Author
Madhuri Bhalekar
Other affiliations: Maharashtra Institute of Technology
Bio: Madhuri Bhalekar is an academic researcher from Massachusetts Institute of Technology. The author has contributed to research in topics: Artificial intelligence & Computer science. The author has an hindex of 2, co-authored 7 publications receiving 14 citations. Previous affiliations of Madhuri Bhalekar include Maharashtra Institute of Technology.
Papers
More filters
••
01 Jan 2017TL;DR: A system that uses SVM technique along with map-reduce paradigm to achieve a higher accuracy in detection of the spam email is proposed and tries to overcome the two hurdles of the SVM.
Abstract: Phishing is a criminal scheme to steal the user's personal data and other credential information. It is a fraud that acquires victim's confidential information such as password, bank account detail, credit card number, financial username and password etc. and later it can be misuse by attacker. We aim to use fundamental visual features of a web page's appearance as the basis of detecting page similarities. We propose a novel solution, to efficiently detect phishing web pages. Note that page layouts and contents are fundamental feature of web pages' appearance. Since the standard way to specify page layouts is through the style sheet (CSS), we develop an algorithm to detect similarities in key elements related to CSS. In this paper, we proposed a system that uses SVM technique along with map-reduce paradigm to achieve a higher accuracy in detection of the spam email. By using the map-reduce technique we also try to overcome the two hurdles of the SVM.
15 citations
••
09 Apr 2022
TL;DR: An image captioning system that generates detailed captions and extracts text from an image, if any, and uses it as a part of the caption to provide a more precise description of the image.
Abstract: Automatically describing the information of an image using properly constructed sentences is a tricky task in any language. However, it has the potential to have a significant effect by enabling visually challenged individuals to better understand their surroundings. This paper proposes an image captioning system that generates detailed captions and extracts text from an image, if any, and uses it as a part of the caption to provide a more precise description of the image. To extract the image features, the proposed model uses Convolutional Neural Networks (CNNs) followed by Long Short-Term Memory (LSTM) that generates corresponding sentences based on the learned image features. Further, using the text extraction module, the extracted text (if any) is included in the image description and the captions are presented in audio form. Publicly available benchmark datasets for image captioning like MS COCO, Flickr-8k, Flickr-30k have a variety of images, but they hardly have images that contain textual information. These datasets are not sufficient for the proposed model and this has resulted in the creation of a new image caption dataset that contains images with textual content. With the newly created dataset, comparative analysis of the experimental results is performed on the proposed model and the existing pre-trained model. The obtained experimental results show that the proposed model is equally effective as the existing one in subtitle image captioning models and provides more insights about the image by performing text extraction.
7 citations
••
01 Jan 2019TL;DR: This paper is presenting an approach which can be used to detect an image is a selfie and the value of foreground and background will be compared with a certain threshold value and according to the obtained result can recognized whether an image are a selfie or not.
Abstract: Selfie is the act of taking self portrait through the front camera of the mobile. By visualizing the captured image one can identify the details regarding the image such as number of objects, location and much more. By the method of object recognition in the image we can identify whether the image taken is a selfie or not. For this we should first segregate both the foreground and background details from an image. From the details of foreground one can identify the object (i.e., the person taking the selfie) and from the background we can tell about the location. Using various object recognition methods such as exhaustive search, segmentation, selective search, Gaussian mixture model the information regarding objects, foreground and background can be detected. And further the value of foreground and background will be compared with a certain threshold value and according to the obtained result can recognized whether an image is a selfie or not. In this paper we are presenting an approach which can be used to detect an image is a selfie.
3 citations
••
07 Sep 2019TL;DR: This work presents an image caption generating system that uses convolutional neural network for extracting the feature embedding of an image and feed that as an input to long short-term memory cells that generates a caption.
