Table 7 Ablation Study: Contribution of Components (Reward Score).

From: A hybrid reinforcement learning and knowledge graph framework for financial risk optimization in healthcare systems

Model Variant

Avg. Reward

Full Model (All Components)

57.19

Without Graph Embedding

49.12

Without VAE Latent Features

45.67

Without Categorical Embeddings

42.30

Flat State Vector Only

38.12

Back to article page