Deep Learning

• 1-Neural Network & optimization
• 2-Structured Machine Learning
• 3-Convolutional Neural Network
• 3-ResNet vs Inception network
• 4-Recurrent Neural Networks
• 4-word2vec
• 4-seq2seq

Keras

• 1-First Look at NN
• 2.1-One-hot&word-embedding in keras
• 2.2-keras RNN
• 2.3-keras analyze Weather sequence data(dropout,fit_generator)
• 2.4-1d convnet for nlp
• 3-keras functional api
• 4-keras callback monitor & tensorboard visualization
• 5-RNN text generation
• 6-Sentiment Analysis in Shopping review(gensim word embedding&RNN)

Tensorflow

• 1-low api introduce
• 2-Variable
• 3-graph & session
• 4-QueueRunner read data

NLP

• HMM Model

  • 1-HMM model
  • 2-HMM Observed sequence probability & parameter estimation & state sequence estimation

• LDA

  • LDA gibbs sampling
  • gensim LDA topic distribution(drop stop words,invisible character,convert SBC-case to DBC-case)

• 3.1 聚类与检索

  • 1-聚类与检索(KNN,TF-IDF,KD-Tree,K-Means,EM
  • 2-LDA与混合关系模型
  • 3-Hierarchical Clustering

Recommand System

• 1-Conception
• 2-Content-based Recommendation
  • 2.1 Keyword Extraction(Tf-IDF&TextRank)
  • 2.2 Entity Identification(HMM)
  • 2.3 fasttext文本分类(对比CBOW和skip-gram)
• 3-Collaborative Filtering
• 4.1-Matrix Factorization-explicit feedback(SVD&SVD++)
• 4.2-Matrix Factorization-implicit feedback(Bayes Personalization Rank)
• 5.1-Ranking-[GBDT+LR]
• 5.2-Ranking-[FM&FMM]
• 5.3-Ranking-[Wide&Deep Model]
• 6.1-MAB problem-[Thompson sampling&UCB&epsilon greed]
• 6.2-MAB problem-[LinUCB(contextual bandit)]
• [6.3-MAB problem-[cofiba(collaborative filtering)]]
• 7-Build scientific ranking list
• 8-drop duplicate in candidate pool

时间序列

• LSTM预测正弦波时序数据