목록NLP (4)
욱이의 냉철한 공부
[논문 요약] RCN : Predict Stock Price with Financial News Based on Recurrent Convolutional Neural Network, 2017
* 논문 Predict Stock Price with Financial News Based on Recurrent Convolutional Neural Network, 2017 1. 입력 데이터 - 입력 데이터 대상 대만증시(1개) : Taiwan Stock Exchange(TWSE) 기업(4개) : TSMC, Hon Hai Precision Industry(Foxconn Technology Group), China Steel Corporation(CSC), Graffe King Bio => 각각 독립적인 모델링 설계 - 입력 데이터 종류 : 일 단위 일 단위의 거래데이터(종가)와 텍스트데이터(뉴스제목) 사용 - 입력 데이터 기간 1. 종가데이터 : 2007-06-21 ~ 2017-02-13, 2,397..
* Word Representation 관점 (Word Embedding) 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 ..
[NLP 개념정리] Word Representation : 카운트 기반 단어표현
* 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..
[NLP 개념정리] Word Embedding : GloVe
* 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..