Off-campus UMass Amherst users: To download dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.
Non-UMass Amherst users, please click the view more button below to purchase a copy of this dissertation from Proquest.
(Some titles may also be available free of charge in our Open Access Dissertation Collection, so please check there first.)
Data reprocessing in signal understanding systems
Signal understanding systems have the difficult task of interpreting environmental signals: decomposing them and explaining their components in terms of an arbitrary number of instances of perceptual object categories whose properties can interact with one another. This dissertation addresses the problem of designing blackboard-based perceptual systems for interpreting signals from complex environments. A "complex environment" is one that can (1) produce signal-to-noise ratios that vary unpredictably over time, and (2) can contain perceptual objects that mutually interfere with each others' signal signature, or have arbitrary time-dependent behaviors. The traditional design paradigm for perceptual systems assumes that some particular set of fixed front-end signal processing algorithms (SPAs) can provide adequate evidence for reliable interpretations regardless of the range of possible scenarios in the environment. In complex environments, with their dynamic character, however, a commitment to parameter values inappropriate to the current scenario can render a perceptual system unable to interpret entire classes of environmental events correctly. To address these problems, this research advocates a new view of signal interpretation as the product of two interacting search processes. The first search process involves the dynamic, context-dependent selection of signal features and interpretation hypotheses, and the second involves the dynamic, context-dependent selection of appropriate SPAs for extracting evidence to support the features. For structuring bidirectional interaction between the search processes, this dissertation presents the Integrated Processing and Understanding of Signals (IPUS) architecture as a formal and domain-independent blackboard-based approach. The architecture is instantiated by a domain's formal signal processing theory, and has four components for organizing and applying signal processing theory: discrepancy detection, discrepancy diagnosis, differential diagnosis, and signal reprocessing. IPUS uses an iterative process of "discrepancy detection, diagnosis, reprocessing" for converging to the appropriate SPAs and interpretations. Convergence is driven by the goal of eliminating or reducing various categories of interpretation uncertainty. This dissertation discusses the IPUS architecture's features, the basic problem of auditory scene analysis (the application domain used in testing IPUS), and evaluates performance results in experiment suites that test the utility of the reprocessing loop and the ability of the architecture to apply special-purpose SPAs effectively. Although the specific research reported herein focuses on acoustic signal understanding, the general IPUS framework appears applicable to the design of perceptual systems for a wide variety of sensory modalities.
Computer science|Acoustics|Electrical engineering
Klassner, Frank Irwin, "Data reprocessing in signal understanding systems" (1996). Doctoral Dissertations Available from Proquest. AAI9709614.