深度学习

计算机科学与技术系研究生选修课, 厦门大学, 2021

该课程为厦门大学计算机科学与技术系研究生选修课,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。

教材

《Deep Learning》, Ian Goodfellow, Bengio Yoshua, and Courville Aaron, MIT press, 2016.

课程大纲

章节主要内容Notebook
Lecture 1 Introduction to Deep LearningDeep learning applications, impacts, researchers, history, and course description. 
Lecture 2 Basics of Machine LearningBasics of machine learning, linear models, neural networks, back-propagation, model selection, model evaluation.Lecture 2
Lecture 3 Regularization and OptimizationGeneralization, overfitting and underfitting, regularization, optimization for deep models, batch normalization, parameter initialization.Lecture 3
Lecture 4 Hardware and SoftwareDeep learning hardware, PyTorch, TensorFlow, Keras.Lecture 4
Lecture 5 Basics of Convolutional Neural NetworksConvolution, padding, stride, parameter sharing, pooling, common CNN patterns.Lecture 5
Lecture 6 CNN ArchitecturesAlexNet, VGG, GoogLeNet, ResNet, SENet, DenseNet. 
Lecture 7 Basics of Recurrent Neural NetworksRNN, Seq2seq, Attention models, LSTMLecture 7
Lecture 8 Language ModelWord2vec, ELMo, Transformer, BERTLecture 8
Lecture 9 Generative Adversarial NetworksGAN, DCGAN, CGAN, WGAN, SAGAN, pix2pix, CycleGAN, SRGAN 
Lecture 10 Deep Reinforcement LearningMarkov Decision Process, Q-Learning, Deep Q Network, Policy Gradient, Actor-Critic, DDPG.Lecture 10
Lecture 11 Deep Learning on GraphsDeepwalk, LINE, Node2vec, GCN, GraphSAGE, GAT 
Lecture 12 Self-Supervised LearningGeneration-Based Methods, Context-Based Methods, Free Semantic Label-Based Methods, Cross Modal-Based Methods, Contrastive Learning 
Lecture 13 Meta-LearningOptimization-Baesd Method, Model-Based Method, Metric-Based Method, MAML, Few-Shot Learning 
Lecture 14 Deep Learning on Incomplete DataFederated Learning, Long-Tail Learning, Noisy-Label Learning, Continual Learning 
Lecture 15 Advanced Topics in Deep LearningKnowledge Distillation, Adversarial Samples, Model Interpretation, Fairness, Privacy 

大作业展示

标题
3D Shape Analysis Using the CNN
A Semi-Supervised Learning Method for Masks Detection
AcGAN_Acg avatar Generation network
Artificial intelligence_Future for Polyp Diagnosis!
A_Little_Improvement_on_Future_Frame_Prediction_for_Anomaly_Detection_Baseline
Boundary Aware PoolNet for Salient Object Detection
Construction_and_Prediction_of_Antimicrobial_Peptide_Predicition_Model_Based_on_BERT
Cuteness is Justice_Pawpularity Evaluation Based On Deep Learning
Garbage Classification Based on ResNet50
I2Pmatch_Matching KeyPoints Across Image and Point Cloud Based On Feature Learning
Image-Denoising-with-Deep-CNNs UDNCNN
Improved SA-UNet_SA-UNet Based Neural Network for Human Retinal Vessel Segmentation
Intelligent Auxiliary Diagnosis System for Oral Caries Images
Introducing Adversarial Training to Improve the Performance of Model in Natural Language Processing Domain
Lightweight Unbiased Review-based Recommendation Based on Knowledge Distillation
Method For Extracting Named Entities in Policy Documents Based on BERT-BiLSTM-CRF
MixLMI_A Multi-level Enhancement Model for Plant lncRNA-miRNA Interaction Prediction
Multi-stream CNN Based Accented English Automatic Speech Recognition System
Multiple U-Net-based CNNs in Ultrasonic Image Segmentation of hemangioma
Network News Sentiment Analysis Based on BERT
Optical Music Recognition System Based on CNN and RNN
PCIT_A Point Cloud Invariance Transformer
Pseudo Label Guided Unsupervised Meta-Learning for Low-Shot Image Classification
Quantitative Cryptocurrency Trading Based On Proximal Policy Optimization
R2Joint_Robust Real-Time Joint Detection Model for Traffic Scenes
Reimplement of SGG and Further Work
Relation-Aware Multi Channel Attention Based Graph Convolutional Network for Breast Cancer Image Classification
Research on Instance Segmentation via Transformer
Research on Pedestrian Fall Detection Based on YOLO
SAPCS_Semantic Attention Empowered Point Cloud Segmentation
Semantic Learning for Facial Expression Recognition
Simulation and Comparison of Target Detection Algorithm
Small Protein Design based on LSTM RNN
Super-resolution Implementation of Anime Pictures
TAI_TAnktrouble reInforcement learning model based on Deep Q-Networks
The masked face recognition with ArcFace
Training for High-fidelity Few-shot Image Synthesis Based On FastGAN
UAV Relay in VANETs Against Smart Jamming With Deep Reinforcement Learning
Using Review Text and a External Knowledge for Explainable Recommendation

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

WechatIMG118

往年资料

2020