Notice
Recent Posts
Recent Comments
Link
일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | ||
6 | 7 | 8 | 9 | 10 | 11 | 12 |
13 | 14 | 15 | 16 | 17 | 18 | 19 |
20 | 21 | 22 | 23 | 24 | 25 | 26 |
27 | 28 | 29 | 30 |
Tags
- 표집분포
- Numpy
- Numpy data I/O
- BOXPLOT
- linalg
- subplot
- python 문법
- 정규분포 MLE
- seaborn
- groupby
- 부스트캠프 AI테크
- unstack
- Python
- scatter
- Python 특징
- namedtuple
- 카테고리분포 MLE
- Array operations
- 딥러닝
- Python 유래
- pivot table
- 가능도
- dtype
- VSCode
- ndarray
- Operation function
- type hints
- 최대가능도 추정법
- Comparisons
- boolean & fancy index
Archives
- Today
- Total
또르르's 개발 Story
부스트캠프 AI 테크 포트폴리오 본문
1️⃣ P Stage (프로젝트)
1) 이미지 분류
- 주어진 이미지를 가지고 Mask 착용, Gender, Age 총 18개의 class로 분류하는 문제
- 전체 사람 명 수 : 4,500
- 한 사람당 사진의 개수: 7 [마스크 착용 5장, 이상하게 착용(코스크, 턱스크) 1장, 미착용 1장]
- 이미지 크기: (384, 512)


- Git Repo
Title | Description | Link |
Git Repo | P Stage 01 Git Repo | https://github.com/bcaitech1/p1-img-JAEWOOSUN |
- 프로젝트
Title | Description | Link |
[Stage1-01] | Baseline 작성 | https://dororo21.tistory.com/114 |
[Stage1-02] | Data Processing | https://dororo21.tistory.com/115 |
[Stage1-03] | Model & Optimizer | https://dororo21.tistory.com/119 |
[Stage1-04] | Focal Loss | https://dororo21.tistory.com/121 |
[Stage1-05] | AutoML 사용 (NNI) | https://dororo21.tistory.com/122 |
[Stage1-06] | 최적의 Hyperparameter 찾기 | https://dororo21.tistory.com/123 |
[Stage1-07] | Ensemble 하기 | https://dororo21.tistory.com/124 |
[Stage1-08] | No Augmentation | https://dororo21.tistory.com/125 |
- 이론
Title | Description | Link |
[Stage1-이론] | Dataset & DataLoader | https://dororo21.tistory.com/116 |
[Stage1-이론] | Model | https://dororo21.tistory.com/117 |
[Stage1-이론] | Training & Inference | https://dororo21.tistory.com/118 |
[Stage1-이론] | Ensemble | https://dororo21.tistory.com/120 |
2) KLUE
- 주어진 문장에서 두 개체의 관계 (ex. 인물-직업 등)를 추출하는 task
- train.tsv: 총 9000개answer: 정답 라벨 (비공개)
- test.tsv: 총 1000개 (정답 라벨 blind)
- 총 42개의 classes


- Git Repo
Title | Description | Link |
Git Repo | P Stage 02 Git Repo | https://github.com/bcaitech1/p2-klue-JAEWOOSUN |
- 프로젝트
Title | Description | Link |
[Stage2-01] | EDA | https://dororo21.tistory.com/128 |
[Stage2-02] | 한국어 전처리 / 토크나이징 | https://dororo21.tistory.com/130 |
[Stage2-03] | BERT | https://dororo21.tistory.com/132 |
[Stage2-04] | BERT MASK & 단일 문장 분류 | https://dororo21.tistory.com/133 |
[Stage2-05] | Entity Special token | https://dororo21.tistory.com/140 |
[Stage2-06] | Pororo 라이브러리 사용 | https://dororo21.tistory.com/141 |
[Stage2-07] | xlm-roberta-large 사용하기 | https://dororo21.tistory.com/146 |
