In order to utilize the machine learning to build a self-operating network, we need a dataset to purposely train ML models to capture network features and behaviors. This directed study will guide students with background of computer network to automatically create such a dataset of computer networks with various features so that these data can be trained and encoded to eventually predict network issues or generate desired networking features.
This would eventually help us to build a self-operating network, that should have the capability of network automation in three phases of plan, build and run in the life cycles of the network infrastructure. Various open source and commercial tools such as GNS3, BatFish, Metha, D3.js, Forward Networks will be used and the programming in Python, C/C++ and JavaScript are required to automate the network creation, data collection and visualization.
Once a dataset is ready, we could utilize the ML models, to train on the obtained dataset and eventually build up capabilities for network prediction and generation. This section of the research might requires students to have knowledge of machine learning related course works.
The research project requires the student to work at about 4 hours per week (equivalent to 1 credit) during the semester including meeting, discussions, self-paced study and programming. The source code and documents will be committed in an open source project. The final delivery also include a technical report that can be a foundation to turn any innovation into a manuscript and submit to a technical conference.
A more detailed roadmap would be continuously developed and used to guide this study.
Notes: EECS 399/499 is the Direct Study course for undergraduate students while EECS 599/699 is the same course but for the graduate students. This topic is under 399/499 - 235. The students need to use the EECS independent research form on the CSE Undergraduate Research page to enroll this course.
4 hours per 1 credit enrolled, including at least 1 hour meeting time.
Brief description of your project: A: see above paragraph.
How will you be evaluated? A: Monthly oral evaluation based on student’s weekly progress report and a final written evaluation.
Will materials from other classes you have taken be used in the project? A: Yes but should be discussed to establish the baseline.
How often will you meet with your faculty mentor? A: one or two weekly meetings based on how many hours/credits.
How will the completion of your project be determined? A: Two parts of the deliveries are expected: (1) The written report or manuscript for a conference submission reporting the background, progress made and lesson learned and/or achievements accomplished. (2) A public repository of open source code and data to share the work with the community.