Task Success Classifier
Objective
Mengklasifikasikan apakah sebuah tugas:
- success
- partial
- failed
berdasarkan metrik performa eksekusi.
Input Features
- task_complexity_level
- execution_time_sec
- energy_consumption_joule
- joint_error_margin_deg
- sensor_noise_index
- retry_count
Output
- task_success
Architecture
- Feature Normalization
- Dense(256) + ReLU
- Dropout(0.3)
- Dense(128) + ReLU
- Dense(3) + Softmax
Loss
Categorical Crossentropy
Optimizer
Adam
Metrics
- Accuracy
- F1 Score
- Confusion Matrix
Use Case
- Failure detection system
- Autonomous performance monitoring