Research Project #1: Bridge
This is an interdisciplinary project, involving both electrical and computer engineering as well as psychology at the University of Rochester. The goal of this project is to automatically classify emotion of a speaker based on vocal features (e.g., pitch, energy, formants). Our system uses a Support Vector Machine (SVM) machine learning algorithm to train various classifiers (e.g., the sample is "angry" or "not angry"; it is "happy" or "not happy", etc.) Then, the outputs of these individual classifiers are fused together to obtain the final emotion classification. While the current system works well on a database of "acted" emotions, we need to experiment with both real data as well as with noisy data obtained in real-world situations (e.g., via a cell phone). We are looking for a student who can work with Matlab code, run experiments and analyze data to determine how well our current system works in new environments and to develop new approaches to improve the emotion classification using real and noisy speech samples.
Research Project #2: GENIUS
Wireless sensor networks have the potential to revolutionize how we interact with the physical world. However, such networks require extremely energy-efficient operation while meeting the required goals of the application. We have developed a suite of passive wake-up radios to wake up wireless sensor "motes" when a sink is ready to receive their data. This can save a considerable amount of energy in the wireless sensor mote. We are looking for a student to help us refine our devices and experiment with their abilities, both in hardware implementation as well as software simulations that use the experimental results to create models for the wake-up radios.
Research Project: Wireless Building Power Monitoring
The goal of this continuing project is to develop an on-line wireless-based electric power monitor for building applications. The system measures AC electrical current, voltage, power, and phase, logs this data, and transmits it by a wireless channel to a server facility for storage, on-demand display, and analysis. The monitor is intended for installation and continuous monitoring of circuits which have an excessive load or power quality problems needing studied. The unit is being designed for easy installation and removal, and the wireless data transmission feature makes it possible to view collected data collect on the web. Considerable progress has been made in the past year.
We have a working prototype that transmits power data to a format accessible anywhere on campus from a browser. Experience with this first prototype has provided the guidance essential to the design of a new system that will work faster and more reliably. The new device is now in the final stages of the design and new prototypes will be assembled in the Spring of 2013.
The next significant task in the project will be to connect the wireless system into a cloud computing architecture. This work, which will demand significant programming skills, has not yet begun. There are intriguing advantages to storing the data in a distributed database. For one thing, the power monitor will be producing lots of data over irregular periods of time, so that there will be the need for flexibility and adaptability that cloud-based computing offers. An additional advantage of cloud computing stems from the fact that a wide range of analyses, many not yet anticipated, will be implemented to take advantage of the data as it becomes available. The cloud should make it possible for multiple users to gain access to the data so that they can perform unique analyses and create their own presentation materials based on their work. The Summer 2013 project for the Xerox Fellow will be focus on establishing and testing a cloud-based data collection system. Issues such as data security will be considered.
Computer Science majors (and ECE majors with strong, advanced programming skills) are encouraged to consider this project.Project advisers are T. B. Jones firstname.lastname@example.org) and W. Heinzelman (email@example.com).
Research Project: CSI Eastman: Image Processing based Characterization of Photolytic Degradation of Daguerreotypes
Introduced in 1839, the daguerreotype became the first commercially viable medium for photographic image capture and served as the primary photographic medium of record for over a decade. The history captured in daguerreotypes provides a unique detailed photorealistic record, which is of interest to social scientists and the general public alike. The recent discovery that daguerreotypes are subject to photolytic degradation, i.e. degradation under exposure to light, has, however, alarmed museums and the majority of daguerreotypes are no longer available for public display because of concerns about light induced damage. The George Eastman House (GEH) has one of the largest collections of daguerreotypes and is actively involved in efforts to analyze and prevent degradations in this medium. In this joint project with GEH, the goal is to characterize photolytic degradations produced by exposure to different wavelengths of light by using image processing on photomicrographs of daguerreotypes and to then use the characterization as a forensic tool to characterize degradations in actual collections.