Research Projects
Some projects have been completed and others are still in progress.
The code and other materials will be uploaded when I have free time. if you are in a rush and need more information or the code of any part of the project, please let me know 🙂
It is worth to mentions, there are many other projects with other people I did not add to this page for some reasons.
DeepSignature
It is a novel approach that combines crowdsourcing and machine learning to improve the security, efficiency, and accuracy of the web. DeepSignature employs character-based word representation to create signatures of web elements that can be used by a classifier element. The DeepSignature classifier model has an accuracy rate of 98%. Moreover, it overcomes many critical security vulnerabilities.
Ad Blocking And Internet User Privacy
By analyzing the crowdsourcing process, we’ve found errors and misunderstandings in the behavior of Ad-blocking systems. Moreover, we found vulnerabilities used by the ad published to circumvent the systems. This project is a lighthouse for the ad-blocking developer and those how interested in the user privacy.
Ad-blocking systems such as AdBlock Plus rely on crowdsourcing to build and maintain filter lists, which are the basis for determining which ads to block on web pages. In this project, we seek to advance our understanding of the ad-blocking community as well as the errors and pitfalls of the crowdsourcing process.
To do so, we collected and analyzed a longitudinal dataset that covered the dynamic changes of the popular filter-list EasyList for eight years and the error reports submitted by the crowd in the same period.
Selector of TIQ Flash ADC
In this work, we build many new algorithms based on a dynamic programming approach and other approaches along with DNL and INL simulation results. In comparison with state-of-the-art methods, 4 times, and 5 times DNL improvements are achieved through the new approach for 6-bit and 8-bit respectively, and others.
Flash analog-to-digital converter (ADC) is known for its high-speed operation among ADCs. A Flash ADC core consists of two sections: the voltage comparator section and the thermometer code (TC) to binary number (BN) encoder section.
At any instant of time, an analog input voltage is compared with many reference voltages. The comparison happens at once using many voltage comparators in parallel, producing the binary number corresponding to the input voltage at that instant. For instance, n-bit ADC requires 2n-1 reference voltages and 2n-1 voltage comparators are needed. For an 8-bit ADC example, there are 255 reference voltages and 255 voltage comparators are needed. By using the data analysis, we aim to build new algorithms that help to improve the voltage comparators.
Digital & Islamic Art Symmetry
Much research has been conducted to study symmetry in Islamic art. In this project, we try to find the relationship between symmetry groups and the regions where the images are from. And we aim to use AI to enhance our perspective of Islamic art.
In best of our knowledge, this is a novel study that contributes an analysis of the relationship using real images from Islamic geometrical Art. In order to analysis the real data, two crawlers have been built to collect data and extracting the features of it. We have prepared and filtered the datasets, which collected. Then, we have analyzed the data statistically. Moreover, a supervised learning technique has been used to classify the data and study the structure of it.
PREDICT AGGREGATE RATINGS
With the rapid expanding of Internet usage, many issues have been arising, which affect many existing systems and algorithms, including the rating systems. In this work, we provide analyses on ratings.
Creditable information conveyed by aggregate ratings is valuable for e-commerce transactions. The considerable and fast expansion of Internet usage boosts the importance of aggregate ratings. However, the acquisition of a sufficient number
of reliable user ratings becomes a critical requirement in building ranking services.
Spamicity and inaccuracy are the primary issues causing unfair ratings, which
impacts the precision of aggregate ratings. In order to find a solution and eliminate
unfair ratings, we perform an analysis to better understand the behavior of users
and rating processes in rating systems. Moreover, we propose a novel algorithm that leverages product categories to predict aggregate ratings. The algorithm has shown statistically significant results in predicting aggregate ratings. This study shows evidence of the benefit of using products categories
Social network Recommendation application
As a part of my research, I built this project as an android app in 2014 to provide a live-recommendation method that allows users to interact with each other to convey information about a specific target at the same time. This idea and implementation of the project were before its time. I mean it was before many similar applications.
Advanced mobile networks and devices today make the internet within reach, affecting the social network and e-commerce marking significantly. Therefore, online recommendations are very important. But and in some cases, we need live recommendations, not just online. Some information will not be helpful since time is the main vector, for example, asking about the weather conditions directly or asking about the security situation in dangerous places. It may become a good idea to ask directly for help. The application is a social application that works on the smartphone to help connect the users in a secure channel that helps the receiver and provide immediate service to the sender.
The application provides liver commands about the places whether they are restaurants, malls, shops, streets, or even celebrations held at specific paces. We use many techniques such as GeoHash algorithm.
