The paper was published in Neural Computation and is available at
For more information, contact the main author Ville Karlsson who is a PhD student in the Vision Group.
New paper about spiking neural networks
Our recent paper introduces neural circuits that can convert between different neural encodings. Different encodings are suitable for different tasks - and perhaps brains use them for optimal computation (in terms of energy consumption). Spiking neural networks can be 1000 times more energy efficient than conventional deep neural networks.