a recommender system is to customize contents for users based on their preferences and interests. For a new user such systems face cold start problem. This is because system knows nothing about this user and is unable to present recommendations. For the above said problem an existing technique, Information Gain through Clustered Neighbors (IGCN), has proved to be productive but this technique uses k-means algorithm for making user clusters. The problem with k-means algorithm is it might get stuck at local optima and has initial value dependency. Genetic k-means Algorithm (GKA), a hybrid clustering technique, converges to global optima faster than traditional Genetic Algorithms (GAs). The performance of this technique was improved by Fast Genetic K-means algorithm (FGKA). As the above mentioned GAs has proved to overcome disadvantages of k-means, the paper intends to use a GA viz. FGKA for clustering instead of k-means due to its better performance. This is why the proposed algorithm is named Information Gain Clustering through Fast Genetic k-means Algorithm (IGCFGKA). We show through our results that IGCFGKA not only overcomes k-means disadvantages but it also provides high quality recommendations and an optimal or near optimal solution. Our paper is first to compare IGCFGKA with various strategies of Information gain in recommender systems.

Information Gain Clustering through Prototype - Embedded Genetic K-Mean Algorithm (IGCPGKA) is a novel heuristic used in Recommendation System (RS) for solving personalization problems.In a bid to generate information on the behavior and effectiveness of Prototype-Embedded Genetic K-mean Algorithm (PGKA) -- a clustering algorithm - in Recommender System (RS) used in personalization of cold start problem, IGCPGKA is proposed, developed and experimented upon in this work/paper.IGCPGKA is derived from IGCEGA (Information Gain Clustering through Elitizt Genetic Algorithm). The main difference between the two algorithms is articulated and exhibited in the clustering stage, and precisely, PGKA is used in IGCPGKA, while EGA (Elitizt Genetic Algorithm) is used in IGCEGA for clustering purposes. The effects of these differences have created positive results in terms of better recommendation for IGCPGKA and this fact is supported by the two evaluation metrics used in this work, namely Expected Utility (EU) and Mean Absolute Error (MAE).A comparison with other heuristics for personalization of cold start problem - such as Information Gain Clustering Neighbor through Bisecting K-Mean Algorithm (IGCN), Information Gain Clustering through Genetic Algorithm (IGCGA), entropy and popularity -- showed that IGCPGKA emerged vector by producing the best recommendation and this fact is also supported by the two evaluation metrics used in this work.

Today Social Networks have an undeniable impact on how users communicate search and share data online. This data has opened up new venues of research on sociology. Unfortunately mining and analyzing this large amount of data can be a difficult task and expensive in terms of computational resources. With the advent of cloud computing, data mining and analysis can be made more accessible due to its cost-effective computational resources.

Information Gain Clustering through Roulette Wheel Genetic Algorithm (IGCRWGA) is a novel heuristic used in Recommender System (RS) for solving personalization problems. In a bid to generate information on the behavior and effects of Roulette Wheel Genetic Algorithm (RWGA) in Recommender System (RS) used in personalization of cold start problem, IGCRWGA is developed and experimented upon in this work / paper. A comparison with other heuristics for personalization of cold start problem - such as Information Gain Clustering Neighbor through Bisecting K-Mean Algorithm (IGCN), Information Gain Clustering through Genetic Algorithm (GCEGA), among others -- showed that IGCRWGA produced the best recommendation for large recommendation size (i.e. greater than 30 items) since it is associated with the least Mean Absolute Error (MAE), the evaluation metric used in this work.

Genetic algorithms are becoming increasingly valuable in solving large-scale, realistic, difficult problems, and selecting replica with multiple selection criteria - availability, security and time- is one of these problems. In this paper, a rank based elitist clustering Genetic Algorithm is proposed named RRWSGA, which alleviates the problem of being trapped in local clustering centroids using k-mean. Simulation results show that the proposed RRWSGA, outperforms k-mean by 9%. Much better performance of RRWSGA is observed.