コード例 #1
0
ファイル: Genome.cs プロジェクト: chrisdurning22/neural-jump
        public void addNodeMutation(Random random)
        {
            // pick a random index
            int splitGeneIndex = innovationNumbers[random.Next(innovationNumbers.Count)];

            ConnectionGene c = connections[splitGeneIndex];

            c.setEnabled(false);

            NodeGene n1 = nodes[c.getIn()];
            NodeGene n2 = nodes[c.getOut()];

            // new node
            NodeGene n3 = new NodeGene(NodeGene.NodeType.HIDDEN, nodes.Count);

            // add connection genes (1) leading into the new node gets a weight of 1 (2) leading out of new node gets the same weight as old connection
            InnovationUtility.incrementGlobalInnovation();
            addConnectionGene(new ConnectionGene(n1.getNodeNumber(), n3.getNodeNumber(), 1f, true, InnovationUtility.getGlobalInnovation()));

            InnovationUtility.incrementGlobalInnovation();
            addConnectionGene(new ConnectionGene(n3.getNodeNumber(), n2.getNodeNumber(), c.getWeight(), true, InnovationUtility.getGlobalInnovation()));
            connections [splitGeneIndex] = c;

            // add node gene
            addNodeGene(n3);
        }
コード例 #2
0
        public void init()
        {
            random = new Random();
            genome = new Genome();

            int node0 = genome.getNodes().Count;

            genome.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node0));
            int node1 = genome.getNodes().Count;

            genome.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node1));
            int node2 = genome.getNodes().Count;

            genome.addNodeGene(new NodeGene(NodeGene.NodeType.HIDDEN, node2));
            int node3 = genome.getNodes().Count;

            genome.addNodeGene(new NodeGene(NodeGene.NodeType.HIDDEN, node3));
            int node4 = genome.getNodes().Count;

            genome.addNodeGene(new NodeGene(NodeGene.NodeType.OUTPUT, node4));

            InnovationUtility.incrementGlobalInnovation();
            genome.addConnectionGene(new ConnectionGene(node0, node2, -0.9f, true, InnovationUtility.getGlobalInnovation()));
            InnovationUtility.incrementGlobalInnovation();
            genome.addConnectionGene(new ConnectionGene(node1, node2, 0.9f, true, InnovationUtility.getGlobalInnovation()));
            InnovationUtility.incrementGlobalInnovation();
            genome.addConnectionGene(new ConnectionGene(node1, node3, -0.1f, true, InnovationUtility.getGlobalInnovation()));
            InnovationUtility.incrementGlobalInnovation();
            genome.addConnectionGene(new ConnectionGene(node2, node3, 0.1f, true, InnovationUtility.getGlobalInnovation()));
            InnovationUtility.incrementGlobalInnovation();
            genome.addConnectionGene(new ConnectionGene(node3, node4, 0.2f, true, InnovationUtility.getGlobalInnovation()));
        }
コード例 #3
0
        private Genome generateInitialGenome()
        {
            Genome initialGenome = new Genome();

            int node0 = initialGenome.getNodes().Count;

            initialGenome.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node0));
            int node1 = initialGenome.getNodes().Count;

            initialGenome.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node1));
            int node2 = initialGenome.getNodes().Count;

            initialGenome.addNodeGene(new NodeGene(NodeGene.NodeType.OUTPUT, node2));

            InnovationUtility.incrementGlobalInnovation();
            initialGenome.addConnectionGene(new ConnectionGene(node0, node2, Genome.getRandomFloatBetweenPoints(random, Utility.RANDOM_VAL_MIN, Utility.RANDOM_VAL_MAX), true, InnovationUtility.getGlobalInnovation()));

            InnovationUtility.incrementGlobalInnovation();
            initialGenome.addConnectionGene(new ConnectionGene(node1, node2, Genome.getRandomFloatBetweenPoints(random, Utility.RANDOM_VAL_MIN, Utility.RANDOM_VAL_MAX), true, InnovationUtility.getGlobalInnovation()));

            return(initialGenome);
        }
コード例 #4
0
        public static void Main(string[] args)
        {
            Random random = new Random();