[Stage2-08] | 최적의 Hyperparameter 찾기 | https://dororo21.tistory.com/149 |
- 이론
Title | Description | Link |
[Stage2-이론] | 자연어 단어 임베딩 | https://dororo21.tistory.com/127 |
[Stage2-이론] | 한국어 토큰화 | https://dororo21.tistory.com/129 |
[Stage2-이론] | BERT | https://dororo21.tistory.com/131 |
[Stage2-이론] | 의존 구문 분석 | https://dororo21.tistory.com/134 |
[Stage2-이론] | 한국어 BERT 모델 학습하기 | https://dororo21.tistory.com/135 |
[Stage2-이론] | BERT를 활용한 단일 문장 분류 | https://dororo21.tistory.com/136 |
[Stage2-이론] | IRQA 챗봇 실습 | https://dororo21.tistory.com/138 |
[Stage2-이론] | 문장 토큰 관계 분류 task | https://dororo21.tistory.com/139 |
[Stage2-이론] | 문장 토큰 단위 분류 모델 학습 | https://dororo21.tistory.com/142 |
[Stage2-이론] | 기계독해 모델 학습 | https://dororo21.tistory.com/143 |
[Stage2-이론] | GPT 언어 모델 | https://dororo21.tistory.com/144 |
[Stage2-이론] | 한국어 GPT-2 모델 학습하기 | https://dororo21.tistory.com/145 |
[Stage2-이론] | 다양한 NLP 모델들 | https://dororo21.tistory.com/147 |
[Stage2-이론] |
KoGPT-2 챗봇 만들기 | https://dororo21.tistory.com/148 |
3) Dialogue State Tracking (DST)
- 주어진 대화 문장에서 domain-slot에 맞는 value를 추출해내는 task
- train_dials.json: 7000개의 대화 (label 포함)
- public/eval_dials.json: 1000개의 대화 (label 미포함 / public test set)
- private/eval_dials.json: 1000개의 대화 (label 미포함 / private test set)
- ontology.json: Ontology-based DST model을 위한 pre-defined ontology입니다.
- 총 45개의 state가 존재하며, 해당 turn에서 domain-slot에 맞는 value를 찾아야함


- Git Repo
Title | Description | Link |
Git Repo | P Stage 03 Git Repo (Team Project) | https://github.com/bcaitech1/p3-dst-chatting-day |
- 프로젝트
Title | Description | Link |
[Stage3-프로젝트] | Notion / Git Discussion 정리 | https://dororo21.tistory.com/157 |
- 논문 리뷰
Title | Description | Link |
[Stage3-논문 리뷰] | TRADE 논문 리뷰 | https://dororo21.tistory.com/158 |
[Stage3-논문 리뷰] | SOM-DST 논문 리뷰 | https://dororo21.tistory.com/159 |
[Stage3-논문 리뷰] | CHAN-DST 논문 리뷰 | https://dororo21.tistory.com/160 |
[Stage3-논문 리뷰] | Transformer-DST 논문 리뷰 | https://dororo21.tistory.com/161 |
- 이론
Title | Description | Link |
[Stage3-이론] | Task-Oriented Dialogue System | https://dororo21.tistory.com/150 |
[Stage3-이론] | Ontology-based DST Models | https://dororo21.tistory.com/151 |
[Stage3-이론] | Open-Vocab based DST Model | https://dororo21.tistory.com/152 |
[Stage3-이론] | Hybrid Approach | https://dororo21.tistory.com/153 |
[Stage3-이론] | DST의 Computational Complexity | https://dororo21.tistory.com/154 |
[Stage3-이론] | Advanced DST Models | https://dororo21.tistory.com/155 |
[Stage3-이론] | DST의 한계점 | https://dororo21.tistory.com/156 |
4) Deep Knowledge Tracing (DKT)
- 주어진 시계열 정형 데이터에서 다음 문제를 맞출 수 있을 지 예측
- train/test 합쳐서 총 7,442명의 사용자가 존재
- 총 2개 (0/1) Answer code로 되어있으며, Classification 문제


