Imbalanced Data를 다루기 위해 SMOTE 그만 사용하기
[공지사항] “출처: https://syslog.ravelin.com/classification-with-tabnet-deep-dive-49a0dcc8f7e8”
Category
ML
TabNet으로 하는 Classification: Deep Dive
[공지사항] “출처: https://syslog.ravelin.com/classification-with-tabnet-deep-dive-49a0dcc8f7e8”
Scikit-learn으로 ML 모델링부터 ML Pipeline까지
[공지사항] “출처: https://towardsdatascience.com/from-ml-model-to-ml-pipeline-9f95c32c6512”
Hyperparameter Optimization Framework인 Optuna 구현 소개
[공지사항] “출처: https://medium.com/optuna/an-introduction-to-the-implementation-of-optuna-a-hyperparameter-optimization-framework-33995d9ec354”
어떤 features가 Overfitting이 되는 걸까요?
[공지사항] “출처: https://towardsdatascience.com/which-of-your-features-are-overfitting-c46d0762e769”
Random Forest 아니면 XGBoost? 이제는 Local Cascade Ensemble (LCE)를 살펴봐야 할 시간입니다
[공지사항] “출처: https://towardsdatascience.com/random-forest-or-xgboost-it-is-time-to-explore-lce-2fed913eafb8”
Fraud Machine Learning Modelling을 돕기 위한 합성 데이터
[공지사항] “출처: https://towardsdatascience.com/synthetic-data-to-help-fraud-machine-learning-modelling-c28cdf04e12a”
단 2줄의 코드로 machine learning model 40개 적용하기
[공지사항] “출처: https://medium.com/@fareedkhandev/apply-40-machine-learning-models-in-two-lines-of-code-c01dad24ad99”
Boruta SHAP: 모든 data scientist가 알아야 할 feature selection을 위한 도구
출처: https://towardsdatascience.com/boruta-shap-an-amazing-tool-for-feature-selection-every-data-scientist-should-know-33a5f01285c0
SHAP 값을 사용하여 machine learning 모델 작동 방식 설명
출처: https://towardsdatascience.com/using-shap-values-to-explain-how-your-machine-learning-model-works-732b3f40e137
Feature selection 알고리즘인 Boruta algorithm에 대한 설명
출처: https://towardsdatascience.com/boruta-explained-the-way-i-wish-someone-explained-it-to-me-4489d70e154a
Class imbalance를 다루기 위해 SMOTE 그만 사용하기
Stop Using SMOTE to Treat Class Imbalance
Automated Feature Engineering in Python
[공지사항] “출처: https://towardsdatascience.com/automated-feature-engineering-in-python-99baf11cc219”
Interpreting Interactions in Regression
[공지사항] “출처: https://www.theanalysisfactor.com/interpreting-interactions-in-regression/”
LIGHTGBM 이란? 그리고 PARAMETER 튜닝하기
[공지사항] “출처: https://medium.com/@pushkarmandot/https-medium-com-pushkarmandot-what-is-lightgbm-how-to-implement-it-how-to-fine-tune-the-parameters-60347819b7f...
Binary Encoding과 Target-based Encoding
“출처: http://www.saedsayad.com/encoding.htm”
XGBoost란
“출처: https://dacon.io/competitions/open/235698/talkboard/404176”
AdaBoost, ExtraTrees, GradientBoosing, LDA
AdaBoost (Adaptive Boosing)
cross validation, KNN, SVM, DecisionTree, NaiveBayes, 앙상블 기법, 비지도 학습
K-폴드 교차 검증(K-fold cross validation) 학습 데이터의 일정 부분을 검증데이터로 쓰되, n번의 검증 과정을 통해 학습 데이터의 모든 데이터를 한 번씩 검증데이터로 사용하는 방식 Training Set, Validation Set, Test Set의 ...
첫 포스팅
[공지사항] “첫 공지사항”
Kaggle
[Feedback Prize - English Language Learning] Speeding up Transformer w/ Optimization Strategies
Introduction state-of-the-art 또는 SOTA model을 훈련시키기 위해서는 GPU가 절대적으로 필요합니다. 그리고 Google Colab이나 Kaggle에서 사용할 수 있다고 해도 메모리 제약 문제가 여전히 발생합니다.
