[Legacy Report] What do deep learning and Computer Tomograph (CAT) have in common?

#Deap #learning #computer #tomograph
Share

This talk will discuss the Universal Approximation Theorem, which is the basis of Neural Nets and show that it is related to the Radon transform, and its inverse, used in CT reconstruction.  This gives some way of intuitively understanding the Universal approximation theorem, and why neural nets work.  

Essentially, the UAT says that any function can be implemented using a two fully-connected layers with a fairly general activation function. The Radon transform is a mathematical formulation of CT image capture, and the inverse Radon transform may be seen as a method for reconstructing and image from projections.


  Date and Time

  Location

  Hosts

  Registration


  • Add_To_Calendar_icon Add Event to Calendar
  • Canberra, Australian Capital Territory
  • Australia
  • Contact Event Host
  • Co-sponsored by Shaodi You

  Speakers

Richard Hartley of Autralian National University, Data61

Topic:

What do deep learning and Computer Tomograph (CAT) have in common?

Biography:

Email:

Address:Canberra, Australia