- Git Repo
Title | Description | Link |
Git Repo | P Stage 04 Git Repo (Team Project) | https://github.com/bcaitech1/p4-dkt-no_caffeine_no_gain |
- 프로젝트
Title | Description | Link |
[Stage4-프로젝트] | Notion / Git Discussion 정리 | https://www.notion.so/DKT-10-No_Caffeine_No_Gain-4c2908c79f8f4470a2c87c6d2a7cce08 |
2️⃣ U Stage (이론)
1) Python
Title | Description | Link |
[01] 개발환경 & Python | Computer System & Python 특징 | 이론 : https://dororo21.tistory.com/1 |
[02] Python 기본 문법 | Python의 변수, 함수, 문자열 | 이론 : https://dororo21.tistory.com/2 실습 : https://dororo21.tistory.com/3 |
[03] Python data structure | Python data structure | 이론 : https://dororo21.tistory.com/4 |
[03-1] Pythonic code | Python 특유의 문법 활용하여 효율적 코드 표현 | 이론 : https://dororo21.tistory.com/5 |
[04] Python의 OOP / Module |
Python 객체지향 프로그래밍(OOP)와 모듈 (module) | 이론 : https://dororo21.tistory.com/6 |
[04-1] Python 정규 표현식 | Python의 Regular Expression | 이론 : https://dororo21.tistory.com/7 |
[05] Python Exception | Python의 Exception handling, File handling, Log handling, Data handling | 이론 : https://dororo21.tistory.com/8 |
[06] Python Numpy | Python의 이학/공학 계산용 패키지 | 실습 : https://dororo21.tistory.com/9 |
[06-1] 벡터와 행렬 | Python Numpy를 사용한 벡터와 행렬 | 이론 : https://dororo21.tistory.com/10 |
[08] Python Pandas (1) | Python의 데이터 처리 라이브러리 (1) | 실습 : https://dororo21.tistory.com/12 |
[09] Python Pandas (2) | Python의 데이터 처리 라이브러리 (2) | 실습 : https://dororo21.tistory.com/14 |
[10] Python 시각화 모듈 | Matplotlib, Seaborn 등 시각화 모듈 | 실습 : https://dororo21.tistory.com/16 |
[11-2] PyTorch 기본문법 | Pytorch의 장점과 다양한 문법 | 실습 : https://dororo21.tistory.com/20 |
[13-2] DataSet 다루기 | Colab에서 강아지 Dataset 다루기 실습 | 실습 : https://dororo21.tistory.com/27 |
[13-3] Google Image Data | Goole에서 image data 웹 크롤링 하기 | 실습 : https://dororo21.tistory.com/28 |
2) AI Math
Title | Description | Link |
[07] 경사하강법 | 미분, 편미분, 경사하강법에 대한 설명 | 이론 : https://dororo21.tistory.com/11 |
[08-1] 신경망 (neural network) | 신경망을 수학적으로 분해해보기 | 이론 : https://dororo21.tistory.com/13 |
[09-1] 딥러닝에 사용되는 확률론 | 확률분포, 조건부확률, 몬테카를로 등 | 이론 : https://dororo21.tistory.com/15 |
[10-1] 딥러닝에 사용되는 통계학 | 모수, 가능도 (Likelihood), MLE 등 | 이론 : https://dororo21.tistory.com/17 |
[11] 조건부 확률과 베이즈 정리 | 조건부 확률, 베이즈 정리에 대한 설명 | 이론 : https://dororo21.tistory.com/18 |
[12-1] Convolution 연산 이해하기 | Convolution 연산에 대한 수학적 이해 | 이론 : https://dororo21.tistory.com/23 |
3) DL Basic
Title | Description | Link |
[11-1] Deep Learning의 발전 | Deep Learning 역사의 중요한 알고리즘 | 이론 : https://dororo21.tistory.com/19 |
[11-3] MLP using PyTorch | PyTorch를 사용한 신경망 실습 | 실습 : https://dororo21.tistory.com/21 |
