What we started with

20
What we started with To develop an application that facilitates user to draw interactive graphs (e.g. Flowcharts ,class diagrams, UML diagrams etc ). Interactive graphs arrange Nodes themselves in the "canvas" dynamically as the user adds diagrams ( calculates the best positions for different nodes, resizes them, and fits them canvas in such a way that it looks aesthetically pleasing to the user. It should also arrange according to the groups. ( Grouping similar Nodes).

description

What we started with. To develop an application that facilitates user to draw interactive graphs (e.g. Flowcharts ,class diagrams, UML diagrams etc ). - PowerPoint PPT Presentation

Transcript of What we started with

Page 1: What  we started  with

What we started with• To develop an application that facilitates user to draw interactive graphs

(e.g. Flowcharts ,class diagrams, UML diagrams etc ).• Interactive graphs arrange Nodes themselves in the "canvas"

dynamically as the user adds diagrams ( calculates the best positions for different nodes, resizes them, and fits them canvas in such a way that it looks aesthetically pleasing to the user.

• It should also arrange according to the groups.( Grouping similar Nodes).

Page 2: What  we started  with

Various techniques analyzed to tackle this problem statement.

• Fundamental forces (Force Based Algorithms)– Result from exchange of “carrier particles”– Equilibrium via combination of attraction & repulsion

• Diffusion of gaseous matter(Simulated Annealing)– Particles tend toward a uniform distribution

• Multi-agent swarms & crowd motion (Genetic & PSO)– Global motion emerges from local behavior– “Thinking fluids” exhibit unique other properties

Page 3: What  we started  with

Meta-Heuristics

• A name for any stochastic optimization algorithm intended to be the last resort before giving up and using random or brute-force search.

• Such algorithms are used for problems where you don't know how to find a good solution, but if shown a candidate solution, you can give it a grade.

• The algorithmic family includes genetic algorithms, hill-climbing, simulated annealing, ant colony optimization, particle swarm optimization, and so on.

Page 4: What  we started  with

Meta-Heuristics

Figure 0 The Mona Lisa, estimatedwith the (5 + 1) EvolutionStrategy. The objective was tofind a set of fifty polygons whichmost closely approximated theoriginal image. Reference: Essentials of Metaheuristics

Page 5: What  we started  with

Why Meta-Heuristics?

• We don’t have a definite solution.• Our objective functions are dynamically changing.• Our needs multi-objective functions to achieve the desired

goal.• Third approach gave more opportunity for research and

solution for variety of problems.• Existing successful applications.• Brief explanation of how this approach can solve our problem

statement.

Page 6: What  we started  with

Our Motivation-Social Behaviour

• 1.For Communication-Swarmming Behaviour of Bees.• A video or an animation.

Page 7: What  we started  with

Our Motivation-Social Behaviour

• For Layout- Flocking of Bird• “Basic models of flocking behavior are controlled by three simple rules:• Separation - avoid crowding neighbors (short range repulsion)• Alignment - steer towards average heading of neighbors• Cohesion - steer towards average position of neighbors (long range attraction)• With these three simple rules, the flock moves in an extremely realistic way, creating

complex motion and interaction that would be extremely hard to create otherwise.”” Wiki

• A video showing how flocks maintain above parameters.

Page 8: What  we started  with

Basic PSO Algorithm

Flowchart of this.• 1. Create a ‘population’ of agents (called particles) uniformly distributed over X.• 2. Evaluate each particle’s position according to the objective function.• 3. If a particle’s current position is better than its previous best position, update it.• 4. Determine the best particle (according to the particle’s previous• best positions).• 5. Update particles’ velocities according to equation:• 6. Move particles to their new positions according to equation:• Go to step 2 until stopping criteria are satisfied.

• Use the term Social Influence. Fully informed PSO.

Page 9: What  we started  with

Our approach• Variations of PSO.• Our approach.(PAPSO). • The sequential synchronous PSO algorithm updates all particle velocities and positions at the• end of every optimization iteration (Figure 1). In contrast, the sequential asynchronous PSO• algorithm updates particle positions and velocities continuously based on currently available• Less costly computationally to achieve convergence.• Fig of page 5.

• Since PAPSO incorporates a dynamic load balancing, scheme, parallel performance is dramatically increased for (1) heterogeneous computing environments,(2) user-loaded computing environments and (3) problems producing run-time load.

• was 3.5 times faster than was PSPSO for the biomechanical test problem executed on a heterogeneous• cluster with 20 processors. Overall, PAPSO exhibits excellent parallel performance when a• large number of processors (more than about 15) is utilized and either (1) heterogeneity exists in the• computational task or environment, or (2) the computation-to-communication time ratio is relatively• small.

Page 10: What  we started  with

Our first attempt(Algorithm)

• Design of various classes.

Page 11: What  we started  with

Class Diagram

Page 12: What  we started  with

Model-View/Observer Design Pattern

Page 13: What  we started  with

Factory, Iterator, Thread, Singleton, Façade Pattern use at different places.

• Every swimmable runs in Thread.• All swimmable created from same base class. etc.

Page 14: What  we started  with

Swimmable Behavior

Page 15: What  we started  with

Testing and Observations

• 1.Demo.• 2.Problems seen which hindered to achieve our goal.

– Particles got stuck due to memory of previous best.– Tried to Converge at a single point.– Single Test case not successful.(Particle bumped off the screen after

being equidistant from its neighbors)– Trying to fit both objectives at the same was time not successful.– –

Page 16: What  we started  with

2nd Attempt.

• Meta-swarm using existing architecture.• Obstacles(1-2)

Page 17: What  we started  with

Latest Attempt.

• Eliminate manager.

Page 18: What  we started  with

TODOs

• 1.More adaptive tuning.• Dynamic sleep interval• Dynamic inertia/learning• Global objective(meta-swarm)• Performance Analysis.• Not working-• Missing-Global Objective(meta-swarm)-Meta

swarm class.

Page 19: What  we started  with

Futurework

• Real truly distributed.(RMI)• Multi-Objective• Other applications for designs.

Page 20: What  we started  with

Application Domain of PSO/Swarm Intelligence

• Crowd simulation- develop crowd controlling strategies. Emergency response teams such as policemen, the National Guard, military

• Telecommunication Networks ,Network (Ant-based routing)so as to achieve minimum loss of information without degrading robustness , performance and efficiency.(Fault tolerating)

• It influencs Emergence behaviour Emergent processes or behaviours can be seen in many places, such as traffic patterns, cities, political systems of governance, cabal and market-dominant minority phenomena in politics and economics(Stock-market),WWW and Internet.

• drug trafficking exhibit similar self-organizing properties. Parallel examples exist in the world of privacy-preserving computer networks such as Tor. In each case, the network as a whole exhibits distinctive synergistic behavior through the combination of the behaviors of individual actors in the network. Usually the growth of such networks is fueled by an ideology or sociological force that is adhered to or shared by all participants in the network