Presentation is loading. Please wait.

Presentation is loading. Please wait.

Neural Cross Correlation For Radio Astronomy Chipo N Ngongoni Supervisor: Professor J Tapson Department of Electrical Engineering, University of Cape Town.

Similar presentations


Presentation on theme: "Neural Cross Correlation For Radio Astronomy Chipo N Ngongoni Supervisor: Professor J Tapson Department of Electrical Engineering, University of Cape Town."— Presentation transcript:

1 Neural Cross Correlation For Radio Astronomy Chipo N Ngongoni Supervisor: Professor J Tapson Department of Electrical Engineering, University of Cape Town Rondebosch, 7701, South Africa Chipo.Ngongoni@uct.ac.za jonathan.tapson@uct.ac.za

2 Neural Cross Correlation For Radio Astronomy

3 Outline Neural Computation description Outline of Research Relevance to Radio Astronomy Work Update

4 Neural Computation... Modelling of systems according to brain response and neural system in living organisms Types of models: compartmental models, rate models, spiking models Modeling platforms: mathematical, hardware and software Application areas: Wireless communications, biomedical prosthetics, pattern and speech recognition, financial analysis….

5 Neural Computation... Not all neural networks are based on training and evolving an algorithm J Tapson( 1998)¹, J Tapson (2009) Benefits are found inherently from modelling close likeness of a biological model and extracting relevant information J. Tapson,1998,Autocorrelation Properties of Single neurons J.Tapson, C.Jin et.al..2009 A First Order Non-Homogeneous Markov Model for the response of Spiking Neurons Stimulated by small phase continuous signals

6 Research Outline Neural based analysis of auto/cross correlation Simulate/ build a biologically inspired correlator module ( ASIC to Reconfigurable) Test applicability to Radio Astronomy correlation requirements

7 Spiking Neuron Basic function of spiking neuron Integrate-and-fire model: membrane potential Stochastic Resonance drift noise signal

8 Spiking Neuron

9 Neuron Spike Interpretation Wolfgang Maass: Information contained in spikes Spike information is contained in the spike time independent of shape and size of the spike. Spikes analyzed in the form ISIH and post processing logic

10 The Selected Model Equivalent analog electronic circuit model Leaky integrator which resets at hysteretic comparator thresholds x(t)‏ n x (t)‏ m x y(t)‏ n y (t)‏ m y

11 The Selected Model Digital analogy of the same model adopted from FPGA Based Silicon Neural Array by Andrew Cassidy et al… Built on Altera FPGA with VHDL and Quartus software

12 Digital Platforms 1. Digital neuron implemented on VHDL-AMS (Analog Mixed Signal). –Ease of modelling 2. Field Programmable Analog Arrays:- availability

13 Proposed Architecture Signal processor CMAC in correlator Based on the functionalities of analog correlators and neurons

14 Proposed Architecture

15 Model Results Cross correlation Mathematical Cross Correlation Signals Neural Cross Correlation

16 Model Results Cross correlation Mathematical Cross Correlation Signals Neural Cross Correlation

17 Relevance to Radio Astronomy Neural networks not a new phenomenon to astronomy. Used in cluster identification, signal processing Spike interpretation can be analysed as bit stream correlators.

18 Relevance to Radio Astronomy Alternative technique for correlation that can switch from parallel to serial‏ Cost-space allocation on FPGA Power Consumption Computation effectiveness


Download ppt "Neural Cross Correlation For Radio Astronomy Chipo N Ngongoni Supervisor: Professor J Tapson Department of Electrical Engineering, University of Cape Town."

Similar presentations


Ads by Google