parametersGenetic.json

{
  "layers": {
    "Dense": {
      "in" : "vector",
      "out" : "vector",
      "parameters": {
        "kernel_initializer" : {
          "values" : ["uniform", "lecun_uniform", "normal", "zero", "glorot_normal", "glorot_uniform", "he_normal", "he_uniform"],
          "type" : "categorical"
        },
        "activation" : {
          "values" : ["relu", "softmax", "softplus", "softsign", "tanh", "sigmoid", "hard_sigmoid", "linear"],
          "type" : "categorical"
        },
        "units" : {
          "values" : [10, 500, 10],
          "type" : "range"
        }
      }
    },
    "Dropout" : {
      "in" : "vector",
      "out" : "vector",
      "parameters" : {
        "rate": {
          "values": [0.1, 0.8],
          "type": "rangeDouble"
        }
      }
    },
    "Convolution2D" : {
      "in" : "matrix4D",
      "out" : "matrix4D",
      "parameters" : {
        "filters" : {
          "values" : [5, 50, 5],
          "type" : "range"
        },
        "kernel_size" : {
          "values" : [3, 15, 2],
          "type" : "range"
        },
        "kernel_initializer" : {
          "values" : ["uniform", "lecun_uniform", "normal", "zero", "glorot_normal", "glorot_uniform", "he_normal", "he_uniform"],
          "type" : "categorical"
        },
        "activation" : {
          "values" : ["relu", "softplus", "softsign", "tanh", "sigmoid", "hard_sigmoid", "linear"],
          "type" : "categorical"
        },
        "padding" : {
          "values" : ["same"],
          "type" : "categorical"
        },
        "use_bias" : {
          "type" : "boolean"
        },
        "data_format" : {
          "values" : ["channels_last"],
          "type" : "categorical"
        }
      }

    },
    "MaxPooling2D" : {
      "in" : "matrix4D",
      "out" : "matrix4D",
      "parameters" : {
        "pool_size" : {
          "values" : [2,6,1],
          "size" : 2,
          "type" : "2Drange"
        },
        "strides" : {
          "values" : [2, 3, 4, 5, 6, null],
          "type" : "categoricalNumeric"
        },
        "padding" : {
          "values" : ["same"],
          "type" : "categorical"
        },
        "data_format" : {
          "values" : ["channels_last"],
          "type" : "categorical"
        }
      }
    },
    "Reshape" : {
      "in" : "vector",
      "out" : "matrix",
      "parameters" : {
        "target_shape" : {
          "aspect_ratio" : 1.0,
          "type" : "matrixRatio"
        }
      }
    },
    "Flatten" : {
      "in" : "matrix",
      "out" : "vector"
    }
  },
  "global_parameters" : {
        "optimizer" : {
          "values" : ["adam", "sgd", "rmsprop", "adagrad", "adamax", "nadam"],
          "type" : "categorical"
        },
        "epochs": {
          "values" : [2, 20, 2],
          "type" : "range"
        },
        "batch_size": {
          "values" : [100, 5000, 100],
          "type" : "range"
        },
        "number_layers": {
          "values" : [3, 20, 1],
          "type" : "range"
        }
  },
  "fsm": {
    "alphabet": ["Reshape", "Dense", "Dropout", "Convolution2D", "MaxPooling2D", "Flatten"],
    "states": ["inicial", "Dense", "Reshape", "Dropout", "Flatten", "Convolution2D", "MaxPooling2D", "Dense_"],
    "initial": "inicial",
    "finals": ["Dense"],
    "map": {
      "inicial": {"Dense": "Dense_", "Reshape": "Reshape", "Dropout": "Dropout"},
      "Dense": {"Dense": "Dense", "Reshape": "Reshape", "Dropout": "Dropout"},
      "Reshape": {"Convolution2D": "Convolution2D"},
      "Dropout": {"Reshape": "Reshape", "Dense": "Dense"},
      "Flatten": {"Dense": "Dense", "Dropout": "Dropout"},
      "Convolution2D": {"Flatten": "Flatten", "Convolution2D": "Convolution2D", "MaxPooling2D": "MaxPooling2D"},
      "MaxPooling2D": {"Flatten": "Flatten", "Convolution2D": "Convolution2D"},
      "Dense_": {"Dense": "Dense", "Dropout": "Dropout", "Reshape": "Reshape"}}
  }
}