ITC/USA 2005 Conference Proceedings / The Forty-First Annual International Telemetering Conference and Technical Exhibition / October 24-27, 2005 / Riviera Hotel & Convention Center, Las Vegas, Nevada / Often it is necessary to retrieve segments of video with certain characteristics, or features, from a large archive of footage. This paper discusses how image processing algorithms can be used to automatically create a relational database, which indexes the video archive. This feature extraction can be performed either upon acquisition or in post processing. The database can then be queried to quickly locate and recover video segments with certain specified key features

No description available.

No description available.

Prior work on command recommendations for feature-rich software has relied on data supplied by a large community of users to generate personalized recommendations. In this work, I explored the feasibility of using an alternative data source: web documentation. Specifically, the proposed approach uses QF-Graphs, a previously introduced technique that maps higher-level tasks (i.e., search queries) to commands referenced in online documentation. The proposed approach uses these command-to-task mappings as an automatically generated plan library, enabling our prototype system to make personalized recommendations for task-relevant commands. Through both offline and online evaluations, I explored potential benefits and drawbacks of this approach.

No description available.

No description available.

We propose a scalable and fexible hardware architecture for the extraction of image features, used in conjunction with an attentional cascade classifier for appearance-based object detection. Individual feature processors calculate feature-values in parallel, using parameter-sets and image data that is distributed via BRAM buffers. This approach can provide high utilization- and throughput-rates for a cascade classifier. Unlike previous hardware implementations, we are able to flexibly assign feature processors to either work on a single- or multiple image windows in parallel, depending on the complexity of the current cascade stage. The core of the architecture was implemented in the form of a streaming based FPGA design, and validated in simulation, synthesis, as well as via the use of a Logic Analyser for the verification of the on-chip functionality. For the given implementation, we focused on the design of Haar-like feature processors, but feature processors for a variety of heterogenous feature types, such as Gabor-like features, can also be accomodated by the proposed hardware architecture.

No description available.

Domain-independent feature learning is a hard problem. This is reflected by lack of broad research in the area. The goal of General Game Playing (GGP) can be described as designing computer programs that can play a variety of games given only a logical game description. Any learning has to be domain-independent in the GGP framework. Learning algorithms have not been an essential part of all successful GGP programs. This thesis presents a feature learning approach, GIFL, for 2-player, alternating move games using state differences. The algorithm is simple, robust and improves the quality of play. GIFL is implemented in a GGP program, Maligne. The experiments show that GIFL outperforms standard UCT algorithm in nine out of fifteen games and loses performance only in one game.

The goal of this thesis is to study and model the Diel Vertical Migration (DVM) pattern using machine learning methods. We choose an Almost Periodic Function as the mathematical model and fit the monthly averaged migration data into a 5-term Fourier series whose coefficients and frequency are functions of time. The resulting function captures the general characteristics of the DVM pattern whose period is similar yet undergoes gradual changes over time. Further correlation analyses show that the monthly averaged distribution of zooplankton and various environmental factors are strongly correlated. Therefore, we adjust the function so that the coefficients and frequency are functions of environmental factors. Besides, we also examine the pattern on finer time scales using classification algorithms. We build classifiers which predict zooplankton existence at different depths based on a set of environmental measurements. Experiments demonstrate that both of the above methods are valid in modeling the DVM pattern.