Home > News > Computing with nanotechnology may get a boost from neural networks
August 20th, 2007
Computing with nanotechnology may get a boost from neural networks
In this paper, a researcher from Hewlett-Packard is suggesting an alternative approach to the problem. Instead of characterizing the device post-characterization, it is treated rather like a neural network. However, the nanoscale building blocks are all linear (they multiply and add), which would normally preclude a neural network. Snider proposes that these are integrated around nonlinear components fabricated using standard lithographic techniques. This works because not many nonlinear elements are needed and they can be spaced such that the density of the nanoscale building blocks is not greatly affected. The linear elements then act to weight the input into the nonlinear elements.
Quantum effects help cells capture light, but the details are obscure: Ultrashort laser pulses reveal that 'coherence' plays a subtle role in energy transfers December 6th, 2013
Coal yields plenty of graphene quantum dots: Rice U. scientists find simple method for producing dots in bulk from coal, coke December 6th, 2013
The gene sequencing that everyone can afford in future December 6th, 2013
Silvija Gradečak seeks to better the world through new materials December 6th, 2013