How ayhanoruc/ml-collection works
Project Type and Purpose
• This is an open‑source, educational “ML‑from‑scratch” toolkit and code repository. Its goal is to provide minimal, self‑contained Python implementations of core machine‑learning and deep‑learning algorithms (linear & logistic regression, MLP, CNN, RNN, LSTM/GRU) plus Jupyter notebooks for hands‑on study.
File‑Structure Analysis
• Top‐level directories:
• data/ – sample datasets and utility scripts
• ml‑from‑scratch/ – basic ML algorithms (linear & logistic regression)
• dl‑from‑scratch/ – deep‑learning builds (subfolders: mlp/, cnn/)
• genai‑from‑scratch/– generative AI RNN/LSTM notebooks & code
• Patterns:
• Each algorithm family lives in its own namespace.
• There are pure‐Python modules (e.g., auto_tensor.py, neural_structures.py) plus Jupyter notebooks demonstrating usage.
• No external services, just local code + data.
README Insights
• Check‑list style shows which components are implemented vs. planned.
• Data Layer:
– ten_animals/ (dataset)
– markdown_images/ (assets)
– translate.py (helper script)
• Compute Layer – ML Core (green rectangle)
• linear_regression.py
• logistic_regression.py
• Compute Layer – DL Core (green rectangle subdivided):
• MLP Submodule: auto_tensor.py → neural_structures.py → trainer.py
• CNN Submodule: utils.py + notebooks (backprop, pooling, modern_architectures…)
• Compute Layer – GenAI Core (green rectangle subdivided):
• RNN Submodule: dataset.py + rnn notebooks
• LSTM/GRU Submodule: lstm.ipynb, gru.ipynb (future scope)
• Presentation Layer:
• All .ipynb demo files, grouped under each core