Michael Buro's 2020-2021 MSc Projects
[2020 and 2021 summer funding is secured]
Project 1
---------
Topic: Learning Motion Tracking via Traffic Simulations
Description:
Tracking mobile objects is crucial for effective traffic navigation.
Humans excel at this task and autonomous driving will require it.
In this project we will investigate how to efficiently track objects
by means of supervised learning based on video footage generated by
traffic simulators.
Prerequisites:
- Machine learning expertise - in particular deep-network learning
- Familiarity with Tensorflow, Python, and C++
- Computer vision background will also help
=============================================================================
Project 2
---------
Topic: Applying Machine Learning to Variable Selection in SAT Solvers
Description:
Many important real-world combinatorial problems can be reduced to
determining whether a propositional formula has a satisfying variable
assignment. Satisfiability (SAT) solvers have come a long way in
solving large formulas, but haven't used modern machine learning
techniques much.
In this project we will study how deep-neural network learning can be
used to select split-variables more effectively - which has the potential
to speed-up SAT solvers considerably.
Prerequisites:
- Machine learning expertise - in particular deep-network learning
- Familiarity with Tensorflow, Python, and C++
- Theoretical CS background will also help