🧠 [2025-05-30] Revisiting CNN Fundamentals with CIFAR-10

Today, I didn’t have time to work on the Real-Time Activity Recognizer.
So I focused solely on reinforcing my basic CNN knowledge using the CIFAR-10 dataset.

🧪 CIFAR-10 CNN Classifier (Book-based Practice)

As part of deepening my CNN fundamentals, I went back to basics using a guided book tutorial on CIFAR-10.

✅ What I Did

• Loaded CIFAR-10 via keras.datasets
• Converted labels with to_categorical
• Built a CNN with:
  • Conv2D, MaxPooling2D, Flatten, Dense, and Dropout
• Practiced with different layer settings and resolved:
  • Syntax mistakes (e.g., input_shape=(...))
  • Typos like y_trian → y_train

🧠 Key Takeaways

• Practicing on simpler datasets like CIFAR-10 helps build layer-wise intuition
• Debugging small bugs enhances attention to detail and logic
• This practice serves as a solid foundation for more complex CV models

🎯 Next Steps

• ⏭️ Resume Real-Time Activity Recognizer training tomorrow
• 📊 Compare baseline CNN vs MobileNetV2 performance on same-class subsets
• 📁 Keep logs and commit updated notebook versions on GitHub

Small step today, but important.
Deep learning foundations matter — and this was a good brush-up session.