Grid e-services for Multi-Layer SOM Neural Network Simulation

Grid e-services for Multi-Layer SOM Neural Network Simulation,, Rui Silva Faculdade de Engenharia 4760-108 V. N. Famalicão, Portugal {rml,rsilva}@fam.ulusiada.pt 2007

Outline Overview Multi-Layer SOM Background Architecture and Design Local Implementation Grid Proposal Environment - GRIDSOM Conclusions Grid e-services for Multi-Layer SOM Neural Network Simulation 2

Overview Vila Nova de Famalicão ILID Instituto de Investigação e Desenvolvimento ( R&D Institute) 8 - R&D Centers (Lisboa, Porto, Famalicão) 1. Energy e Gestão Industrial (CLEGI) 2. Ciências Sociais e Humanas 3. Ciências Jurídicas 4. Arquitectura 5. Centro de Matemática 6. Centro de Recursos Humanos (CIDRHU) 7. Centro de Turismo, Inovação e Serviços (TIS) 8. Centro de Economia e Gestão Intelligent Tool Wear Monitoring http://www.fam.ulusiada.pt/ilid/mifc/index.htm Grid e-services for Multi-Layer SOM Neural Network Simulation 3

Intelligent Tool Wear Monitoring Cutting Process Wear monitoring Grid e-services for Multi-Layer SOM Neural Network Simulation 4

Our Problem Experiments were conducted on a CNC Non intrusive sensors Modeling technique based on artificial neural networks SOM Grid e-services for Multi-Layer SOM Neural Network Simulation 5

Multi-Layer SOM Background Clustering Method 1. Randomly initiate weights; 2. Present input vector; 3. Determine the distance between input feature vector and each output node; 4. Adjust weights according to winning frequency; 5. Select neuron with minimum distance to input feature vector; 6. Update weights; 7. Back to step 2. 7. Division of the output map into several sub-maps Grid e-services for Multi-Layer SOM Neural Network Simulation 6

Multi-Layer SOM Results 9.00 8.00 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 9.00 8.00 7.00 Contour plot - tool with VB B =0.05 mm 9.00 8.00 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 9.00 8.00 7.00 Contour plot - tool with VB B =0.17 mm The sub-maps are treated as jobs in the GRID computing environment 6.00 6.00 5.00 5.00 4.00 4.00 3.00 3.00 2.00 2.00 1.00 1.00 0.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 0.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 Contour plot - tool with VB B =0.21mm Contour plot - tool with VB B =0.31 mm Grid e-services for Multi-Layer SOM Neural Network Simulation 7

Our Aim A lightweight architecture to support the execution of neural networks simulation tasks across the Grid infrastructure. Security Authentication Encryption Model Design Build Simulation Monitor Results Grid e-services for Multi-Layer SOM Neural Network Simulation 8

Architecture and Design Local Resources: Architecture (processor type) P IV 2GHz Number of Nodes 15 Operating system SL (Linux( Linux) Memory per Node 512 MB Grid middleware glite Disk per node 80 GB Job Manager LCG Cluster Storage 1 TB Grid e-services for Multi-Layer SOM Neural Network Simulation 9

Architecture and Design Layer Model Grid e-services for Multi-Layer SOM Neural Network Simulation 10

Architecture and Design Grid e-services for Multi-Layer SOM Neural Network Simulation 11

Neural Network Agent Manage and control tasks. Distribution of work load. Control neural network learning and recognition. Interface the communication between neural network layers. Local Service Local Database Grid e-services for Multi-Layer SOM Neural Network Simulation 12

Monitor Agent Submit jobs that are waiting in the Database. Querying the Logging and Bookkeeping. Collect those that have already been successfully finished. Re-submit unfinished jobs. Update the database with the jobs states. Scheduling System Local Service Local Database Grid e-services for Multi-Layer SOM Neural Network Simulation 13

Hybrid Approach Network Agent Simulation Algorithm Database Store Results Monitor Looks for new Jobs in Database Launch Jobs to the GRID Collect Results of job execution Fast Application Migration to GRID Grid e-services for Multi-Layer SOM Neural Network Simulation 14

GRIDSOM Prototype WEB Registration http:// ://gridsom.fam.ulusiada.pt Grid e-services for Multi-Layer SOM Neural Network Simulation 15

GRIDSOM Start Simulation Grid e-services for Multi-Layer SOM Neural Network Simulation 16

Generate.JDL File Grid e-services for Multi-Layer SOM Neural Network Simulation 17

Tasks in Database Current Tasks In the Outbox The Neural Network Agent will queue the initial job of this task Grid e-services for Multi-Layer SOM Neural Network Simulation 18

Monitor Launch Jobs Grid e-services for Multi-Layer SOM Neural Network Simulation 19

Queued Jobs in Database Job status stored in (grid_status( grid_status) Grid e-services for Multi-Layer SOM Neural Network Simulation 20

Monitor Getting Job Status Job Done! But FAIL Job re-submit! Grid e-services for Multi-Layer SOM Neural Network Simulation 21

Monitor Getting Results Job Done! Neural Network Agent Generate a new iteration Until the Monitor Submit this new iteration: end of simulation Grid e-services for Multi-Layer SOM Neural Network Simulation 22

End Simulation Neural Network Agent Publish the final results. Grid e-services for Multi-Layer SOM Neural Network Simulation 23

User Collect Results Download de Final Data Grid e-services for Multi-Layer SOM Neural Network Simulation 24

Data Visualization Data Analysis (outside GRIDSOM) Grid e-services for Multi-Layer SOM Neural Network Simulation 25

Conclusions and Future Developments Prototype successfully tested Hybrid approach SSL layer Migrate to New Scheduling System Virtual Organization Core services Grid e-services for Multi-Layer SOM Neural Network Simulation 26

Many Thanks Grid e-services for Multi-Layer SOM Neural Network Simulation 27