深度学习
计算机科学与技术系研究生选修课, 厦门大学, 2023-09
该课程为厦门大学信息学院/人工智能研究院研究生选修课，以基于深度学习平台TensorFlow和PyTorch的编程实践为主，算法理论为辅，使学生能够领悟深度学习的基本原理以及适用场景，并且对使用深度学习方法来解决问题具有一定的动手能力，为学生今后开展科研工作和业界求职打下基础。
课程大纲
| 章节 | 主要内容 | Notebook | 视频链接 |
|---|---|---|---|
| Lecture 1 Introduction to Deep Learning | Deep learning applications, impacts, researchers, history, and course description. | Lecture 1 | |
| Lecture 2 Basics of Machine Learning | Basics of machine learning, linear models, neural networks, back-propagation, model selection, model evaluation. | Lecture 2 | Lecture 2 |
| Lecture 3 Regularization and Optimization | Generalization, overfitting and underfitting, regularization, optimization for deep models, batch normalization, parameter initialization. | Lecture 3 | Lecture |
| Lecture 4 Hardware and Software | Deep learning hardware, PyTorch, TensorFlow, Keras. | Lecture 4 | Lecture 4 |
| Lecture 5 Basics of Convolutional Neural Networks | Convolution, padding, stride, parameter sharing, pooling, common CNN patterns. | Lecture 5 | Lecture 5 |
| Lecture 6 CNN Architectures | AlexNet, VGG, GoogLeNet, ResNet, SENet, DenseNet. | Lecture 6 | |
| Lecture 7 Basics of Recurrent Neural Networks | RNN, Seq2seq, Attention models, LSTM | Lecture 7 | Lecture 7 |
| Lecture 8 Language Model | Word2vec, ELMo, Transformer, BERT | Lecture 8 | Lecture 8 |
| Lecture 9 Deep Reinforcement Learning | Markov Decision Process, Q-Learning, Deep Q Network, Policy Gradient, Actor-Critic, DDPG. | Lecture 9 | |
| Lecture 10 Generative Models | GAN, DCGAN, CGAN, WGAN, SAGAN, pix2pix, CycleGAN, SRGAN | Lecture 10 | Lecture 10 |
| Lecture 11 Deep Learning on Graphs | Deepwalk, LINE, Node2vec, GCN, GraphSAGE, GAT | Lecture 11 | |
| Lecture 12 Self-Supervised Learning | Generation-Based Methods, Context-Based Methods, Free Semantic Label-Based Methods, Cross Modal-Based Methods, Contrastive Learning | Lecture 12 | |
| Lecture 13 Meta-Learning | Optimization-Baesd Method, Model-Based Method, Metric-Based Method, MAML, Few-Shot Learning | Lecture 13 | |
| Lecture 14 Deep Learning on Incomplete Data | Federated Learning, Long-Tail Learning, Noisy-Label Learning, Continual Learning | Lecture 14 | |
| Lecture 15 Advanced Topics in Deep Learning | Knowledge Distillation, Adversarial Samples, Model Interpretation, Fairness, Privacy | Lecture 15 |
大作业展示
参考资料
本课程的课件参考了许多著名的深度学习课程，非常感谢这些课程的教授对课件进行无私的分享。
CS231n: Convolutional Neural Networks for Visual Recognition, Stanford University
CS224n: Natural Language Processing with Deep Learning, Stanford University
CS224w: Machine Learning with Graphs, Stanford University
CMSC 35246: Deep Learning, University of Chicago
Introduction to Reinforcement Learning with David Silver, DeepMind
MGMTMSA-434: Advanced Workshop on Machine Learning, UCLA
期末合影
