We use machine learning and control/optimization techniques, based on network measurement and user behavior data, to address network resource allocation and management challenges. One project is on prediction-assisted dynamic bandwidth allocation mechanisms in wireline networks. Another project is on analyzing real network performance data for better cellular network configurations. The third project is to conduct experiments on a remote wireless testbed, Orbit.
A GSR position is available in Professor Ravani’s group for MS or PhD students in Computer Science for the above titled research project.
We have made significant discoveries in finding genes contributing to neurodevelopmental disorders. However, each gene can be considered as several domains with various intolerance levels for mutations. We want to develop a novel computational method to discover domains that are intolerant to missense mutations, utilizing the available large-scale sequencing projects. Furthermore, we will try to study these domains in the context genetic networks to find potential domain-domain interactions that are crucial for neurodevelopment.
We have access to 2 data sets:
The goal of this project is to determine the effect of anonymization on signature-based intrusion detection analysis. First, run Bro on both sets of data to establish a baseline of alerts from the data. Then, anonymize the packets in each data set using black marker anonymization, and rerun the analyses. Record the differences between the results and the baseline (false positives and false negatives). Repeat this with prefix-preserving anonymization. Then, explain any differences between the analyses of the anonymized data and the raw data sets.
Not only is hardware technology changing rapidly, but the workloads that execute on this hardware are also evolving. This set of projects will extend the open-source architecture simulator gem5 to simulate systems in their native environment with close to zero overhead so computer system designers in industry and academia can evaluate new hardware on real applications. gem5 is used by many different universities and in industry including at ARM, Google, and AMD. If you want to work on a vibrant open-source project with real impact on research and industry, I’d be happy to work with you! These projects will give you experience working on a large distributed-development C++ open-source application.
See my research page for more information.
This is a project funded by the California Department of Water Resources for effective modeling of the groundwater and surface water resources of California. We will use Intel’s VTune Amplifer for a performance profile of IWFM to identify where in the code time is being spent in both serial and threaded applications.
The timeline of this project is from January 1 to June 30, 2018. This project is funded and can support one Master’s student.
We hope to construct simplified geometric models of subcellular components of a cell, like the nucleus or mitochondria. The input will be high resolution Focused Ion Beam Scanning Electron Microscope (FIB-SEM) images of plane sections through the cell, which are assembled into a 3D voxelized volume. These voxel volumes have already been segmented to identify the voxels belonging to the subcellular objects of interest. The output will be a simplified geometric approximation to the voxel volumes in these objects, bounded by pieces of planar, spherical, or conical surfaces. Such a simplified description can serve as a compressed summary of the volume, taking much less data to specify, and can also be used to compute things like the surface area of the object.
Initially, the project will involve applying existing published algorithms to this data to compute the approximations, which will be used to create images for grant proposals, since I have no funding to support a research student on this project. If we get a grant, there may be money to support a student to carry the research further.