🔨 Active Development Notice
This repository is under continuous improvement. Recent updates include [Data Mining].Feel free to open an issue for suggestions!
A modular tech stack navigation for deep learning self-learners, implementing classical/cutting-edge models across seven core domains to build a closed-loop understanding from theory to practice.
Each project includes:
- 🧠 Model Architecture Analysis
- ⚡ Minimalist Implementation Code
- 📈 Training/Inference Scripts
Exclusions:
- 📚 Dataset Download Links (Competition/Undisclosed Data)
- 📦 Environment Configuration (Long-term Spanning Multiple Environments)
- Data Processing:
Pandas
/Seaborn
/ydata-profiling
- Predictive Models:
- EV Charging Demand Prediction (
LightGBM
/KFold
/LSTM
/ARIMA
) - Li-ion Battery Temp Prediction (
FNN
/XGBoost
/Stacking
) - Quantum Chemistry Molecular Property Prediction (
Joblib
/TF-IDF
/CatBoost
)
- EV Charging Demand Prediction (
- Classification:
- Diabetes Risk Prediction (
CatBoost
) - CNS Drug Prediction (
QSAR
/LightGBM
)
- Diabetes Risk Prediction (
- Regression:
- House Price Prediction (
Linear Regression
) - Taiyuan Heating Prediction (
SVR
)
- House Price Prediction (
- Classification:
- Titanic Survival Prediction (
Decision Tree
/Random Forest
) - Advertisement Classification (
PCA
/XGBoost
)
- Titanic Survival Prediction (
- Clustering:
- Airline Customer Segmentation (
KMeans
/KMeans++
/DBSCAN
)
- Airline Customer Segmentation (
- Image Processing:
- Image Denoising (
Markov
)
- Image Denoising (
- Financial:
- Stock Prediction (
LSTM
/GRU
) - Used Car Price Prediction (
CNN-LSTM
)
- Stock Prediction (
- Text Generation:
- Acrostic Poetry Generation (
LSTM
) - Shakespeare-style Generation (
LSTM
)
- Acrostic Poetry Generation (
- Environmental:
- Weather Forecasting (
BiLSTM
)
- Weather Forecasting (
- Pretrained Models:
- Grammar Correction/Sentiment Analysis (
BERT
/HuggingFace
)
- Grammar Correction/Sentiment Analysis (
- Traditional Methods:
- Tweet Sentiment Recognition (
n-gram
) - Shakespeare Text Mining (
Word2Vec
)
- Tweet Sentiment Recognition (
- Classification:
- News Categorization (
ELMo
)
- News Categorization (
- Medical Imaging:
- Brain PET Diagnosis (
ResNet
/VGG
) - Skin Lesion Segmentation (
UNet
/UNet++
)
- Brain PET Diagnosis (
- Object Detection:
- Fire Detection (
YOLOv5
) - Infrared Small Target Detection (
YOLOv8
)
- Fire Detection (
- Image Generation:
- Fashion MNIST Inpainting (
VAE
) - Style Transfer (
VGG
)
- Fashion MNIST Inpainting (
- Other Model Analyses
- Cross-modal Retrieval:
- Autonomous Driving Video Understanding (
CLIP
) - Image Search System (
CLIP
/Milvus
)
- Autonomous Driving Video Understanding (
- Generation:
- Stable Diffusion Analysis (
BLIP2
)
- Stable Diffusion Analysis (
- Education:
- AI Teaching Assistant (
LangChain
/RAG
/CoT
) - Paper Writing Assistant (
LoRA
/vLLM
)
- AI Teaching Assistant (
- Industry:
- Digital Tourism (
LazyGraphRAG
/CogVideoX
) - Tourism Recommendation (
RAG
/Tools
) - Immersive Tourism RPG (
MetaGPT
)
- Digital Tourism (
- Optimization:
- DeepSeek-R1 4090 Deployment (
Unsloth
/LoRA
) - Llama2 from Scratch
- DeepSeek-R1 4090 Deployment (
- AIoT:
- Smart Home (
RAG
/CoT
/blockly
)
- Smart Home (
- Model Analyses:
- Qwen (
Qwen2
/Qwen2.5
) - GPT (
GPT1-4
) - Llama (
Llama3
/Llava
) - DeepSee (
DeepSeek v3
/DeepSeek R1
) - Others
- Qwen (
- Generative Models:
- CIFAR10 Training (
Diffusion
) - GAN Implementation
- CIFAR10 Training (
- Speech:
- ChatTTS Generation
- EchoMimic Deployment
- Graph Networks:
- GCN Molecular Property Prediction
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License