| ---
|
| language: en
|
| tags:
|
| - ad-creative
|
| - ctr-prediction
|
| - survival-analysis
|
| - multi-task-learning
|
| - clip
|
| - fatigue-prediction
|
| license: mit
|
| ---
|
|
|
| # Creative Intelligence Scorer
|
|
|
| **Multi-Task Creative Lifespan Prediction** — predicts ad creative CTR score and
|
| fatigue half-life from a raw image using a frozen CLIP backbone and a trainable
|
| multi-task head.
|
|
|
| ## Architecture
|
|
|
| ```
|
| Input image (224×224 RGB)
|
| ↓
|
| [FROZEN] CLIP-ViT-B/32 (openai/clip-vit-base-patch32)
|
| ↓ 512-dim embedding
|
| Projection: Linear(512→256) → ReLU → Dropout(0.2)
|
| ↓ 256-dim shared representation
|
| ┌─────────────────────┐
|
| ↓ ↓
|
| CTR head Fatigue head
|
| Linear(256→1) Linear(256→2)
|
| Sigmoid Weibull params (log_scale, log_shape)
|
| ```
|
|
|
| Loss = 0.5 × BCELoss(ctr) + 0.5 × WeibullNLLLoss(fatigue, right-censored)
|
|
|
| ## Training data
|
|
|
| - Meta Ad Library (Apify scrape): 3,502 real ad images — gaming, ecommerce, finance verticals
|
| - PIL-generated synthetic ads: 18,746 images with rule-based CTR and half-life labels
|
| - Total: 22,248 images | 80/10/10 train/val/test split
|
|
|
| ## Metrics (test set)
|
|
|
| | Metric | Value | Target |
|
| |--------|-------|--------|
|
| | Spearman r (CTR ranking) | TBD | > 0.30 |
|
| | MAE (CTR calibration) | TBD | < 0.15 |
|
|
|
| ## Limitations
|
|
|
| - **CTR labels are proxy scores**, not real click-through rates — derived from ad
|
| activity signals, not A/B test data.
|
| - **GradCAM is a spatial approximation** — CLIP's pooler_output discards spatial
|
| structure; the 16×16 heatmap is gradient-weighted feature attribution on the
|
| projection layer, not true spatial GradCAM.
|
| - Trained on a dataset with known label imbalance (wear-out >> cut-out).
|
|
|
| ## Intended use
|
|
|
| Portfolio project demonstrating Multi-Task Creative Lifespan Prediction for ad
|
| creative scoring. Not intended for production ad serving decisions.
|
| |