            Genome initialGenome = new Genome();

            int node0 = initialGenome.getNodes().Count;

            initialGenome.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node0));
            int node1 = initialGenome.getNodes().Count;

            initialGenome.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node1));
            int node2 = initialGenome.getNodes().Count;

            initialGenome.addNodeGene(new NodeGene(NodeGene.NodeType.OUTPUT, node2));

            InnovationUtility.incrementGlobalInnovation();
            initialGenome.addConnectionGene(new ConnectionGene(node0, node2, Genome.getRandomFloatBetweenPoints(random, -1.0, 1.0), true, InnovationUtility.getGlobalInnovation()));

            InnovationUtility.incrementGlobalInnovation();
            initialGenome.addConnectionGene(new ConnectionGene(node1, node2, Genome.getRandomFloatBetweenPoints(random, -1.0, 1.0), true, InnovationUtility.getGlobalInnovation()));

            Genome initialGenome1 = new Genome();

            node0 = initialGenome1.getNodes().Count;
            initialGenome1.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node0));
            node1 = initialGenome1.getNodes().Count;
            initialGenome1.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node1));
            node2 = initialGenome1.getNodes().Count;
            initialGenome1.addNodeGene(new NodeGene(NodeGene.NodeType.OUTPUT, node2));

            initialGenome1.addConnectionGene(new ConnectionGene(node0, node2, Genome.getRandomFloatBetweenPoints(random, -1.0, 1.0), true, 1));

            initialGenome1.addConnectionGene(new ConnectionGene(node1, node2, Genome.getRandomFloatBetweenPoints(random, -1.0, 1.0), true, 2));

            Genome initialGenome2 = new Genome();

            node0 = initialGenome2.getNodes().Count;
            initialGenome2.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node0));
            node1 = initialGenome2.getNodes().Count;
            initialGenome2.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node1));
            node2 = initialGenome2.getNodes().Count;
            initialGenome2.addNodeGene(new NodeGene(NodeGene.NodeType.OUTPUT, node2));

            initialGenome2.addConnectionGene(new ConnectionGene(node0, node2, Genome.getRandomFloatBetweenPoints(random, -1.0, 1.0), true, 1));

            initialGenome2.addConnectionGene(new ConnectionGene(node1, node2, Genome.getRandomFloatBetweenPoints(random, -1.0, 1.0), true, 2));

            Genome initialGenome3 = new Genome();

            node0 = initialGenome3.getNodes().Count;
            initialGenome3.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node0));
            node1 = initialGenome3.getNodes().Count;
            initialGenome3.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node1));
            node2 = initialGenome3.getNodes().Count;
            initialGenome3.addNodeGene(new NodeGene(NodeGene.NodeType.OUTPUT, node2));

            initialGenome3.addConnectionGene(new ConnectionGene(node0, node2, Genome.getRandomFloatBetweenPoints(random, -1.0, 1.0), true, 1));

            initialGenome3.addConnectionGene(new ConnectionGene(node1, node2, Genome.getRandomFloatBetweenPoints(random, -1.0, 1.0), true, 2));


            // //-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

            // Genome initialGenome2 = new Genome();

            // node1 = initialGenome2.getNodes().Count;
            // initialGenome2.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node1));
            // node2 = initialGenome2.getNodes().Count;
            // initialGenome2.addNodeGene(new NodeGene(NodeGene.NodeType.INPUT, node2));
            // node3 = initialGenome2.getNodes().Count;
            // initialGenome2.addNodeGene(new NodeGene(NodeGene.NodeType.OUTPUT, node3));

            // InnovationUtility.incrementGlobalInnovation();
            // initialGenome2.addConnectionGene(new ConnectionGene(node1, node3, Genome.getRandomFloatBetweenPoints(random, -1.0, 1.0), true, 1));

