Deep learning to filter SMS spam

Improving the performance of machine learning algorithms has been always the topic of interest in data mining. The ensemble learning is one of the machine learning methods that, according to the subject literature, it yields better performance than the single base learner in the accuracy parameter. In the ensemble learning, all base learners are considered at the same level in terms of power and separation capabilities. However, whether the ensemble is made of homogeneous based learners or it is made of heterogeneous base learners, in either case, weaknesses and strengths of the base learners are ignored. To overcome this challenge, the stronger coefficient of influence should be assigned to stronger base learners and the lower coefficient of influence should be assigned to weaker base learners. However, given that the data is associated with uncertainty in the real-world issues, it is impossible to determine which base learner performs better than the others under these circumstances. Learning Automata is one of the desirable options of reinforcement learning subject literature to dealing with dynamic environments. The learning automata works by receiving feedback from the environment. In this paper, a method named LAbEL has been proposed which allows the assignment of the coefficient of influence to each base learner in the ensemble dynamically. Due to the use of learning automata, the proposed method works adjusted to the problem space conditions. The LAbEL is based on learning automata and according to its ability to dealing with dynamic environments, it is possible to apply it to issues where data has nonlinear and unpredictable behavior.

A new ensemble learning method based on learning automata

Spam and non-spam email identification are one of the most challenging tasks for both email service providers and consumers. The spammers try to spread misleading facts through irritating messages by attracting user's attention. Several spam identification-models have previously been proposed and tested but the recorded accuracy has shown that further work in this direction is needed to achieve improved accuracy, low training time, and less error rate. In this research work, we have proposed a model that classifies the e-mail into spam and ham. DBSCAN and Isolation Forest are used to identify the extreme values outside of the specific range. Heatmap, Recursive Feature Elimination, and Chi-Square feature selection techniques are used to select the effective features. The proposed model is implemented in both machine learning and deep learning to establish a comparative analysis. Multinomial Naive Bayes (MNB), Random Forest (RF), K-Nearest Neighbor (KNN), Gradient Boosting (GB) are used to introduce ensemble method in machine learning implementation. Recurrent Neural Network (RNN), Gradient Descent (GD), Artificial Neural Network (ANN) for deep learning implementation. An ensemble method is constructed to combine multiple classifiers' output. The ensemble methods allow producing better prediction accuracy compared to a single classifier. Our proposed model obtained an accuracy of 100%, AUC=100, MSE error = 0 and RMSE error = 0 for machine learning implementation and accuracy of 99%, loss value= 0.0165 for deep learning implementation based on an email spam base dataset collected from the UCI machine learning repository.

Analysis of Optimized Machine Learning and Deep Learning Techniques for Spam Detection

Text mining, a section of the synthetic intelligence, is gaining grounds nowadays in terms of the applications in business and analysis. Varied sectors and domains across industries understand the potential of text mining in gaining information, mining helpful data and in enhancing the choice creating method in terms of speed and potency. In today's world, where the web area is overflowing with data, text mining technology and solutions will persuade be the turning purpose. Most of the recent researches conducted during this space focused principally on advanced or the hybrid of deep neural networks so as to induce economical and higher results. Functioning on such serious models isn't solely time taking however conjointly needs the usage of heap of resources. Therefore, to get the comparable results only basic deep learning models have been used in order to minimize the model's complexity and computational cost. For the same, RNN and LSTM based models have been used and accuracy of proposed models have been enhanced by by varying the hyper parameters and using Glove word embeddings. Moreover, a labeled dataset named Sentiment5k has also been created using semi-supervised learning approach for evaluating the performance of RNN & LSTM based models.

Enhancing Text Mining Using Deep Learning Models

Advances in spam detection for email spam, web spam, social network spam, and review spam: ML-based and nature-inspired-based techniques

In our daily life, we use email and SMS many times to communicate to each other, but due to the increase of spam email and SMS, it becomes a headache for both the sender and receiver. We need spam detection tool to detect the spam, and there are many spam detection tools available in the market but they are not up to the mark because they only emphasize on individual classifier or only one or two combination of classifier. In our research, we present different combinations of four different classifiers, namely “Gaussian Naive Bayes”, “Multinomial Naive Bayes”, “Bernoulli Naive Bayes”, and “Decision Tree”. We have used voting classifier, a type of ensemble learning to calculate the accuracy of different combinations of classifiers. Results show that use of voting classifier produces more accurate prediction than individual classifier. We had also created an android application to serve the purpose. The mobile application works on client–server principle. Basically, the mobile application acts as a client which sends the data clicked by a user from mobile to server. At the server, there is machine learning script which classifies the received data and sends the prediction back to the client.

Spam Detection Using Ensemble Learning

Music, an integral part of our lives, which is not only a source of entertainment but plays an important role in mental well-being by impacting moods, emotions and other affective states. In this age of Big Data, online music streaming services allow us to capture ecologically valid music listening behavior and provide a rich source of information to identify several user-specific aspects. Music preferences and listening strategies have been shown to be associated with the psychological well-being of listeners including internalized symptomatology and depression. However, till date no studies exist that examine music-related social tags, static & time-varying audio-derived features, in terms of acoustic and emotional content, and its association with users well-being. The current work aims at unearthing static and dynamic patterns and trends in the individuals at risk for depression as it manifests in naturally occurring online music consumption. Mental well-being scores and listening histories of 541 Last.fm users were examined. Social tags associated with each listener’s most popular tracks were analyzed to unearth the mood/emotions and genres associated with the users. Results revealed that social tags prevalent in the users at risk for depression were predominantly related to emotions depicting Sadness associated with genre tags representing neo-psychedelic-, avant garde-, dream-pop. Static and dynamic acoustic and emotion-related features were extracted from each user’s listening history and correlated with their mental well-being scores. Results revealed that individuals with greater depression risk resort to higher dependency on music with greater repetitiveness in their listening activity. Furthermore, the affinity of depressed individuals towards music that can be perceived as sad was found to be resistant to change over time. This study has large implications for future work in the area of assessing mental illness risk by exploiting digital footprints of users via online music streaming platforms. This study will open up avenues for an MIR-based approach to characterizing and predicting risk for depression which can be helpful in early detection and additionally provide bases for designing music recommendations accordingly.

Aman Mehta

Papers

Spam Detection Using Ensemble Learning

Online Music Consumption Characterizing Depression Risk