Abstract: Recent advancements in technology have made available a variety of image capturing devices, ranging from handheld mobiles to space-grade rovers. This has generated a tremendous visual data, which has made a necessity to organize and understand this visual data. Thus, there is a need to caption thousands of such images. This has resulted in tremendous research in computer vision and deep learning. Inspired by such recent works, we present an image caption generating system that uses convolutional neural network (CNN) for extracting the feature embedding of an image and feed that as an input to long short-term memory cells that generates a caption. We are using two pre-trained CNN models on ImageNet, VGG16 and ResNet-101. Both the models were tested and compared on Flickr8K dataset. Experimental results on this dataset demonstrate that the proposed architecture is as efficient as the state-of-the-art multi-label classification models. Experimental results on public benchmark dataset demonstrate that the proposed architecture performs as efficiently as the state-of-the-art image captioning model.
3 citations
01 Jan 2014
TL;DR: TARF is designed and implemented for dynamic WSN to provide the trustworthiness and Energy efficient route without any known geographic information neither require tight time synchronization and proved to be effective against such attacks.
Abstract: In multi hop routing, Wireless sensor Networks (WSNs) plays a vital role by preventing the routing information against the identity deception. The attacks such as sinkhole attack, wormhole attack, Sybil attack etc. are launched against the routing protocol which damages the network. Traditional cryptographic techniques or even trust aware routing protocols could not solve these severe problems. Thus to secure the WSN against such attacks, Trust aware Routing Frame work (TARF) is designed and implemented for dynamic WSN. TARF provide the trustworthiness and Energy efficient route without any known geographic information neither require tight time synchronization and proved to be effective against such attacks.
2 citations
Cited by
More filters
••
TL;DR: The goal is to explore the current research state and identify open research issues by surveying proposed schemes and evaluate proposed schemes based on two important factors, which are energy consumption and attack resiliency.
Abstract: Routing is one of the most important operations in wireless sensor networks (WSNs) as it deals with data delivery to base stations. Routing attacks can cripple it easily and degrade the operation of WSNs significantly. Traditional security mechanisms such as cryptography and authentication alone cannot cope with some of the routing attacks as they come from compromised nodes mostly. Recently, trust mechanism is introduced to enhance security and improve cooperation among nodes. In routing, trust mechanism avoids/includes nodes in routing operation based on the estimated trust value. Many trust-based routing protocols are proposed to secure routing, in which they consider different routing attacks. In this research work, our goal is to explore the current research state and identify open research issues by surveying proposed schemes. To achieve our goal we extensively analyze and discuss proposed schemes based on the proposed framework. Moreover, we evaluate proposed schemes based on two important factors, which are energy consumption and attack resiliency. We discuss and present open research issues in the proposed schemes and research field.
51 citations
••
TL;DR: This study focused on an algorithm that was thoroughly made and the methods in implementing this algorithm are discussed in detail, which can be used in the machine learning method to prevent phishing attacks.
Abstract: The development of computer networks today has increased rapidly. This can be seen based on the trend of computer users around the world, whereby they need to connect their computer to the Internet. This shows that the use of Internet networks is very important, whether for work purposes or access to social media accounts. However, in widely using this computer network, the privacy of computer users is in danger, especially for computer users who do not install security systems in their computer. This problem will allow hackers to hack and commit network attacks. This is very dangerous, especially for Internet users because hackers can steal confidential information such as bank login account or social media login account. The attacks that can be made include phishing attacks. The goal of this study is to review the types of phishing attacks and current methods used in preventing them. Based on the literature, the machine learning method is widely used to prevent phishing attacks. There are several algorithms that can be used in the machine learning method to prevent these attacks. This study focused on an algorithm that was thoroughly made and the methods in implementing this algorithm are discussed in detail.
18 citations
••
TL;DR: The e-mail phishing detection is performed in this paper using the optimization-based deep learning networks and it is clear that the proposed method acquired the maximal accuracy, sensitivity and specificity compared with the existing methods.
