In the final project, my teammates and I want to have a deeper research regarding to a question about link prediction: how’s the importance of geography information, such as location and graph features. We use two famous social network from Stanford database: Gowalla and Brightkite. The features we take into consider can be categories into
Continue Reading “Social Network and Analysis Final Project”
This homework is about predicting the existence of edges. In the training data, one is given some pair of nodes, which indicates the existing nodes in the network. Later, we are asked to predict whether there is edge for a pair of given node or not. There are many strategies for this kind of problem.
In this project, we are going to play a game in a predefined network; while all information are hidden. Initially, two players is in charge of a fixed amount of nodes randomly. Later on, players will extend the influence from existing nodes across the network based on their derived algorithm round by round. At the
There are two parts in this project. In the first part, I need to calculate some basic properties of a given network, including alpha value, pair rate, closeness centrality, and betweenness centrality. In the second part, I need to implement Independent Cascade Model (IC Model) and Linear Threshold Model (LT Model).
Before this project starts, each student is asked to write each Chinese Zodiac word ten times. After that, these handwriting images are collected as data for this project. As one can imagine, these written words might be tilted or not centered. Thus, we use a preprocessing pipeline to deal with as much as corner case
In this final project, we need to pipeline all the tasks we have done before. It might sound easy in the beginning; however, we soon notice that the error propagation issue is quite significant. Thus, we need to go back and improve each previous project model individually. For Hw1, we try three other models: Improved
We are given a raw dataset in xml formats. Thus, we not only need to work on preprocessing, such as deleting headers, punctuations, and special characters, but also make use of some existing library, ex: NTLK in Python and `word2vec`, to do the advanced parsing. There are three proposed solutions from us, including Random Guess,
In this project, the given data is multiple sequential sound records. In other words, feature `x` is a vector of length `k`; while `y` is a scalar label. There are some known solutions to this kind of learning problem, including structured Support Vector Machine, Deep Learning with Hidden Markov Model, and Recurrent Neural Network. We
In this project, my teammates and I want to research a relatively new security vulnerability in bash: Shellshock. According our survey, this vulnerability is most exploited in Common Gateway Interface-based (CGI-based) web server, Dynamic Host Configuration Protocol (DHCP), and Limited OpenSSH servers. To present this vulnerability, we build up a website based on CGI and
Continue Reading “Cryptography and Network Security Final Project”
In the final project, our teammates decide to work on a open machine learning problem on Kaggle — Titanic: Survival Prediction. We try different models in Sklearn package and it turns out that Adaboost algorithm has the best performance.
Recent Comments