Cuantificación, proyección y poda

Comprima una red neuronal profunda mediante el proceso de cuantificación, proyección o poda.

Utilice Deep Learning Toolbox™ con el paquete de soporte Deep Learning Toolbox Model Quantization Library para reducir el uso de memoria y los requisitos computacionales de una red neuronal profunda de las siguientes formas:

Podando filtros desde las capas de convolución mediante una aproximación de Taylor de primer orden. Después, podrá generar código C/C++, CUDA^® o HDL a partir de esta red podada.
Proyectando capas realizando un análisis de componentes principales (PCA) en las activaciones de capa empleando un conjunto de datos representativo de los datos de entrenamiento y aplicando proyecciones lineales en los parámetros que se pueden aprender de la capa. Los pasos hacia adelante de una red neuronal profunda proyectada suelen ser más rápidos cuando se despliega la red en hardware integrado utilizando la generación de código C/C++ sin biblioteca.
Cuantificando los pesos, los sesgos y las activaciones de capas en tipos de datos de precisión enteros escalados. Después, podrá generar código C/C++, CUDA o HDL a partir de esta red cuantificada.
Para generación de código C/C++ y CUDA, el software genera código para una red neuronal profunda convolucional cuantificando los pesos, los sesgos y las activaciones de las capas convolucionales en tipos de datos enteros escalados de 8 bits. La cuantificación se lleva a cabo proporcionando el archivo de resultados de la calibración generado por la función calibrate al comando codegen (MATLAB Coder).
La generación de código no es compatible con redes neuronales profundas cuantificadas producidas por la función quantize.

Funciones

expandir todo

Poda

`taylorPrunableNetwork`	Network that can be pruned by using first-order Taylor approximation (desde R2022a)
`forward`	Compute deep learning network output for training (desde R2019b)
`predict`	Compute deep learning network output for inference (desde R2019b)
`updatePrunables`	Remove filters from prunable layers based on importance scores (desde R2022a)
`updateScore`	Compute and accumulate Taylor-based importance scores for pruning (desde R2022a)
`dlnetwork`	Redes neuronales de deep learning (desde R2019b)

Proyección

`compressNetworkUsingProjection`	Compress neural network using projection (desde R2022b)
`neuronPCA`	Principal component analysis of neuron activations (desde R2022b)
`unpackProjectedLayers`	Unpack projected layers of neural network (desde R2023b)
`ProjectedLayer`	Compressed neural network layer using projection (desde R2023b)
`gruProjectedLayer`	Gated recurrent unit (GRU) projected layer for recurrent neural network (RNN) (desde R2023b)
`lstmProjectedLayer`	Long short-term memory (LSTM) projected layer for recurrent neural network (RNN) (desde R2022b)

Cuantificación

`dlquantizer`	Quantize a deep neural network to 8-bit scaled integer data types (desde R2020a)
`dlquantizationOptions`	Options for quantizing a trained deep neural network (desde R2020a)
`calibrate`	Simulate and collect ranges of a deep neural network (desde R2020a)
`quantize`	Quantize deep neural network (desde R2022a)
`validate`	Quantize and validate a deep neural network (desde R2020a)
`quantizationDetails`	Display quantization details for a neural network (desde R2022a)
`estimateNetworkMetrics`	Estimate network metrics for specific layers of a neural network (desde R2022a)
`equalizeLayers`	Equalize layer parameters of deep neural network (desde R2022b)

Apps

Deep Network Quantizer

Quantize deep neural network to 8-bit scaled integer data types (desde R2020a)

Temas

Poda

Parameter Pruning and Quantization of Image Classification Network
Use parameter pruning and quantization to reduce network size.
Prune Image Classification Network Using Taylor Scores
This example shows how to reduce the size of a deep neural network using Taylor pruning.
Prune Filters in a Detection Network Using Taylor Scores
This example shows how to reduce network size and increase inference speed by pruning convolutional filters in a you only look once (YOLO) v3 object detection network.
Prune and Quantize Convolutional Neural Network for Speech Recognition
This example shows how to compress a convolutional neural nework (CNN) to prepare it for deployment on an embedded system.

