Computer Science Masters Papershttps://hdl.handle.net/10365/325522024-03-28T12:23:49Z2024-03-28T12:23:49ZAlternative Clustering Algorithms in Sensor NetworksGupta, Divyahttps://hdl.handle.net/10365/337232024-03-08T19:59:45Z2010-01-01T00:00:00ZAlternative Clustering Algorithms in Sensor Networks
Gupta, Divya
A wireless sensor network is composed of a large number of tiny sensor
nodes that can be deployed in a variety of environments like battle fields, water,
large fields, and the like, and can transmit data to a Base station (BS). In a clusterbased network organization, sensor nodes are organized into clusters and one
sensor node is selected as a sensor head (SH) in each cluster. Each SH denotes a
facility and sends useful information to the Base Station (BS) through other SHs
via the shortest path. In this paper, we study two clustering techniques, namely kmedian clustering and k-center clustering for a wireless sensor network. All the
sensor nodes are static and homogeneous (having the same specifications) and
SHs are assumed to be heterogeneous with respect to other sensor nodes in their
respective clusters (but homogeneous to other SHs once they are located). The
focus of this paper is to compare the k-median and k-center clustering techniques
based on shortest path and total intra-cluster distance. We have implemented the
two clustering techniques using the Java language and necessary experimental
and statistical results are provided.
2010-01-01T00:00:00ZDigital Deception and the Illusion of Choice: How Dark Patterns Undermine Informed Consent GDPRKhan, Wajeehahttps://hdl.handle.net/10365/337132024-03-04T21:36:33Z2023-01-01T00:00:00ZDigital Deception and the Illusion of Choice: How Dark Patterns Undermine Informed Consent GDPR
Khan, Wajeeha
Our investigation builds on prior research to examine global e-commerce data privacy, focusing on compliance with GDPR and CCPA laws introduced in 2018 and 2020. This study reveals uneven adherence to GDPR and CCPA regulations across e-commerce platforms, underscoring the persistent use of dark patterns. UK and French sites lead in GDPR compliance at 85% and 80%, while U.S. sites showed 65% adherence to CCPA. Turkish websites displayed a surprising 85% - 95% compliance with European standards. In contrast, South African platforms showed a low 30% compliance, often utilizing implicit consent methods. These findings expose significant gaps and inconsistencies in the application of data privacy laws across continents and nations. We advocate for a global standardization of data protection regulations to protect consumers and create a level playing field for businesses in the digital marketplace.
2023-01-01T00:00:00ZEvaluation of Convolutional Neural Networks Against Deepfakes Using Transfer LearningKrishan, Siddharthhttps://hdl.handle.net/10365/337102024-03-04T20:44:51Z2023-01-01T00:00:00ZEvaluation of Convolutional Neural Networks Against Deepfakes Using Transfer Learning
Krishan, Siddharth
The main objective of this paper is to evaluate ResNets, DenseNet, Inception and VGG, against deepfake images, to answer the question: How effectively these Convolutional Neural Network can distinguish between deepfake images and real images.
The dataset was acquired from FaceForensics++ and CelebA datasets for manipulated and unmanipulated images respectively. A custom script using Python and OpenCV was applied to create the final dataset for modelling.
Transfer learning is a technique of applying the learned features by a network to a similar approach. It is employed to save time and resources in training, as it does not require a large dataset to allow the network to learn effectively.
The Convolutional Neural Networks are tested against different deep fakes and the networks are evaluated using metrics like precision, recall, accuracy, loss, and f-1 score. It was observed that all the networks used in the experiment performed exceptionally well, but Inception network was slightly better than the other networks in separating the real and fake images.
2023-01-01T00:00:00ZSentiment Analysis of Tweets for Hate Speech Detection Using Binary Classification Algorithms and BERTKaur, Manveerhttps://hdl.handle.net/10365/337082024-03-04T20:32:23Z2023-01-01T00:00:00ZSentiment Analysis of Tweets for Hate Speech Detection Using Binary Classification Algorithms and BERT
Kaur, Manveer
In the modern world, social media wields a lot of power. Twitter, particularly, has provided people a platform to express their opinions about everything under the sun from mundane everyday life to politics, race, religion etc. It has often come under scrutiny for unabashed propagation of hate speech. This project employs natural language processing techniques on a corpus of tweets to detect hate speech. A total of 3538 unique tokens are identified that appear only in tweets classified as hate speech. With the help of data visualization techniques like word clouds and frequency distribution plots, it became evident that the occurrence of sexist, homophobic, and racist slurs is the most frequent in hate tweets. This implies that women, LGBTQ+ community, and people of color are the most targeted sections of society.
2023-01-01T00:00:00Z