Skip to content

sting1000/ntds_project

Repository files navigation

A Network Tour of NIPS Conference papers

Group 26: Yueran Liang, Peilin Kang, Yawen Hou, Zhechen Su

Machine learning has always been a very popularsubject since the last decade. In the last few years, thenumber of papers submitted to the top machine learningconferences has been increasing exponentially. More andmore researchers become interested to this field andnew concepts and algorithms are proposed every yearat conferences. Among all the conferences, Neural In-formation Processing Systems (NIPS) is the one of theworld’s largest machine learning conference (see sectionIV). For our current project, we will build our networksbased on NIPS papers. We will study the papers’ topicsand their authors’ favorite topics by building a networkof topics using Latent Dirichlet allocation (LDA) andassociate each researcher to the topics they write the mostabout using Author-Topic modeling (ATM). We will alsostudy the connections between individual researchers bybuilding a co-authorship graph using adjacent matrices. Tobetter visualize the networks, we will apply dimensional-ity reduction to our network models using T-distributedStochastic Neighbor Embedding (t-SNE). We will exploitthe distribution the distribution of the papers’ topics overyears and the distribution of topic preference per author.As a result, we will attempt to predict the main topic of anew paper with our LDA model. Given a particular author,we will also attempt to look for other authors that havethe same research orientation as them

We did not include the dataset in our repository as it exceeds the maximum file size that Github allows. Here is the link to download it: https://drive.google.com/drive/folders/1olqiQqQCjWm5MtteurADoaenn-7neo3n. Do not forget to change the path in the jupyter notebook to load correctly the dataset.

Here is an overview of our repository's architecture:

.   
├── model                                   # Folder containing the models we've trained and some processed file that takes a long time to generate from scratch.
├── figure                                  # Folder containing the saved figures we generated
├── **Topic_and_Author_Network.ipynb**      # Jupyter notebook having all the code for our project
├── A Network Tour of NIPS Papers.pdf       # Our report
├── environment.yml                         # Environment file for Anaconda
└── README.md                               # This file

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published