Proyección y destilación de conocimiento

Compress Neural Network Using Projection
This example shows how to compress a neural network using projection and principal component analysis.
Compress Network for Estimating Battery State of Charge
This example shows how to compress a neural network for predicting the state of charge of a battery using projection and principal component analysis. (desde R2023b)
Train Smaller Neural Network Using Knowledge Distillation
This example shows how to reduce the memory footprint of a deep learning network by using knowledge distillation. (desde R2023b)

Cuantificación

Quantization of Deep Neural Networks
Understand effects of quantization and how to visualize dynamic ranges of network convolution layers.
Quantization Workflow Prerequisites
Products required for the quantization of deep learning networks.
Prepare Data for Quantizing Networks
Supported datastores for quantization workflows.
Quantize Multiple-Input Network Using Image and Feature Data
Quantize Multiple Input Network Using Image and Feature Data

Cuantificación para un objetivo GPU

Generate INT8 Code for Deep Learning Networks (GPU Coder)
Quantize and generate code for a pretrained convolutional neural network.
Quantize Residual Network Trained for Image Classification and Generate CUDA Code
This example shows how to quantize the learnable parameters in the convolution layers of a deep learning neural network that has residual connections and has been trained for image classification with CIFAR-10 data.
Quantize Layers in Object Detectors and Generate CUDA Code
This example shows how to generate CUDA® code for an SSD vehicle detector and a YOLO v2 vehicle detector that performs inference computations in 8-bit integers for the convolutional layers.
Quantize Semantic Segmentation Network and Generate CUDA Code
Quantize Convolutional Neural Network Trained for Semantic Segmentation and Generate CUDA Code

Cuantificación para un objetivo FPGA

Quantize Network for FPGA Deployment (Deep Learning HDL Toolbox)
Reduce the memory footprint of a deep neural network by quantizing the weights, biases, and activations of convolution layers to 8-bit scaled integer data types.
Classify Images on FPGA Using Quantized Neural Network (Deep Learning HDL Toolbox)
This example shows how to use Deep Learning HDL Toolbox™ to deploy a quantized deep convolutional neural network (CNN) to an FPGA.
Classify Images on FPGA by Using Quantized GoogLeNet Network (Deep Learning HDL Toolbox)
This example show how to use the Deep Learning HDL Toolbox™ to deploy a quantized GoogleNet network to classify an image.

Cuantificación para un objetivo CPU

Generate int8 Code for Deep Learning Networks (MATLAB Coder)
Quantize and generate code for a pretrained convolutional neural network.
Generate INT8 Code for Deep Learning Network on Raspberry Pi (MATLAB Coder)
Generate code for deep learning network that performs inference computations in 8-bit integers.

Ejemplos destacados

Prune Image Classification Network Using Taylor Scores

Reduce the size of a deep neural network using Taylor pruning. By using the taylorPrunableNetwork function to remove convolution layer filters, you can reduce the overall network size and increase the inference speed.

Abrir script en vivo

Prune Filters in a Detection Network Using Taylor Scores

Reduce network size and increase inference speed by pruning convolutional filters in a you only look once (YOLO) v3 object detection network.

Abrir script en vivo

Compress Neural Network Using Projection

Compress a neural network using projection and principal component analysis.

Abrir script en vivo

Quantize Residual Network Trained for Image Classification and Generate CUDA Code

Quantize the learnable parameters in the convolution layers of a deep learning neural network that has residual connections and has been trained for image classification with CIFAR-10 data.

Abrir script en vivo

Prune and Quantize Semantic Segmentation Network

Reduce the memory footprint of a semantic segmentation network and speed-up inference by compressing the network using pruning and quantization.

Abrir script en vivo

Explore Quantized Semantic Segmentation Network Using Grad-CAM

Compare the predictions of a quantized semantic segmentation network to the original network using the gradient-weighted class activation mapping (Grad-CAM) interpretability method.

Abrir script en vivo