The Deep Learning Toolbox for MATLAB 2023 provides a complete toolchain that enables you to model, train and deploy deep learning in one integrated environment. Compared with Python, MATLAB has a concise syntax, is easy to use, and does not require cumbersome configuration and installation, allowing you to achieve deep learning tasks faster.
MATLAB's deep learning toolbox provides a wealth of functions and algorithms, covering the whole process from data preprocessing to model training. You can easily import and process large-scale datasets, using batch import and Datastore-like functions for efficient data manipulation. MATLAB also provides an intuitive deep network designer, allowing you to quickly build and customize network structures without writing tedious code. At the same time, the collaborative work function of MATLAB and deep learning frameworks such as TensorFlow and PyTorch allows you to flexibly interact with other platforms and give full play to their respective advantages. In addition, MATLAB's deep learning toolbox also has outstanding advantages in model interpretability and feature visualization. You can gain a deep understanding of the working principles and decision-making process of deep learning models through feature map visualization, convolution kernel visualization, and category activation visualization. MATLAB also provides commonly used interpretability methods such as CAM, LIME, GRAD-CAM, etc. to help you explain and interpret the prediction results of the model. These capabilities will bring greater insight and understanding to your research and projects.
Original link: The latest machine learning and deep learning based on MATLAB 2023a
| Chapter One Introduction to the New Features of MATLAB 2023a Deep Learning Toolbox |
1. Overview of MATLAB Deep Learning Toolbox 2. Function introduction and demonstration of Live Script and Control 3. Introduction and demonstration of batch big data import and Datastore functions 4. Data Cleaning (Data Cleaning) function introduction and demonstration 5. Introduction and demonstration of Deep Network Designer functions 6. Function introduction and demonstration of Experiment Manager 7. Introduction to MATLAB Deep Learning Model Hub 8. Introduction and demonstration of MATLAB, TensorFlow, PyTorch and other deep learning frameworks working together 9、MATLAB Deep Learning Toolbox Examples简介 |
|
| Chapter two convolutional neural network （Convolutional Neural Network, CNN） |
1. The difference and connection between deep learning and traditional machine learning 2. The basic principle of convolutional neural network (what is a convolution kernel? What is the typical topology of CNN? What is the weight sharing mechanism of CNN? What are the features extracted by CNN?) 3. Differences and connections of classic deep neural networks such as LeNet, AlexNet, Vgg-16/19, GoogLeNet, ResNet 4. Download and installation of pre-trained models (Alexnet, Vgg-16/19, GoogLeNet, ResNet, etc.) 5. Optimization algorithm (gradient descent, stochastic gradient descent, small batch stochastic gradient descent, momentum method, Adam, etc.) 6. Parameter adjustment skills (parameter initialization, data preprocessing, data amplification, batch normalization, hyperparameter optimization, network regularization, etc.) 7. Case explanation: (1) CNN pre-training model realizes object recognition (2) Using convolutional neural network to extract abstract features (3) Customized convolutional neural network topology (4) 1D CNN model solves regression fitting prediction problem 8. Practical exercises |
|
| third chapter Model Interpretability and Feature Visualization Model Explanation and Feature Visualization |
1. What is model interpretability? Why is an explanation of the CNN model needed? 2. What are the commonly used visualization methods (feature map visualization, convolution kernel visualization, category activation visualization, etc.)? 3. Explanation of the principles of CAM (Class Activation Mapping), LIME (Local Interpretable Model-agnostic Explanation), GRAD-CAM and other methods 4. Case explanation 5. Practical exercises |
|
| Chapter Four transfer learning algorithm （Transfer Learning） |
1. The basic principle of transfer learning algorithm (why transfer learning is needed? What is the basic idea of transfer learning?) 2. Migration learning algorithm based on deep neural network model 3、案例讲解：基于Alexnet预训练模型的模型迁移 4、实操练习 |
|
| 第五章 循环神经网络与长短时记忆神经网络 （RNN & LSTM） |
1. 循环神经网络（RNN）与长短时记忆神经网络（LSTM）的基本原理 2. RNN与LSTM的区别与联系 3. 案例讲解： 1）时间序列预测 2）序列-序列分类 4. 实操练习 |
|
| 第六章 时间卷积网络（Temporal Convolutional Network, TCN） |
1. 时间卷积网络（TCN）的基本原理 2. TCN与1D CNN、LSTM的区别与联系 3. 案例讲解： 1）时间序列预测：新冠肺炎疫情预测 2）序列-序列分类：人体动作识别 4. 实操练习 |
|
| 第七章 生成式对抗网络 （Generative Adversarial Network） |
1、生成式对抗网络GAN（什么是对抗生成网络？为什么需要对抗生成网络？对抗生成网络可以做什么？） 2、案例讲解：向日葵花图像的自动生成 3、实操练习 |
|
| 第八章 自编码器 （AutoEncoder） |
1、自编码器的组成及基本工作原理 2、经典自编码器（栈式自编码器、稀疏自编码器、去噪自编码器、卷积自编码器、掩码自编码器等） 3、案例讲解：基于自编码器的图像分类 4、实操练习 |
|
| 第九章 目标检测YOLO模型 |
1、什么是目标检测？目标检测与目标识别的区别与联系？YOLO模型的工作原理 2、案例讲解：(1)标注工具Image Labeler功能简介与演示 (2)使用预训练模型实现图像、视频等实时目标检测 (3)训练自己的数据集：新冠疫情佩戴口罩识别 3、实操练习 |
|
| 第十章 U-Net模型 |
1、语义分割（Semantic Segmentation）简介 2、U-Net模型的基本原理 3、案例讲解：基于U-Net的多光谱图像语义分割 |
|
| 第十一章 讨论与答疑 |
1、如何查阅文献资料？（你会使用Google Scholar、Sci-Hub、ResearchGate吗？应该去哪些地方查找与论文配套的数据和代码？） 2、如何提炼与挖掘创新点？（如果在算法层面上难以做出原创性的工作，如何结合自己的实际问题提炼与挖掘创新点？） 3. Sharing and copying of relevant learning materials (book recommendation, online course recommendation, etc.) 4. Establish a WeChat group for later discussion and Q&A |