The TinyML Neural Network Ontology

A neural network is a computer system modeled on the biological neural networks that form the brains of humans. An NN consists of layers, which can be divided into three categories: the input layer, the middle layer (hidden layer), and the output layer. Layers are composed of many interconnected neurons, each with its weights for participating computation. The input data circulates between layers and results in output.

• s3n:Algorithm
• schema:SoftwareSourceCode
• rdfs:Class

• inputLayer ^p
• middleLayer ^p
• outputLayer ^p
• hasMetric ^p
• trainingDataset ^p
• hasMultiplyAccumulateOps ^p
• hasCategory ^p

• ssn:hasInputonly
  
• ssn:hasOutputonly
  

A layer of a neural network.

• shapeIn ^p
• shapeOut ^p
• hasIndex ^p
• hasActivation ^p
• hasQuantization ^p
• hasType ^p

• inputLayer ^p
• middleLayer ^p
• outputLayer ^p

The type of activation of a layer.

• schema:Enumeration
• rdfs:Class

The type of quantization of a layer.

• schema:Enumeration
• rdfs:Class

The type of a layer.

• schema:Enumeration
• rdfs:Class

The metric type used for training the neural network

• schema:Enumeration
• rdfs:Class

The category of the neural network

• schema:Enumeration
• rdfs:Class

A description of the input to the NN.

A description of the output from the NN.

The input layer of a neural network.

The middle layer of a neural network.

The output layer of a neural network.

The shape of input of a layer.

The shape of output of a layer.

The index of a layer.

The activation function of a layer.

The quantization of a layer.

The type of a layer of a neural network.

The information about the performance of a neural network.

The value of the metric of a neural network.

The information about the used training dataset.

The muliply accumulate operations count of the neural network.

The category of a neural network.

The extra information about the NN input.

The extra information about the NN output.