Deep Learning Toolbox

Design, train, and analyze deep learning networks

Deep Learning Toolbox™ provides a framework for designing and implementing deep neural networks with algorithms, pretrained models, and apps. You can use convolutional neural networks (ConvNets, CNNs) and long short-term memory (LSTM) networks to perform classification and regression on image, time-series, and text data. You can build network architectures such as generative adversarial networks (GANs) and Siamese networks using automatic differentiation, custom training loops, and shared weights. With the Deep Network Designer app, you can design, analyze, and train networks graphically. The Experiment Manager app helps you manage multiple deep learning experiments, keep track of training parameters, analyze results, and compare code from different experiments. You can visualize layer activations and graphically monitor training progress.

You can exchange models with TensorFlow™ and PyTorch through the ONNX format and import models from TensorFlow-Keras and Caffe. The toolbox supports transfer learning with DarkNet-53, ResNet-50, NASNet, SqueezeNet and many other pretrained models.

You can speed up training on a single- or multiple-GPU workstation (with Parallel Computing Toolbox™), or scale up to clusters and clouds, including NVIDIA^® GPU Cloud and Amazon EC2^® GPU instances (with MATLAB Parallel Server™).

What Is Deep Learning Toolbox?

YOLO v2 object detection network of two cars driving on a street with double yellow lines and pedestrians walking in a crosswalk.

Deep Learning Applications

Train deep learning models for classification, regression, and feature learning applications for automated driving, signal and audio processing, wireless communications, image processing, and more.

Introduction to Deep Learning: What Are Convolutional Neural Networks? (4:44)

Documentation | Examples

Screenshot of the Deep Network Designer app.

Network Design and Model Management

Speed up the development of deep learning models using low-code apps. Create, train, analyze, and debug a network using Deep Network Designer app. Tune and compare multiple models using Experiment Manager app.

Interactively Build, Visualize, and Edit Deep Learning Networks (3:54)

Documentation | Examples

Examples of other pretrained networks in Deep Network Designer.

Pretrained Models

Access popular models with a single line of code in MATLAB. Use PyTorch™ via ONNX and TensorFlow™ to import any model into MATLAB.

Deep Learning with MATLAB: Transfer Learning with Neural Networks in MATLAB (4:05)

Documentation | Examples

Four images of the same scene that represent the test image, semantic segmentation, Grad-CAM of the road, and Grad-CAM of the pavement.

Explainability

Visualize training progress and activations of the learned features in a deep learning network. Use Grad-CAM, Occlusion Mapping, and LIME to explain deep learning model results.

Documentation | Examples

Four images of a dog that shows how to perform common kinds of image and bounding box augmentation as part of object detection workflows.

Preprocessing

Label, process, and augment data for network training. Automate data labeling with built-in algorithms.

Documentation | Examples

Two graphs used for training progress show one with a blue curve in an upward trajectory tracking accuracy and the second has an orange curve in a downward trajectory tracking loss.

Training Acceleration

Speed up deep learning training using GPUs, cloud acceleration, and distributed computing.

Documentation | Examples

Code Generation

Automatically generate optimized CUDA^® code with GPU Coder™, and generate C and C++ code with MATLAB Coder™ to deploy deep learning networks to NVIDIA GPUs and various processors. Prototype and implement deep learning networks on FPGAs and SoCs using Deep Learning HDL Toolbox™.

Generate Generic C/C++ Code for Deep Learning Networks in Simulink (5:10)

Documentation | Examples

Three screenshots that show virtual sensor modeling using deep learning and one screenshot of a line graph plotting variables such as truth, E K F, deep learning F N N, and deep learning L S T M.

Simulation with Simulink

Simulate deep learning networks with control, signal processing, and sensor fusion components to assess the impact of your deep learning model on system-level performance.

Deep Learning in Simulink (3:45)

Documentation | Examples

Screenshot of the Deep Network Quantizer window used to reduce the model size, validate, and visualize the results. There are three sections on the screen: Net Layer Graph, Calibration Statistics, and Validation Summary.

Deep Learning Compression

Quantize and prune your deep learning network to reduce memory usage and increase inference performance. Analyze and visualize the tradeoff between increased performance and inference accuracy using the Deep Network Quantizer app.

Quantizing a Deep Learning Network in MATLAB (2:50)

Documentation | Examples