Applications of Machine Learning for Networking Research
| dc.contributor.advisor | Ramakrishnan Durairajan Dr. Walter Willinger | en |
| dc.contributor.advisor | ||
| dc.contributor.advisor | ||
| dc.contributor.author | Knofczynski, Jared | |
| dc.date.accessioned | 2022-09-28T22:53:55Z | |
| dc.date.available | 2022-09-28T22:53:55Z | |
| dc.date.issued | 2022-02 | |
| dc.description | 51 pages | en |
| dc.description.abstract | The application of machine learning (ML) to mitigate network-related problems continues to pose significant challenges for researchers and operators alike. For one, there is a general lack of labeled training data in networking communities, and labeling techniques popular in other domains are ill-suited due to the scarcity of operators’ domain expertise. Additionally, networking issues are typically multi-tasked in nature, requiring the development of multiple ML models (one per task) and resulting in multiplicative increases in training times as the number of tasks increases. To address these challenges, we propose Arise, a multi-task weak supervision framework for network measurements. Arise uses weak supervision-based data programming to label network data at scale and applies multi-task learning (MTL) to facilitate information sharing between tasks as well as reduce overall training time. Using community datasets from the Center for Applied Internet Data Analysis (CAIDA), we show that Arise can create MTL models that demonstrate improved classification accuracy and reduced training times when compared to multiple single-task learning (STL) models. | en |
| dc.identifier.uri | https://hdl.handle.net/1794/27533 | |
| dc.language.iso | en | en |
| dc.publisher | University of Oregon | en |
| dc.rights | Creative Commons BY-NC-ND 4.0-US | en |
| dc.subject | Math and Computer Science | en |
| dc.subject | Machine Learning | en |
| dc.subject | Networking | en |
| dc.subject | Data Science | en |
| dc.title | Applications of Machine Learning for Networking Research | en |
| dc.type | Thesis / Dissertation | en |