# multilayer perceptron

> type of feedforward neural network with multiple fully connected layers

**Wikidata**: [Q2991667](https://www.wikidata.org/wiki/Q2991667)  
**Wikipedia**: [English](https://en.wikipedia.org/wiki/Multilayer_perceptron)  
**Source**: https://4ort.xyz/entity/multilayer-perceptron

## Summary
A multilayer perceptron (MLP) is a type of feedforward artificial neural network characterized by having multiple fully connected layers of nodes. As a foundational class of neural network, it is an extension of the original perceptron and is capable of learning complex, non-linear patterns in data. MLPs are used for tasks like statistical classification, data mining, and speech synthesis.

## Key Facts
- **Classification**: Subclass of feedforward neural network.
- **Inventor**: Frank Rosenblatt is credited as the discoverer in 1957.
- **Foundation**: Based on the perceptron, a 1957 algorithm for supervised learning.
- **Core Components**: Uses backpropagation for training and employs activation functions.
- **Structure**: Consists of multiple layers where the nodes in each layer are fully connected to the nodes in the next.
- **Common Alias**: Frequently abbreviated as MLP.
- **Applications**: Used for statistical classification, fitness approximation, speech synthesis, and data mining.

## FAQs
### Q: What is a multilayer perceptron (MLP)?
A: A multilayer perceptron, or MLP, is a class of feedforward artificial neural network. It consists of multiple layers of nodes, with each layer being fully connected to the next one, allowing it to model and learn from complex data.

### Q: Who invented the multilayer perceptron?
A: The American psychologist and computer scientist Frank Rosenblatt is credited with discovering the multilayer perceptron in 1957.

### Q: What is the difference between a perceptron and a multilayer perceptron?
A: A multilayer perceptron is based on the original perceptron but is distinguished by having multiple layers of nodes, whereas the original perceptron has only a single layer. This multi-layer structure allows the MLP to solve more complex problems that are not linearly separable.

## Why It Matters
The multilayer perceptron represents a critical step in the evolution of artificial intelligence. The original perceptron, while groundbreaking, was limited to solving linearly separable problems. The introduction of multiple layers and the use of activation functions in the MLP allowed neural networks to learn and model complex, non-linear relationships in data for the first time.

This breakthrough overcame major limitations of earlier models and paved the way for more advanced deep learning architectures. The use of the backpropagation algorithm for training MLPs became a standard and essential technique for nearly all subsequent neural networks. Its ability to handle a wide variety of tasks, from statistical classification to data mining, makes it a foundational and versatile tool in machine learning.

## Notable For
- **Multiple Layers**: Its defining feature is the presence of at least one hidden layer between the input and output layers, distinguishing it from the single-layer perceptron.
- **Fully Connected Architecture**: In an MLP, every node in a given layer is connected to every node in the subsequent layer, creating a dense network structure.
- **Use of Backpropagation**: It is a classic example of a network trained using the backpropagation algorithm, which is fundamental to modern deep learning.
- **Modeling Non-Linearity**: By employing activation functions, MLPs can capture and model non-linear patterns, a capability that the original perceptron lacked.

## Body
### Classification and Architecture
A multilayer perceptron (MLP) is a subclass of a feedforward neural network. In this architecture, the connections between nodes do not form a cycle. The structure is composed of multiple layers of nodes, with each layer being fully connected to the one that follows it.

### Origins
- **Based On**: The MLP is based on the perceptron, an algorithm for supervised learning of binary classifiers developed in 1957.
- **Inventor**: Frank Rosenblatt is credited as the discoverer of the MLP, with an associated date of 1957.

### Mechanism
An MLP utilizes two key mechanisms for its operation:
- **Backpropagation**: It employs the backpropagation algorithm for training the network.
- **Activation Function**: It uses activation functions within its nodes to allow the network to learn non-linear models.

### Applications
MLPs are used in a variety of computational tasks, including:
- Fitness approximation
- Statistical classification
- Speech synthesis
- Data mining

### Identifiers and Aliases
- **Common Aliases**: MLP, multi-layer perceptron, perceptron multicapa, 多层感知器
- **MeSH Descriptor ID**: D000098421
- **Freebase ID**: /m/03bx6t8
- **Golden ID**: Multilayer_perceptron-Y3RWAW

## References

1. Freebase Data Dumps. 2013
2. [Source](https://golden.com/wiki/Multilayer_perceptron-Y3RWAW)
3. [OpenAlex](https://docs.openalex.org/download-snapshot/snapshot-data-format)