This thesis introduces the Named-State Register File, a fine-grain, fully-associative register file. The NSF allows fast context switching between concurrent threads as well as efficient sequential program performance. The NSF holds more live data than conventional register files, and requires less spill and reload traffic to switch between contexts. This thesis demonstrates an implementation of the Named-State Register File and estimates the access time and chip area required for different organizations. Architectural simulations of large sequential and parallel applications show that the NSF can reduce execution time by 9% to 17% compared to alternative register files.

This thesis consists of four chapters. In this thesis we explored associative plasticity of human motor cortex with the use of noninvasive transcranial magnetic stimulation (TMS). Paired Associative Stimulation (PAS) has grown in popularity because of its potential clinical applications. We used TMS techniques in combination with electromyographic (EMG) measurements to study cortical excitability and kinematic features of arm movement. This work has focused in a cohesive approach to answer certain fundamental questions about a) the rules of cortical plasticity and mechanism of PAS, b) the interaction between the state of neuronal excitability at the targeted cortical network and the effects of PAS, and c) translation of these effects into obvious measurable kinematic changes starting from network level changes and ending up with the behavioral modulation of arm movement. First we explored the role of GABAergic intracortical networks and intracortical facilitation on modulation of cortical excitability by showing for the first time that PAS can be conditioned by these inhibitory and facilitatory intracortical networks. Next, using standard indirect approaches utilizing peripheral EMG measures, we showed a graded excitability response for the PAS technique and showed that interactions of PAS with motor learning depends on the degree as well as the state of cortical excitability. Rules governing the interactions of brain stimulation techniques and motor learning are important because brain stimulation techniques can be used to modify, improve or disrupt motor adaptation and skill learning with great potential for clinical applications such as facilitation of recovery after stroke. TMS provide us with a unique opportunity to study the rules of plasticity at a systems level, which is a combination of synaptic and nonsynaptic (metaplastic) changes. These changes can occur either in the direction to limit the physiological range of neuronal functioning (homeostatic) or against the direction established state of neurons.

No description available.

During the past decade, a new class of knowledge representation has emerged known as structured distributed representation (SDR). A number of schemes for encoding and manipulating such representations have been developed; e. g. Pollack's Recursive Auto-Associative Memory (RAAM), Kanerva's Binary Spatter Code (BSC), Gayler's MAP encoding, and Plate's Holographically Reduced Representations (HRR). All such schemes encode structural information throughout the elements of high dimensional vectors, and are manipulated with rudimentary algebraic operations. <br /><br /> Most SDRs are very compact; components and compositions of components are all represented as fixed-width vectors. However, such compact compositions are unavoidably noisy. As a result, resolving constituent components requires a cleanup memory. In its simplest form, cleanup is performed with a list of vectors that are sequentially compared using a similarity metric. The closest match is deemed the cleaned codevector. <br /><br /> While SDR schemes were originally designed to perform cognitive tasks, none of them have been demonstrated in a neurobiologically plausible substrate. Potentially, mathematically proven properties of these systems may not be neurally realistic. Using Eliasmith and Anderson's (2003) Neural Engineering Framework, I construct various spiking neural networks to simulate a general cleanup memory that is suitable for many schemes. <br /><br /> Importantly, previous work has not taken advantage of parallelization or the high-dimensional properties of neural networks. Nor have they considered the effect of noise within these systems. As well, additional improvements to the cleanup operation may be possible by more efficiently structuring the memory itself. In this thesis I address these lacuna, provide an analysis of systems accuracy, capacity, scalability, and robustness to noise, and explore ways to improve the search efficiency.

