DeepGenome

Basis of the DeepGenome experiment

Flow chart

This subgraph represents the calculate_total_size function, which calculates the total size of a given shape.
It starts by initializing the total_size variable to 1, then iterates through each dimension in the input shape, multiplying the total_size by the current dimension.
Finally, it returns the calculated total_size.

This subgraph represents the DeepDNA class, which is the main class for the DNA generation and manipulation.
It starts by initializing the class attributes, such as inputs, outputs, flow, and operations.
It then defines the different operations (LINEAR, ADD, ASSIGNOUT) and their associated properties, including the result_shape_of_tensor_sum function for the ADD operation and the opAssignOut_validation function for the ASSIGNOUT operation.
The class also includes methods for normalizing the inputs and outputs, generating the flow, optimizing the flow, merging with another flow, and printing the flow.

This subgraph represents the DeepDNA_Var class, which is used to represent variables in the DNA flow.
It initializes the attributes of the DeepDNA_Var class, including the size and dependsOn properties.

This subgraph represents the DeepDNA_Operation class, which is used to represent the different operations in the DNA flow.
It initializes the attributes of the DeepDNA_Operation class, including the name, numInputs, args, and validationFunc properties.
It also defines the validation method, which can be overridden by specific operation types.

This subgraph represents the DeepDNA_Sequence class, which is used to represent a sequence of operations in the DNA flow.
It initializes the attributes of the DeepDNA_Sequence class, including the reference to the DeepDNA instance, the output, inputs, operation, and arguments.
It also includes the choose method, which is responsible for choosing the appropriate operation and inputs based on the available operations and variables.

This subgraph represents the example usage of the DeepDNA class, including the creation of two DNA instances (DNA 1 and DNA 2), the generation and optimization of their flows, and the merging of the two flows into a new DNA instance.
It demonstrates the step-by-step process of working with the DeepDNA class, including normalizing the inputs and outputs, generating the flow, optimizing the flow, and printing the flow for each DNA instance.
Finally, it shows the merging of the two DNA flows into a new DNA instance and the printing of the merged flow.