Creating and sharing knowledge for telecommunications

Seminar - Shining light on cortical connections


on 18-12-2014

... Leopoldo Petreanu (Champalimaud Neuroscience Programme)

Date & time: 18/12/2014 at 10:30.

Location: Seminar Room (2.8.3), Physics Department, Instituto Superior Técnico, Lisbon.

Abstract:
The cerebral neocortex underlies human's unique cognitive abilities. Understanding how the neuronal circuits of the neocortex allow so many complex behaviors is one of the central challenges of neuroscience. Functionally specialized cortical areas communicate with each others through an extensive network of long-range connections. We developed novel optical-methods that allow studying the connectivity and function of long-range cortical connections with unprecedented detail. Using these cutting-edge techniques we found rules organizing the connectivity of long-range projections linking distant cortical areas and we recorded the signals relayed by these projections in behaving mice.

Physics of Information Seminar

Supported with funding from FCT, FEDER and EU FP7, specifically through FCT strategic project FCT PEst-OE/EEI/LA0008/2013 and the FP7 projects Landauer (GA 318287) and PAPETS (GA 323901). More Information..

Colloquium: Quantum machine learning


on 10-12-2014

... Seth Lloyd, Massachusetts Institute of Technology


Wednesday 10/12/2014 at 15:00

Amphitheatre VA1, Civil Engineering Building, IST


Machine learning algorithms look for patterns in data. Frequently, that data comes in the form of large arrays of high-dimensional vectors. Quantum computers are adept at manipulating large arrays of high-dimensional vectors. This talk presents a series of quantum algorithms for big data analysis. The ability of quantum computers to perform Fourier transforms, find eigenvectors and eigenvalues, and invert matrices translates into quantum algorithms for clustering, principal component analysis, and for identifying topological features such as numbers of connected components, holes and voids. These quantum algorithms are exponentially faster than their classical counterparts: complex patterns in datasets of size N can be identified in time O(logN). The talk will discuss methods for implementing quantum machine learning algorithms on the current generation of quantum information processors.
More Information..