            // InnovationUtility.incrementGlobalInnovation();
            // initialGenome2.addConnectionGene(new ConnectionGene(node2, node3, Genome.getRandomFloatBetweenPoints(random, -1.0, 1.0), true, 2));


//-------------------------------------------------------------------

            initialGenome.addNodeMutation(random);
            initialGenome3.addNodeMutation(random);


            Console.Write("initialGenome: ");
            foreach (KeyValuePair <int, ConnectionGene> c in initialGenome.getConnections())
            {
                Console.Write("[" + c.Key + "]");
            }

            Console.Write("\n" + "initialGenome1: ");
            foreach (KeyValuePair <int, ConnectionGene> c in initialGenome1.getConnections())
            {
                Console.Write("[" + c.Key + "]");
            }
            Console.Write("\n");

            Console.WriteLine("disjoint: " + Speciation.disjointCount(initialGenome, initialGenome1));
            Console.WriteLine("excess: " + Speciation.excessCount(initialGenome, initialGenome1));

            Console.Write("\n");

            GeneratePopulation population = new GeneratePopulation(4);

            population.addGenomeToPopulation(new Genome(initialGenome));
            population.addGenomeToPopulation(new Genome(initialGenome1));
            population.addGenomeToPopulation(new Genome(initialGenome2));
            population.addGenomeToPopulation(new Genome(initialGenome3));

            for (int i = 0; i < 60; i++)
            {
                population.sortGenomesIntoSpecies(random);
                population.addAdjustedFitness();
                population.fillNextGeneration(random);
                Console.WriteLine("\n Generation: " + i);
                // foreach(Species s in population.getSpeciesList()) {
                //  Console.WriteLine("New Species ");
                //  foreach(Genome g in s.getGenomesList()) {
                //      Console.WriteLine("Genome fitness: " + g.getFitness());
                //  }
                // }
            }



//--------------------------------------------------------------------------------


            // Console.WriteLine("Parent1");
            // foreach(KeyValuePair<int, NodeGene> n in initialGenome.getNodes()) {
            //  Console.WriteLine("ID: " + n.Value.getNodeNumber() + " Layer: " + n.Value.getNodeType());
            // }

            // Console.Write("\n");

            // foreach(KeyValuePair<int, ConnectionGene> c in initialGenome.getConnections()) {
            //  Console.WriteLine("In: " + c.Value.getIn() + " Out: " + c.Value.getOut() + " Innov: " + c.Value.getInnovationNumber() + " Enabled: " + c.Value.getEnabled() + " Weight: " + c.Value.getWeight());
            // }

            // Console.Write("\n");

            // Console.WriteLine("Parent2");
            // foreach(KeyValuePair<int, NodeGene> n in initialGenome2.getNodes()) {
            //  Console.WriteLine("ID: " + n.Value.getNodeNumber() + " Layer: " + n.Value.getNodeType());
            // }

            // Console.Write("\n");

            // foreach(KeyValuePair<int, ConnectionGene> c in initialGenome2.getConnections()) {
            //  Console.WriteLine("In: " + c.Value.getIn() + " Out: " + c.Value.getOut() + " Innov: " + c.Value.getInnovationNumber() + " Enabled: " + c.Value.getEnabled() + " Weight: " + c.Value.getWeight());
            // }

            // Console.Write("\n");

            // Genome child = Evolution.crossover(random, initialGenome, initialGenome2);

            // Console.WriteLine("Child");
            // foreach(KeyValuePair<int, NodeGene> n in child.getNodes()) {
            //  Console.WriteLine("ID: " + n.Value.getNodeNumber() + " Layer: " + n.Value.getNodeType());
            // }

            // Console.Write("\n");

            // foreach(KeyValuePair<int, ConnectionGene> c in child.getConnections()) {
            //  Console.WriteLine("In: " + c.Value.getIn() + " Out: " + c.Value.getOut() + " Innov: " + c.Value.getInnovationNumber() + " Enabled: " + c.Value.getEnabled() + " Weight: " + c.Value.getWeight());
            // }


