Neural Network Playground
مثال توضيحي مصغّر — يُظهر كيف تتعلم الشبكة العصبية من الصفر. مشاريعنا الحقيقية أكبر وأكثر تعقيداً.
معاينة مباشرة — تفاعلية
Neural Network Playground
Deep Learning - Neural networks learn from scratch (XOR problem)
Ready
0.50
Network ArchitectureEpoch: 0
1.0000
MSE Loss
0
Epochs
0/4
XOR Correct
Loss Curve (MSE)
XOR Truth Table — Predictions
| x | y | Target | Output | Status |
|---|---|---|---|---|
| 0 | 0 | 0 | ... | ✓ |
| 0 | 1 | 1 | ... | ✗ |
| 1 | 0 | 1 | ... | ✗ |
| 1 | 1 | 0 | ... | ✓ |
How does this Deep Learning work?
- Problem: XOR is a non-linear function that a simple Perceptron cannot learn — it requires a hidden layer.
- Forward pass: Inputs (x, y) → multiply by weights → add bias → Sigmoid activation → output.
- Loss calculation: Error = target - output, then MSE = Error².
- Backward pass: Compute gradient of the loss with respect to each weight (chain rule).
- Update weights: w_new = w_old - learning_rate × gradient.