[12] 최적화 종류와 용어들 | Optimization에 대한 용어 정리 | 이론 : https://dororo21.tistory.com/22 실습 : https://dororo21.tistory.com/24 |
[13] CNN | CNN 알고리즘, Semantic Segmentation 등 | 이론 : https://dororo21.tistory.com/25 실습 : https://dororo21.tistory.com/26 |
[14] RNN, LSTM, GRU 정리 (1) | Sequence data, RNN, LSTM, GRU에 대한 이해 | 이론 : https://dororo21.tistory.com/29 실습 : https://dororo21.tistory.com/31 |
[14-1] Transformer 이해하기 (1) | Trasnformer 구조, Self-Attention 등 | 이론 : https://dororo21.tistory.com/30 실습 : https://dororo21.tistory.com/32 |
[15] Generative Models | 생성모델 (GAN)에 대한 이해 | 이론 : https://dororo21.tistory.com/33 |
4) NLP
Title | Description | Link |
[16] Bag-of-words / Word Embedding |
NLP 역사, 단어를 mapping 해주는 방법 | 이론 : https://dororo21.tistory.com/35 실습 (1) : https://dororo21.tistory.com/36 실습 (2) : https://dororo21.tistory.com/37 실습 (3) : https://dororo21.tistory.com/38 |
[17] RNN, LSTM, GRU 정리 (2) | Sequence data, RNN, LSTM, GRU에 대한 이해 | 이론 : https://dororo21.tistory.com/39 실습 (1) : https://dororo21.tistory.com/40 실습 (2) : https://dororo21.tistory.com/41 |
[18] Seq2Seq Model | Sequence to Sequence 모델에 대한 이해 | 이론 : https://dororo21.tistory.com/43 실습 : https://dororo21.tistory.com/45 |
[18-1] Beam Search와 BLUE | 자연어 생성 Decoding 방법에 대한 이해 | 이론 : https://dororo21.tistory.com/44 실습 : https://dororo21.tistory.com/46 |
[19] Transformer 이해하기 (2) | Self-Attention, Multi-head attention 등 | 이론 : https://dororo21.tistory.com/47 실습 (1) : https://dororo21.tistory.com/49 실습 (2) : https://dororo21.tistory.com/50 |
[19-1] Byte Pair Encoding | 기본적인 Sub-word tokenization | 실습 : https://dororo21.tistory.com/48 |
[20] Pre-Training Models | GPT, BERT, ELECTRA 등 범용 NLP 모델 | 이론 : https://dororo21.tistory.com/51 실습 (1) : https://dororo21.tistory.com/53 실습 (2) : https://dororo21.tistory.com/55 |
[29] GLUE와 KLUE | 영어와 한국어의 NLP 평가지표 비교 | 이론 : https://dororo21.tistory.com/74 |
5) Graph
Title | Description | Link |
[21] 그래프 (Graph 개념) | 그래프에 대한 정의와 군집 | 이론 : https://dororo21.tistory.com/56 실습 : https://dororo21.tistory.com/57 |
[22] 페이지랭크 (PageRank) | 페이지랭크에 대한 이해 | 이론 : https://dororo21.tistory.com/58 실습 : https://dororo21.tistory.com/60 |
[22-1] 그래프 전파 모형 | 선형 임계치 모형, 독립 전파 모형 등 | 이론 : https://dororo21.tistory.com/59 |
[23] 군집 구조 | 군집, 군집성, 군집 탐색 알고리즘 등 | 이론 : https://dororo21.tistory.com/61 실습 : https://dororo21.tistory.com/63 |
[23-1] 추천시스템 | Graph를 활용한 추천 시스템에 대한 이해 | 이론 : https://dororo21.tistory.com/62 실습 : https://dororo21.tistory.com/64 |
[24] 정점 임베딩 | 정점들을 vector로 표현하는 방법 | 이론 : https://dororo21.tistory.com/65 실습 : https://dororo21.tistory.com/67 |
[24-1] 넷플릭스 추천시스템 | 넷플릭스에서 사용하는 추천시스템 | 이론 : https://dororo21.tistory.com/66 실습 : https://dororo21.tistory.com/68 |
