욱이의 냉철한 공부

* 논문 Deep Learning for Stock Prediction Using Numerical and Textual Information, 2016 * 목차 1. 입력데이터 2. 출력데이터 3. 모델링 4. Trading simulation 5. contribution 6. 한계점 0. 방향성 Many of previous works used only one of textual, numerical, or image information for stock price prediction, and their model was trained with consideration about a single company. Nevertheless, it is desirable for the prediction m..

* Word Representation 분류체계 1. Discrete Representation : Local Representation 1) One - hot Vector - One - hot Vector 2) Count Based - Bag of Words (BoW) - Document-Term Matrix (DTM) - (TDM) - Term Frequency-Inverse Document Frequency (TF - IDF) - N-gram Language Model (N-gram) ​ 2. Continuous Representation 1) Prediction Based (Distributed Representation) - Neural Network Language Model (NNLM) or..

* 자료출저 및 참고논문 - 논문 GloVe : 2014, Global Vectors for Word Representation * Word Representation 분류체계 1. Discrete Representation : Local Representation 1) One - hot Vector - One - hot Vector 2) Count Based - Bag of Words (BoW) - Document-Term Matrix (DTM) - (TDM) - Term Frequency-Inverse Document Frequency (TF - IDF) - N-gram Language Model (N-gram) ​ 2. Continuous Representation 1) Prediction Base..

* 자료출저 및 참고논문 - 강의 Coursera, Andrew Ng 교수님 인터넷 강의 - 논문 Word2Vec: 2013, Effcient Estimation of Word Representations in Vector Space * Word Representation 관점 : Word Embedding 만들기 1. Discrete Representation : Local Representation 1) One - hot Vector - One - hot Vector 2) Count Based - Bag of Words (BoW) - Document-Term Matrix (DTM) - (TDM) - Term Frequency-Inverse Document Frequency (TF - IDF) - N-..

* 자료출저 및 참고논문 - 강의 Coursera, Andrew Ng 교수님 인터넷 강의 - 논문 NPLM : A Neural Probabilistic Language Model * Word Representation 분류체계 1. Discrete Representation : Local Representation 1) One - hot Vector - One - hot Vector 2) Count Based - Bag of Words (BoW) - Document-Term Matrix (DTM) - (TDM) - Term Frequency-Inverse Document Frequency (TF - IDF) - N-gram Language Model (N-gram) ​ 2. Continuous Repre..