348 research outputs found
The Phase Space Dynamics of Neuronal Systems: I, Model and Experiments.
No full textWe investigate the phase space dynamics of local systems of biological neurons in order to deduce the salient computational characteristics of such systems. In this rst report, we develop an abstract physical system that models local systems of spiking biological neurons. The system is based on a limited set of realistic assumptions and in consequence accommodates a wide range of neuronal models. Simulations of the model demonstrate that the dynamical behavior of the system is akin to that observed in neurophysiological experiments. In an upcoming report, we shall demonstrate that the dynamics of the model exhibits the classic characteristics of a chaotic system, namely, contraction, expansion, and folding. We view this research as a rst step towards understanding the basis for symbolic computation in the brain.Technical report DCS-TR-34
Initializing neural networks using decision trees
No full textExisting approaches to the inductive learning problem include Symbolic and Connectionist algorithms. While the Symbolic approach is generally found to run significantly faster during learning, the Connectionist algorithms are often more accurate at classifying novel examples in the presence of noisy data. This paper presents a technique that determines the topology and initial weightsof a neural network using a decision tree, thus combining both approaches. Experimental results on benchmark real-world datasets indicate that this technique outperforms the above mentioned approaches both in efficiency and accuracy.Technical report lcsr-tr-22
On The Phase Space Dynamics of Neuronal Systems: Model, Experiments, and Analysis.
No full textWe investigate the phase space dynamics of local systems of biological neurons in order to deduce the salient computational characteristics of such systems. We develop an abstract physical system that models local systems of spiking biological neurons. The system is based on a limited set of realistic assumptions and in consequence accommodates a wide range of neuronal models. An appropriate instantiation of the system is used to simulate the dynamics of a typical column in the neocortex. The results of the simulations demonstrate that the dynamical behavior of the system is akin to that observed in neurophysiological experiments. Analysis of local properties of flows in the phase space of the system reveals the classic characteristics of a chaotic system, namely, contraction, expansion, and folding. The criterion for the dynamics of the system to be sensitive to initial conditions is identified. Based on physiological parameters it is deduced that (a) periodic orbits in the region of the phase space corresponding to "normal operational conditions" are almost surely (with probability=1) unstable, (b) periodic orbits in the region of the phase space corresponding to "seizure like conditions" are almost surely stable, and (c) trajectories in the region of the phase space corresponding to "normal operational conditions" are almost surely sensitive to initial conditions. Based on these results preliminary conclusions are drawn about the computational nature of neocortical neuronal systems.Technical report DCS-TR-37
Visual attention and retinal fixation: preprocessing modules for enhanced performance in real-time vision
No full textIn systems that operate in real time, the manner in which resources are allocated often determines the success of a system or the failure thereof. The human visual system manages to operate in real time by repeating in order, processes of Fixation, Attention and Recognition. In essence, it chooses regions of interest in the visual scene and subsequently concentrates its computational resources on them. In this paper, we attempt to emulate this characteristic of the human visual system. We present operators for Visual Attention and Retinal Fixation, neural architectures that implement them, and simulation results that confirm the effectiveness of the underlying mechanisms. We also discuss the biological plausibility of our model in light of some of the available data on the functional characteristics of the visual system.Technical report lcsr-tr-23
Inductive learning of feature-tracking rules for scientific visualization
No full textNumerical simulation and scientific visualization are often used by scientists to help them understand physical phenomena. One approach taken by some visualization systems is to identify and quantify coherent features in a simulation and track their trajectories as they evolve over time. Such feature-tracking systems operate either by relying on manual (human) efforts, or by utilizing ad hoc programs embodying heuristics that are computationally expensive to use. Our research demonstrates the use of inductive learning to construct feature-tracking programs for fluid flows. Our approach uses manually generated feature trajectories as training data, and applies inductive learning to construct feature-tracking rules that can then be incorporated into a feature-tracking program. This results in a more efficient system that can match up objects across large time steps without inspecting intermediate steps. We demonstrate our approach on the problem of tracking vortices in turbulent viscous fluids.Technical report hpcd-tr-2
Experimental studies of oxygen incorporation during growth of Y-Ba-Cu-O films by pulsed-laser deposition (Abstract Only)
No full text
Novel antimony-based Heuslers with potential spintronic applications
No full textElectronic Thesis or DissertationThe Heusler alloys, or compounds to be precise, have become a well-known name among materials scientists, solid-state physicists and other researchers of the field. In this set of studies, we examined a ternary system, i.e. Fe-Ti-Sb using experimental methodology comprised of different microstructural and structural tests alongside theoretical evaluations. The results showed that there was a compound present in the system, Fe1.5TiSb, which had the structural hall marks of a Heusler, albeit with vacancies in the tetragonal positions. This meant that the compound not only was most probably ‘layered’ between half a full Heusler, but also it was paramagnetic, the compound exhibited a novel Slater-Pauling behavior. The next system studied was FexCo1-xTiSb, eventually leading to observation of a novel half Heusler compound, Fe0.5Co0.5TiSb. The compound showed a C1b signature and no sign of any other phase were witnessed microstructurally. Magnetometry and transport measurements revealed the nature of the material to be an antiferromagnetic material with possible semiconductivity. Theoretical evaluations further validated the possibility of an antiferromagnetic coupling between iron and cobalt in this compound. Following the discovery of the Fe1.5TiSb, colleagues in our theory group encouraged us to look into another possible ‘layered’ Heusler, Co1.5TiSn, based on the former compound’s prototype. The experiments not only found that the Co1.5TiSn has all the hallmarks of a Heusler, but also magnetometry tests, especially the interesting evaluation of a Rhodes-Wohlfarth model, led to conclusion that the compound has half-metallic characteristics. The bandstructure calculated through first-principles methods also initially predicated this half metallic nature
- …
