140 likes | 244 Views
Vector Space Embedding of Graphs via Statistics of Labelling Information. Ernest Valveny Computer Vision Center Computer Science Departament - UAB. Problem to be solved. How do we solve the problem?. Motivation ( discrete case ). Extension to continuous case.
E N D
Vector Space Embedding of Graphs viaStatistics of LabellingInformation Ernest Valveny Computer VisionCenter Computer Science Departament - UAB
Extension to continuouscase (softversion) • Softassignment of nodes to representatives • Thefrequency of eachword is theaccumulation for all nodes
Someissues • Selection of the set of representatives • K-means • Fuzzy k-means • Spanningprototypes • Mixture of gaussians • Sparsityandhigh-dimensionality of representation • Featureselection • Combination of different sets of representatives • Multipleclassifiersystemswithdifferentnumber of representatives
Conclusions • Simple embeddingmethodology • Computationallyefficient • Based on unaryandbinaryrelationsbetween nodes • Goodresults for classificationandclustering