[Feedback Prize - English Language Learning] SWA, Apex AMP & Interpreting Transformers in Torch
SWA-LP & Interpreting Transformer Interactively
[Feedback Prize - English Language Learning] DeBERTa LayerwiseLR LastLayerReinit in TensorFlow
DeBERTa LLRD + LastLayerReinit with TensorFlow MultilabelStratifiedKFold split of the data
[Home Credit Default Risk] Tuning Automated Feature Engineering (Exploratory)
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-tuning-automated-feature-engineering”
[Home Credit Default Risk] Automated Feature Engineering Basics
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-automated-feature-engineering-basics”
[Home Credit Default Risk] Introduction: Manual Feature Engineering (part two)
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-manual-feature-engineering-pt2”
[Home Credit Default Risk] Introduction to Manual Feature Engineering
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-introduction-manual-feature-engineering”
[Home Credit Default Risk] Start Here: A Gentle Introduction
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-start-here-a-gentle-introduction”
[Porto Seguro] Porto Seguro Exploratory Analysis and Prediction
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-exploratory-analysis-and-prediction”
[Porto Seguro] XGBoost_CV(LB.284)
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-xgboost-cv-lb-284”
[Porto Seguro] Interactive Porto Insights - A Plot.ly Tutorial
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-interactive-porto-insights-plot-ly”
[Porto Seguro] Data Preparation Exploration
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-data-preparation-exploration”
[Titanic] Introduction to Ensembling Stacking
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-introduction-to-ensembling-stacking”
[Titanic] Top 4% with ensemble modeling
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-titanic-top-4-with-ensemble-modeling”
[Titanic] EDA To Prediction(DieTanic)
[공지사항] “출처: https://www.kaggle.com/code/jundthird/kor-eda-to-prediction”
Papers
[논문] DeBERTa: Decoding-enhanced BERT with Disentangled Attention (2021)
출처: https://arxiv.org/abs/2006.03654
[논문] RoBERTa: A Robustly Optimized BERT Pretraining Approach (2019)
출처: https://arxiv.org/abs/1907.11692
[논문] Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks (2019)
출처: https://arxiv.org/abs/1908.10084
[논문] BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding (2019)
출처: https://arxiv.org/pdf/1810.04805.pdf
[논문] Deep Feature Synthesis: Towards Automating Data Science Endeavors (2015)
출처 : https://dm-gatech.github.io/CS8803-Fall2018-DML-Papers/deep-feature-synthesis.pdf
[논문] NEURAL MACHINE TRANSLATION BY JOINTLY LEARNING TO ALIGN AND TRANSLATE (2016)
NEURAL MACHINE TRANSLATION BY JOINTLY LEARNING TO ALIGN AND TRANSLATE (2016)
[논문] Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates (2018)
Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates (2018)
[논문] Google’s Neural Machine Translation System: Bridging the Gap between Human and Machine Translation (2016)
Google’s Neural Machine Translation System: Bridging the Gap between Human and Machine Translation (2016)
[논문] SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing (2018)
SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing (2018)
[논문] Neural Machine Translation of Rare Words with Subword Units (2016)
Neural Machine Translation of Rare Words with Subword Units (2016)
[논문] Bidirectional LSTM-CRF Models for Sequence Tagging (2015)
Bidirectional LSTM-CRF Models for Sequence Tagging (2015)
DL
트랜스포머를 이용한 한국어 챗봇(Transformer Chatbot Tutorial)
[공지사항] “출처: https://wikidocs.net/89786”
트랜스포머(Transformer) 설명
[공지사항] “출처: https://wikidocs.net/31379”
논문에서 설명하지 않은 Transformer Details
[공지사항] “출처: https://tunz.kr/post/4”
BLEU Score(Bilingual Evaluation Understudy Score)
[공지사항] “출처: https://wikidocs.net/31695”
[tensorflow] 센텐스피스(SentencePiece)
[공지사항] “출처: https://wikidocs.net/86657”