[25] 그래프 신경망 | 신경망을 사용한 그래프 이론 | 이론 : https://dororo21.tistory.com/69 실습 : https://dororo21.tistory.com/71 |
6) Computer Vision
Title | Description | Link |
[31] Image Classification (1) | K-NN, CNN 등에 대한 이해 | 이론 : https://dororo21.tistory.com/80 실습 : https://dororo21.tistory.com/82 |
[31-1] Annotation 방법들 | 적은 수의 데이터로 효율적으로 학습하는 방법 | 이론 : https://dororo21.tistory.com/81 |
[31-3] Scratch vs Fine-tuning | 처음부터 학습 vs Fine-tuning의 성능 비교 | 실습 : https://dororo21.tistory.com/83 |
[32] Image Classification (2) | Image Calssification 알고리즘에 대한 이해 | 이론 : https://dororo21.tistory.com/85 |
[32-1] Semantic Segmentation | Semantic Segmentation 정의와 알고리즘 | 이론 : https://dororo21.tistory.com/86 실습 : https://dororo21.tistory.com/87 |
[33] Object detection | Object Detection 정의와 알고리즘 | 이론 : https://dororo21.tistory.com/88 |
[33-1] CNN visualization | CNN의 weight를 시각화하는 방법 | 이론 : https://dororo21.tistory.com/89 실습 : https://dororo21.tistory.com/94 |
[33-2] Pytorch Autograd | Pytorch에서 제공하는 Automatic gradient | 이론 : https://dororo21.tistory.com/90 |
[34-1] Instance/Panoptic segmentation |
Instance/Panoptic segmentation에 대한 이해와 알고리즘 | 이론 : https://dororo21.tistory.com/91 실습 : https://dororo21.tistory.com/95 |
[34-1] Conditional GAN | Conditional input을 추가한 GAN 모델 | 이론 : https://dororo21.tistory.com/92 |
[35] Multi-modal | Multi-modal 정의와 알고리즘 | 이론 : https://dororo21.tistory.com/96 |
[35-1] 3D task | CV가 3D에 쓰이는 task들에 대한 설명 | 이론 : https://dororo21.tistory.com/97 |
7) 모델 경량화
Title | Description | Link |
[36] 결정기 | Decision making machine에 대한 이해 | 이론 : https://dororo21.tistory.com/99 |
[36-1] Model Conversion | 다양한 딥러닝 라이브러리로 바꾸는 방법 | 실습 : https://dororo21.tistory.com/100 |
[37] 딥러닝 관점의 Entropy | Entropy에 대한 다양한 관점 | 이론 : https://dororo21.tistory.com/101 |
[37-1] Hyperparameter Search | Hyperparameter를 찾는 여러 방법들 | 이론 : https://dororo21.tistory.com/102 |
[38] DL Compiler Acceleration | DL 알고리즘을 가속화하는 방법들 | 이론 : https://dororo21.tistory.com/104 실습 : https://dororo21.tistory.com/106 |
[38-1] 경량화 (1) - Pruning | Pruning 정의와 방법 | 이론 : https://dororo21.tistory.com/105 실습 : https://dororo21.tistory.com/107 |
[39] 경량화 (2) - 양자화 | Quantization 정의와 방법 | 이론 : https://dororo21.tistory.com/108 실습 : https://dororo21.tistory.com/110 |
[39-3] 경량화 (3) - 지식 증류 | Teacher-Student 모델의 정의와 방법 | 실습 : https://dororo21.tistory.com/111 |
[40] 경량화 (4) - 행렬분해 | 행렬분해 정의와 방법 | 이론 : https://dororo21.tistory.com/112 |
8) Kaggle 대회
Title | Description | Link |
[17-3] CNN을 사용한 Kaggle 데이터 분류 |
피어세션 팀원들과 U Stage 기간동안 참여한 대회 (종료된 대회) |
https://dororo21.tistory.com/42 |
Comments