How deepmancer/resnet-cifar-classification works
This repository is a pair of notebook-based image-classification experiments. The documented primary workflow trains a from-scratch ResNet-18 on CIFAR-10, then examines penultimate-layer features with t-SNE and k-NN. The tree also contains a separate simple-CNN classification notebook. Source excerpts were not available, so the graph follows README-documented relationships and avoids asserting notebook-internal call order beyond those relationships.
Main entry points: CIFAR10_Classification.ipynb ResNet Notebook · Images Feature Visualizations · Image_Classification_CNN.ipynb Simple CNN Notebook
Components 11
Classification Experiments
- ResNet Notebook
ResNet-18/CIFAR10_Classification.ipynb - Simple CNN Notebook
Simple_Conv/Image_Classification_CNN.ipynb - CIFAR-10
- ResNet-18 Model
ResNet-18/CIFAR10_Classification.ipynb - CNN Classifier
Simple_Conv/Image_Classification_CNN.ipynb
Feature Analysis
- Penultimate Features
ResNet-18/CIFAR10_Classification.ipynb - t-SNE Projection
ResNet-18/CIFAR10_Classification.ipynb - k-NN Analysis
ResNet-18/CIFAR10_Classification.ipynb
Visual Outputs
- Feature Visualizations
Images
Other components
- Notebook User
- PyTorch
Connections 10
- Notebook User → ResNet Notebook: runs
- Notebook User → Simple CNN Notebook: runs
- ResNet Notebook → CIFAR-10: trains on
- ResNet Notebook → ResNet-18 Model: implements and trains
- ResNet Notebook → PyTorch: uses
- ResNet Notebook → Penultimate Features: explores
- Penultimate Features → t-SNE Projection: projects
- Penultimate Features → k-NN Analysis: analyzes
- t-SNE Projection → Feature Visualizations: visualizes in
- k-NN Analysis → Feature Visualizations: visualizes in