WP5 - Network Comparison at Different Areas and Spatio-Temporal Scales
| Participating partner | Total Pm per organization |
|---|---|
| FORTH | 7 |
| PAGNI | - |
| NEUROCOM | 2 |
| UZH | - |
| PROMETRONICS | 19 |
| UA | 4 |
| BRUKER | - |
| HARVARD | 2 |
| UCLA | - |
Tasks 5.1 - 5.4
- Data preparation in the appropriate format for analysis (UA, FORTH)
- Develop appropriate similarity metrics for comparing different network architectures and apply graph similarity algorithms for comparing network topologies. (FORTH, ProMetronics) Combine in an innovative manner, graph theory, subgraph matching, graphlets, deltaconnectivity, and dynamic tensor analysis, and similarity metrics (based on graph kernels, and information theory) for comparing biological network architectures (or patterns) of the same area and scale.
- Comparing Network Topologies in Passive vs. Active Learning within specific layer and across the whole brain (FORTH, ProMetronics) Map specific local subnetworks to global area activities, under the same conditions;
- Perform feature extraction to find the predictive power of certain local subnetworks on global activities (e.g., using regression, deep learning algorithms); (Prometronics, Forth, 13M-36M)
Publications
D5.1 TR/Scientific Publication that presents the changes of the whole brain connectivity across learning stages (UA, PDE,
PU, 24M)
D5.2 TR on how the learning alters cortical network architectures in V1 and PFC before and after learning (FORTH, PDE,
PU, M40)
D5.3 TR/Scientific Publ. on using RNNs to model and predict (1) motifs in L2/3 from the events in L4, (2) local subnetwork
patterns, global activity (signature). (ProMetronics, PDE, PU, M44)
D5.4 Collected traces (T4.1) and software for the analysis made available to the web repository (Neurocom,PDE, PU, M48)