Superb-CSIS 2011 Participants
|
Jasmine Berry
Project Title: Ethernet Controller for Host Communication with Energy-Efficient Microprocessor Final Report: Ethernet Controller for Host Communication with Energy-Efficient Microprocessor |
|
| Abstract: As the chips have grown bigger over time, it has become more difficult to test them. The most common approach is to use programmable logic arrays (FPGAs) to send test information from the host station (PC) to the chip and return the chip reply back to the host. This project will investigate such communication with an extremely energy-efficient microprocessor via ethernet connection. The goal is to provide fast synchronization with the chip, and establish robust protocol that can be re-used for future implementations. | ||
|
Graduate Student Mentor: Ruzica Jevtic | ||
|
|
||
|
Grace Christenbery
Project Title: Visualizing Online RNA-Seq Abundance Estimation Final Report: Visualizing Online RNA-Seq Abundance Estimation |
|
| Abstract: RNA-Seq is a new high-throughput methodology leveraging modern DNA sequencing technology to measure the abundance of different RNA transcripts in a sample. Biologists wanting to use this technology to measure gene expression levels require analysis tools that can estimate abundances at the isoform level using ambiguous mappings of sequencing reads. Many tools for this type of analysis (including Cufflinks, which was developed by the Pachter lab) use the batch EM algorithm to achieve this goal. However, as datasets become larger, a faster solution is needed. We have recently begun to develop CuffExpress, which uses an online version of the EM algorithm of Cufflinks. As we enter the testing phase, we need to determine the convergence properties of the algorithm as sequencing reads are processed. To do so, we will simulate sequencing reads and plot the change in parameters as they are observed. After our analysis of the algorithm is complete, a GUI and visualization tool will be built in order to display these results on actual data in real-time for users of the software. | ||
|
Graduate Student Mentors: Adam Roberts and Harold Pimentel | ||
|
|
||
|
Jonathan Harper
Project Title: Efficient Graph Algorithms in Python Final Report: Efficient Graph Algorithms in Python |
|
| Abstract: Many computations of interest involve graph algorithms, which perform operations on an arbitrary graph of vertices and edges connecting those vertices. Such algorithms are used in a wide variety of domains, from Google's PageRank algorithm to the analysis of friend relationships in social networks. In order to program them productively and with high performance, we are building a graph specializer for Python, which translates high-level graph algorithm code written in Python into parallel low-level code to run efficiently on multicore machines at scales that are impossible in pure Python. The project involves creating a Pythonic graph algorithm API, and then leveraging existing technologies from the Asp project to translate instances of the API into efficient, parallel code. | ||
|
Graduate Student Mentor: Shoaib Kamil | ||
|
|
||
|
Jasmine Jones
Project Title: Developing Tools for Sharing User Research and Concepts in Product Design Teams Final Report: Developing Tools for Sharing User Research and Concepts in Product Design Teams |
|
| Abstract: Product designers contribute their individual user research and concepts to those of their team, in order to develop them further and create a shared understanding of the design problem. However, sharing more rich individual information could lead to better shared understanding, it is also important to protect the rest of the team from information overload. This project will investigate ways to leverage cloud computing and mobile phones to support sharing individual design information during co-located design team meetings. The goal is to develop a system that allows product designers to work with both tangible and digital information benefit shared understanding and fit seamlessly fit into product design practice. | ||
|
Graduate Student Mentor: Lora Oehlberg | ||
|
|
||
|
Linda Lee
Project Title: Energy Management in Buildings Final Report: Energy Management in Buildings |
|
| Abstract: Buildings consume 40% of all energy used in the United States and over 70% of all electricity. Commercial buildings, such as buildings on campus, contain a wealth of sensors and controls that can be accessed programmatically. Using these controls energy consumption can be reduced and buildings can be made into controllable energy storage resources to enable increased use of renewable energy. This project will investigate an online lighting and AC control system in Sutardja Dai Hall. The goal will be to analyze data and produce control schemes that reduce energy consumption while maintaining user comfort | ||
|
Graduate Student Mentor: Andrew Krioukov | ||
|
|
||
|
Rebecca Pottenger
Project Title: Privacy-Preserving Data Mining Final Report: Privacy-Preserving Data Mining |
|
| Abstract: As the growth of the Internet makes access to personal data more straightforward, the issue of maintaining an individual's privacy is becoming a critical concern. The recent and ongoing issue with online privacy violations further indicate the growing awareness that there is a definite need for a way to ensure individual privacy, while still allowing for advertisers to have access to necessary data in order to effectively market products to clients. To that end, this project will extend upon the student's prior research. The student has already performed research that combines differential privacy, online learning and topic modeling to demonstrate a practical implementation of differential privacy on a real world dataset of web browsing activity. This previous work explored a new methodology for the establishment of sensitivity bounds using empirical experimentation, and used this methodology to establish a bound on the sensitivity of the online Latent-Dirichlet algorithm. The goal of this project is to expand upon this work by determining a theoretical bound on the sensitivity of the Latent-Dirichlet algorithm, based on rigorous mathematical analysis. Upon determining this theoretical bound, the student will determine what, if any, changes could be made to the previously established differentially private LDA system, and compare the results thereof. | ||
|
Graduate Student Mentors: Elaine Shi and Ivan Martinovic | ||
|
|
||
|
Virginia Smith
Project Title: Mathematical Modeling of Building-Wide HVAC Systems Final Report: Mathematical Modeling of Building-Wide HVAC Systems |
|
| Abstract: Heating, ventilation, and air conditioning (HVAC) systems represent a significant portion of the electrical energy usage in the developed world. They are an important target for improvements in energy efficiency, and they may be used to develop energy-agile systems that can help integrate renewable energy sources into the power grid. One step towards these goals is the identification of a mathematical model that enables temperature and energy control for multiple objectives. The goal of this project is to use experimentally measured data to develop such a model for a building on the Berkeley campus. | ||
|
Graduate Student Mentor: Anil Aswani | ||
|
|
||
|
Amanda Stowers
Project Title: Airplane for Sensor Distribution Final Report: Airplane for Sensor Distribution |
|
| Abstract: In order to gather data for surveillance or search and rescue missions, it is useful to have sensors which are capable of communicating with one another as well as vehicles. This project aims to drop sensors at particular GPS coordinates, using a small unmanned aircraft, specifically a modified RC plane. These sensors will then be capable of communicating with each other, and later sensors and vehicles which come into the area. This project will involve writing drivers such that the GPS can communicate with the GINA boards, which will be used both as sensors and hopefully as a controller for the airplane. It will also be necessary to design and implement the controllers (initially remotely on a computer, and hopefully eventually on the microcontroller) on the airplane, as well as develop a launching mechanism for the sensors themselves. The ideal end result would be that the airplane flies autonomously using the microcontroller, launching sensors at designated GPS points | ||
|
Graduate Student Mentor: Ankur Mehta | ||