Authors: Josh Hyman, Mark Hansen, Eric Graham, Deborah Estrin
Published in: ImageSense Workshop of SenSys (bibtex)
Abstract: Relative spectral reflectance is an illumination invarient image feature that is related to many ecological phenomena that are difficult to measure, such as plant CO2 uptake. We describe a procedure to estimate the relative spectral reflectance of known subject using color image features. Through application, we show that this procedure produces accurate estimates in the presence of changing field conditions. Using this procedure, we can use imagers as sensors to measure natural phenomena that cannot be easily measured using any other available sensing modality.

Other resources:
: Presentation
Authors: Josh Hyman (Thesis Proposal)
Submitted to: UCLA Computer Science Deptartment (bibtex)
Abstract: There exist many natural phenomena where direct measurement is either impossible or extremely invasive. We propose using imagers as sensors by constructing a procedure that uses images to obtain approximate measurements of these phenomena. This procedure, composed of state-of-the-art computer vision, image processing, and statistical learning algorithms, will be evaluated in the context of a specific application and shown to be general through multiple instantiations. We show through application, that many of these algorithms make unacceptable assuptions about their input. We will describe a methodology that can be used to augment existing algorithms, making them robust to field conditions present in ecological applications. In this paper, we rigorously define the proposed procedure and begin to evaluate its accuracy in the context of an example application.

Authors: Josh Hyman and Dustin Boswell (bibtex)
Abstract: TaskMaster is a system for managing priority-ordered queues that is designed to scale to 1 billion tasks across 100 thousand queues per node. A reliable queuing system, such as TaskMaster, provides a mechanism for distributing units of inherently serial work (tasks) to workers. Priorities are lexicographically ordered strings that give users more power than FIFO or fixed-range integer priorities when defining queue order. TaskMaster provides the ability to atomically transition a task between queues, facilitating the decomposition of processing into a sequence of tasks. Individual tasks are exclusively leased to workers for the duration of processing to minimize system-wide work duplication. TaskMaster helps system designers detect bottlenecks and direct system optimization by providing queue statistics such as enqueue and dequeue rates, as well as queue lengths. TaskMaster allows application designers to focus on the individual stages of data processing instead of task distribution.

Other resources:
: Presentation
: Poster (PDF)
: Alternate paper link
: Alternate poster link
Authors: Josh Hyman, Eric Graham, Mark Hansen, and Deborah Estrin
Published in: Workshop on Data Management in Sensor Networks (bibtex)
Abstract: There exist many natural phenomena where direct measurement is either impossible or extremely invasive. To obtain approximate measurements of these phenomena we can build prediction models based on other sensing modalities such as features extracted from data collected by an imager. These models are derived from controlled experiments performed under laboratory conditions, and can then be applied to the associated event in nature. In this paper we explore various different methods for generating such models and discuss their accuracy, robustness, and computational complexity. Given sufficiently computationally simple models, we can eventually push their computation down towards the sensor nodes themselves to reduce the amount of data required to both flow through the network and be stored in a database. The addition of these models turn in-situ imagers into powerful biological sensors, and image databases into useful records of biological activity.

Authors: Josh Hyman (bibtex)
Abstract: Image databases attempt to allow users to effectively search through large sets of images. These images are typically captured intentionally by a human (vacation pictures, etc) or automatically by cameras attached to sensors (security camera in an airport, etc). As with all other searchable end-to-end database systems, image databases need the following facilities: storage, indexing, querying, and result display. In this survey, I will focus on the method by which image databases can be queried as well as all supporting technologies such as feature extraction, query matching, and ranking.

Authors: Sasand Reddy, Andrew Parker, Josh Hyman, Jeff Burke, Deborah Estrin, Mark Hansen
Published in: Workshop on Embedded Networked Sensing (bibtex)
Abstract: Imagers are an increasingly significant source of sensory observations about human activity and the urban environment. ImageScape is a software tool for processing, clustering, and browsing large sets of images. Implemented as a set of web services with an Adobe Flash-based user interface, it supports clustering by both image features and context tags, as well as re-tagging of images in the user interface. Though expected to be useful in many applications, ImageScape was designed as an analysis component of DietSense, a software system under development at UCLA to support (1) the use of mobile devices for automatic multimedia documentation of dietary choices with just-in-time annotation, (2) efficient post facto review of captured media by participants and researchers, and (3) easy authoring and dis- semination of the automatic data collection protocols.

Authors: Josh Hyman, Mohit Lad, Lixia Zhang (bibtex)
Abstract: In this paper, we take an objective look at the usefulness of visualization for the purpose of security and study the current state of the art. We first argue that visualization is a critical component to counter the unpredictable nature of security breaches. We also argue that data abstraction and correlating different data sets could open up new frontiers and foster innovation in security visualization.

Authors: Josh Hyman (bibtex)
Abstract: In large clusters, it seems that many resources (such as memory, disk, CPU cycles, network I/O, etc) are wasted. I have begun to explore one possible reason related to users over estimating their process' needs. For this project I will look into dynamically scheduling "small" tasks to use the remaining available resources in the cluster, though fragmented across machine boundaries. This problem is very similar to the problem addressed by Condor -- trying to use spare CPU cycles on idle workstations for useful work. If such "small" tasks could effectively use these fragmented resources, then the overall resource utilization in the cluster would increase. The goal of this project is to define the required properties of a "small" tasks and a policy for statically and dynamically scheduling them on a cluster.