Abstract: Phishing is a serious cybersecurity problem, which is widely available through multimedia, such as e-mail and Short Messaging Service (SMS) to collect the personal information of the individual. However, the rapid growth of the unsolicited and unwanted information needs to be addressed, raising the necessity of the technology to develop any effective anti-phishing methods.,The primary intention of this research is to design and develop an approach for preventing phishing by proposing an optimization algorithm. The proposed approach involves four steps, namely preprocessing, feature extraction, feature selection and classification, for dealing with phishing e-mails. Initially, the input data set is subjected to the preprocessing, which removes stop words and stemming in the data and the preprocessed output is given to the feature extraction process. By extracting keyword frequency from the preprocessed, the important words are selected as the features. Then, the feature selection process is carried out using the Bhattacharya distance such that only the significant features that can aid the classification are selected. Using the selected features, the classification is done using the deep belief network (DBN) that is trained using the proposed fractional-earthworm optimization algorithm (EWA). The proposed fractional-EWA is designed by the integration of EWA and fractional calculus to determine the weights in the DBN optimally.,The accuracy of the methods, naive Bayes (NB), DBN, neural network (NN), EWA-DBN and fractional EWA-DBN is 0.5333, 0.5455, 0.5556, 0.5714 and 0.8571, respectively. The sensitivity of the methods, NB, DBN, NN, EWA-DBN and fractional EWA-DBN is 0.4558, 0.5631, 0.7035, 0.7045 and 0.8182, respectively. Likewise, the specificity of the methods, NB, DBN, NN, EWA-DBN and fractional EWA-DBN is 0.5052, 0.5631, 0.7028, 0.7040 and 0.8800, respectively. It is clear from the comparative table that the proposed method acquired the maximal accuracy, sensitivity and specificity compared with the existing methods.,The e-mail phishing detection is performed in this paper using the optimization-based deep learning networks. The e-mails include a number of unwanted messages that are to be detected in order to avoid the storage issues. The importance of the method is that the inclusion of the historical data in the detection process enhances the accuracy of detection.
11 citations
••
01 Nov 2019
TL;DR: The combination of OSS and SMOTE can be a plausible option to handle the imbalanced class problem on the web phishing classification either on binary class and multiclass datasets.
Abstract: From the previous work related to web phishing, the researchers overlook the imbalanced class problem on the dataset. theoretically, the majority of classification methods would assume that the nature of the class distribution is balanced. It caused the classification’s performance of the method will be declining. Therefore, the mechanism of imbalanced class handling is severely needed. In our study, One SidedSelection and Synthetic Minority Over-Sampling Technique are used to handle the imbalanced class condition. Those algorithms work to balancing the class distribution of the dataset so that the accuracy and the gmean score of the classification will be enhanced. Based on the result, the combination of those methods (OSS and SMOTE) can enhance the classification’s result significantly either on binary type class and multiclass type dataset. Hence, the combination of OSS and SMOTE can be a plausible option to handle the imbalanced class problem on the web phishing classification either on binary class and multiclass datasets.
8 citations
••
09 Apr 2022
TL;DR: An image captioning system that generates detailed captions and extracts text from an image, if any, and uses it as a part of the caption to provide a more precise description of the image.
Abstract: Automatically describing the information of an image using properly constructed sentences is a tricky task in any language. However, it has the potential to have a significant effect by enabling visually challenged individuals to better understand their surroundings. This paper proposes an image captioning system that generates detailed captions and extracts text from an image, if any, and uses it as a part of the caption to provide a more precise description of the image. To extract the image features, the proposed model uses Convolutional Neural Networks (CNNs) followed by Long Short-Term Memory (LSTM) that generates corresponding sentences based on the learned image features. Further, using the text extraction module, the extracted text (if any) is included in the image description and the captions are presented in audio form. Publicly available benchmark datasets for image captioning like MS COCO, Flickr-8k, Flickr-30k have a variety of images, but they hardly have images that contain textual information. These datasets are not sufficient for the proposed model and this has resulted in the creation of a new image caption dataset that contains images with textual content. With the newly created dataset, comparative analysis of the experimental results is performed on the proposed model and the existing pre-trained model. The obtained experimental results show that the proposed model is equally effective as the existing one in subtitle image captioning models and provides more insights about the image by performing text extraction.
7 citations