Skip to content

Neural network classification workflow

The Neural network classification workflow describes the standard process for using deep learning to categorize data. This workflow emphasizes the role of deep learning in performing feature extraction to classify inputs^[600-developer__big-data__ai__tensorflow-01.md].

Workflow Steps

The classification process is divided into four distinct stages:

  1. Extract Entity Features: Identify and isolate relevant characteristics or variables from the input data^[600-developer__big-data__ai__tensorflow-01.md].
  2. Define Neural Network Structure: Design the model architecture, organizing layers and nodes^[600-developer__big-data__ai__tensorflow-01.md].
  3. Train Model: Execute the training process, feeding data into the network to adjust internal parameters and minimize error^[600-developer__big-data__ai__tensorflow-01.md].
  4. Predict: Use the trained model to categorize new, unseen data based on learned patterns^[600-developer__big-data__ai__tensorflow-01.md].

Core Concepts

This workflow relies on the fundamental concept of a Tensor, which represents data as multidimensional arrays within the neural network structure^[600-developer__big-data__ai__tensorflow-01.md]. The term "flow" in this context refers to the computational model used to process these tensors^[600-developer__big-data__ai__tensorflow-01.md].

  • [[Deep Learning]]
  • [[Feature Extraction]]

Sources

^[600-developer__big-data__ai__tensorflow-01.md]