IBM - CVUT Student Research Projects SOM for Cell Broadband Engine Petr Hafner (hafnep1@fel.cvut.cz)
SOM – self organizing maps About Illustration • a subtype of artificial neural networks • unsupervised learning • vector quantization • first described by the Finnish professor Teuvo Kohonen IBM - CVUT Student Research Projects 2
Application • Speech, audio, video … processing • Face identification system • Hand-writing text transcription • Searching similarities in an unknown signals • Unknown noise reduction • Automatic sorting IBM - CVUT Student Research Projects 3
Principle • Randomize the map's nodes' weight vectors • Grab an input vector • Traverse each node in the map – Use Euclidean distance formula to find similarity between the input vector and the map's node's weight vector – Track the node that produces the smallest distance (so-called Best Matching Unit – BMU ) • Update the nodes in the neighbourhood of BMU – Wv(t + 1) = Wv(t) + Θ(t)α(t)(D(t) - Wv(t)) IBM - CVUT Student Research Projects 4
Goals of the project • To study the architecture of the Cell Broadband Engine • To create effective vector implementation of the SOM on the Cell IBM - CVUT Student Research Projects 5
What ’s done ? • Understanding and orientation in Cell Broadband Engine architecture IBM - CVUT Student Research Projects 6
What is to be done? • Multithreaded solution • Vector multithreaded solution IBM - CVUT Student Research Projects 7
IBM - CVUT Student Research Projects 8 Thank you for your attention
Recommend
More recommend