How aryamanstha/home-assignment-neural-network-and-deep-learning works
This repository is a Jupyter-based neural-network assignment, not a packaged application. Its documented workflow runs tensor and loss demonstrations, trains MNIST classifiers, compares optimizers, and logs a separate model’s training metrics for TensorBoard inspection. The notebook itself was not included in the sampled source, so the connections below follow README descriptions; implementation details and exact call wiring cannot be verified.
Main entry points: app.ipynb Image Normalization · fit Event Logs
Components 13
Notebook Exercises
- Tensor Operations
app.ipynb - Loss Comparison
app.ipynb
MNIST Training
- MNIST Dataset
- Image Normalization
app.ipynb - CNN Models
app.ipynb - Adam and SGD
app.ipynb - Accuracy Curves
app.ipynb
Training Monitoring
- Feedforward Model
app.ipynb - TensorBoard Logging
app.ipynb - Event Logs
logs/fit - Overfitting Analysis
app.ipynb
Other components
- Learner
- TensorBoard Viewer
Connections 12
- Learner → Tensor Operations: runs exercise
- Learner → Loss Comparison: runs comparison
- Learner → MNIST Dataset: loads dataset
- MNIST Dataset → Image Normalization: provides images
- Image Normalization → CNN Models: feeds images
- CNN Models → Adam and SGD: trains with
- Adam and SGD → Accuracy Curves: compares accuracy
- Learner → Feedforward Model: trains model
- Feedforward Model → TensorBoard Logging: uses callback
- TensorBoard Logging → Event Logs: writes metrics
- TensorBoard Viewer → Event Logs: reads metrics
- TensorBoard Viewer → Overfitting Analysis: supports inspection