VflAggregationMode
Enums · AiDotNet.Models.Options
Specifies how embeddings from multiple parties are combined in vertical federated learning.
For Beginners
In vertical FL, each party computes a local embedding (a compressed representation of its features). These embeddings must be combined before the top model can make a prediction. This enum controls how that combination happens.
Fields
| Field | Summary |
|---|---|
VflAggregationMode Attention |
Attention-weighted combination of embeddings. |
VflAggregationMode Concatenation |
Concatenate embeddings side by side. |
VflAggregationMode Gating |
Gating mechanism that learns a sigmoid gate to blend embeddings. |
VflAggregationMode Sum |
Element-wise sum of embeddings. |
int value__ |