Contact Information

Computer Science Department
Colgate University
McGregory Hall, 3rd Floor
13 Oak Drive
Hamilton, NY 13346
(tel) 315.228.7719
Charlotte Jablonski, Administrative Assistant
cjablonski@colgate.edu

Upcoming events

  • No upcoming events.

Past events

  • 3

    May

    2018

    Join us in celebrating our seniors and wishing them well in their future endeavors! Senior gifts will be distributed at this time.

  • 1

    May

    2018

    Knowledge Discovery From Sensitive Data: Privatizing Bayesian Rule Lists
    Armando Belardo '18
    The utility of machine learning is rising, coming from a growing wealth of data and problems that are becoming harder to solve analytically. With these changes there is also the need for interpretable machine learning in order for users to understand how a machine learning algorithm comes to a specific output. Additionally, with the amount of data collected today, there is a lot of potentially sensitive data that we can learn from. However, to do so, we must guarantee a degree of privacy on the dataset for which we use differential privacy to protect the underlying individuals. Our research presents a differentially privatized version of the interpretable machine learning algorithm Bayesian Rule Lists, along with proofs of privacy and experiments showing its utility.

    Autocompletion for Network Configurations
    Ahsan Mahmood '18
    Human factors are responsible for 50% to 80% of network outages. Yet, compared to software developers, network operators are often left neglected when it comes to development tools for writing network configurations, which could help reduce human errors. Our work tries to bridge that gap by providing a completion engine that could be incorporated into a more extensive tool. In my thesis, we propose a simple yet powerful model inspired by code completion techniques and Natural Language Processing research. We believe our engine is a strong first step in creating a holistic tool similar to IDEs that can assist network operators. In this talk, I will outline how we developed this engine and discuss how our analyses show that despite challenges, the current state of the model gives us encouraging results.

Events history