            //------------------------------------------------------------------------------------------------------------------------------------------------------------

            // Console.WriteLine(Speciation.compatibilityDistance(initialGenome, initialGenome2));


            //------------------------------------------------------------------------------------------------------------------------------------------------------------

            // GeneratePopulation population = new GeneratePopulation(4);

            // population.addGenomeToPopulation(new Genome(initialGenome));
            // population.addGenomeToPopulation(new Genome(initialGenome1));
            // population.addGenomeToPopulation(new Genome(initialGenome2));
            // population.addGenomeToPopulation(new Genome(initialGenome3));


            // GeneratePopulation population = new GeneratePopulation(50);

            // for(int i = 0; i < population.getPopulationSize(); i++) {
            //  population.addGenomeToPopulation(new Genome(initialGenome));
            // }


            // print connection genes
            // Console.WriteLine("CONNECTION GENES: ");
            // foreach(ConnectionGene gene in initialGenome.connections) {
            //  Console.WriteLine("IN: " + gene.getIn() + " OUT: " + gene.getOut() + " WEIGHT: " + gene.getWeight() + " ENABLED: " + gene.getEnabled() + " INNO: " + gene.getInnovationNumber());

            // }

            // Console.WriteLine("\n");



            // for(int i = 1; i < 20; i++) {
            //  bestGen = population.GenerateNextGeneration(random);


            //  Console.Write(" Generation: " + i);
            //  // Console.Write(", Species Count: " + population.getSpeciesList().Count);
            //  Console.Write(", Highest Fitness: " + population.getHighestScore() + "\n");
            // }
        }
コード例 #5
0
ファイル: Genome.cs プロジェクト: chrisdurning22/neural-jump
        public bool addConnectionMutation(Random random)
        {
            bool     uniqueMatch = false;
            NodeGene n1          = null;
            NodeGene n2          = null;

            if (allConnectionsMade())
            {
                return(false);
            }

            // to be used in circular loop checker
            Utility.nodeConnections.Clear();
            foreach (KeyValuePair <int, ConnectionGene> c in connections)
            {
                //if key already exists add to hashset
                if (Utility.nodeConnections.ContainsKey(c.Value.getOut()))
                {
                    Utility.nodeConnections[c.Value.getOut()].Add(c.Value.getIn());
                }
                else
                {
                    Utility.nodeConnections.Add(c.Value.getOut(), new HashSet <int>()
                    {
                        c.Value.getIn()
                    });
                }
            }

            while (!uniqueMatch)
            {
                uniqueMatch = true;
                n1          = nodes[nodeNumbers[random.Next(nodeNumbers.Count)]];
                n2          = getDifferentNode(random, n1, nodes);

                if (circularConnectionFound(n1.getNodeNumber(), n2.getNodeNumber()))
                {
                    return(false);
                }

                // if n1 and n1 aren't INPUT nodes
                if (!(n1.getNodeType() == NodeGene.NodeType.INPUT && n2.getNodeType() == NodeGene.NodeType.INPUT))
                {
                    makeN1PointToN2(ref n1, ref n2);

                    foreach (KeyValuePair <int, ConnectionGene> c in connections)
                    {
                        // checks to see if connection already exists
                        if (c.Value.getIn() == n1.getNodeNumber() && c.Value.getOut() == n2.getNodeNumber())
                        {
                            uniqueMatch = false;
                        }
                    }
                }
                else
                {
                    uniqueMatch = false;
                }
            }

            InnovationUtility.incrementGlobalInnovation();
            addConnectionGene(new ConnectionGene(n1.getNodeNumber(), n2.getNodeNumber(), getRandomFloatBetweenPoints(random, -1.0, 1.0), true, InnovationUtility.getGlobalInnovation()));

            return(true);
        }