During the past decade, a new class of knowledge representation has emerged known as structured distributed representation (SDR). A number of schemes for encoding and manipulating such representations have been developed; e. g. Pollack's Recursive Auto-Associative Memory (RAAM), Kanerva's Binary Spatter Code (BSC), Gayler's MAP encoding, and Plate's Holographically Reduced Representations (HRR). All such schemes encode structural information throughout the elements of high dimensional vectors, and are manipulated with rudimentary algebraic operations. <br /><br /> Most SDRs are very compact; components and compositions of components are all represented as fixed-width vectors. However, such compact compositions are unavoidably noisy. As a result, resolving constituent components requires a cleanup memory. In its simplest form, cleanup is performed with a list of vectors that are sequentially compared using a similarity metric. The closest match is deemed the cleaned codevector. <br /><br /> While SDR schemes were originally designed to perform cognitive tasks, none of them have been demonstrated in a neurobiologically plausible substrate. Potentially, mathematically proven properties of these systems may not be neurally realistic. Using Eliasmith and Anderson's (2003) Neural Engineering Framework, I construct various spiking neural networks to simulate a general cleanup memory that is suitable for many schemes. <br /><br /> Importantly, previous work has not taken advantage of parallelization or the high-dimensional properties of neural networks. Nor have they considered the effect of noise within these systems. As well, additional improvements to the cleanup operation may be possible by more efficiently structuring the memory itself. In this thesis I address these lacuna, provide an analysis of systems accuracy, capacity, scalability, and robustness to noise, and explore ways to improve the search efficiency.

Thèse de doctorat : Sciences du langage : Nancy 2 : 2003. / Bibliographie.

The cockroach, with its large brain and physiological resilience, holds many advantages for the development of behavioral paradigms. The work presented here provides a foundation for, and describes the results of, the implementation of studies of neural correlates of learning and memory on restrained animals.Using the antennal projection response (APR) as an indicator of learning and retention, several learning paradigms have been developed. A visual-olfactory associative and a gustatory-olfactory aversive conditioning paradigm demonstrated a plastic behavior that could be driven in an intact and immobilized cockroach. Conditioning the APR to a visual cue paired with an olfactory cue characterized the role of unilateral and bilateral olfactory input in learning and memory. While unilateral olfactory input is sufficient to learn a visual-olfactory association, bilateral olfactory input is necessary for long-term retention of the association. This comparison identified a critical time period in which memory is consolidated. This time period was subsequently used to analyze gene expression during memory consolidation.The split-brain cockroach preparation was developed to investigate what parts of the brain are necessary and sufficient for learning and retention of a visual-olfactory association; this preparation was also used to examine learning-induced changes in test tissue versus control tissue provided by the same animal. Evidence suggests that half of a brain is sufficient for a visual-olfactory association to be established and sufficient for retention of that association between 12 and 24 hours. However, the entire brain is necessary for long-term memory to be established. Using the split-brain cockroach simultaneously as the control and the test subject, learning-induced alterations in the microglomerular synaptic complexes of the calyces were identified in the trained half, but not in the naïve half.Using the APR, spatial learning and memory was examined. Multiple representations of space were revealed in the brain of the cockroach. Cockroaches represent space in terms of an olfactory gradient map, as well as the visuospatial relationship between objects. When both representations of space can be utilized by the cockroach to localize a cue, the positional visual cue is the one that determines the behavioral response.

Imaging has revealed that brain activation of verbs with verifiable products (‘throw, kick’) activate language areas as well as the motor cortex responsible for the performance of the action described. An exploratory comparison of eye related verbs with no verifiable products (‘observe’) to mouth related verbs with verifiable products (‘shout’) has revealed a similar activation pattern. Thus in order to further study mental action verbs with no verifiable products, the present two-part study used words that were suitable across two modalities (e.g. you can ‘perceive’ both through vision and audition) and compare them to themselves under differing contexts of auditory and visual verbs so as to eliminate any word characteristics differences, as well as explored the two modalities directly. The primary purpose was to delineate whether associative learning or the mirror systems theory might better account for the acquisition of this unique subclass of verbs. Results suggest that Mirror systems theory more likely accounts for the observed cognitive processing differences between the two verbs. Keywords: Verbs, language, Event-related potentials, abstract, associative learning theory, mirror systems theory.

This study explores compositional abstraction in architecture. The act of squinting adopts propositions from Christopher Alexander and structuralist Marcel Mauss. An analytical method based upon element density and regularity is tested through residential facade studies. Observed limitations prompt the introduction of a third axis of exceptionality. Generative possibilities are investigated through a series of facade games. Focus turns to the facades of parking garages as a neutral background for the examination of compositional qualities. Notable garages are examined, and local design guidelines are critiqued. Finally, a garage in Halifax, Nova Scotia is redesigned. Compositional ambiguities inherent in the existing confi guration prompt the introduction of an alternative, associatively rich diaphragm system based upon the compositional and structural logic of trees. A review of the associative method notes its value as an explicit decision-making tool, but suggests that key formal moves remain beyond the generative scope of organizational modelling.

No description available.