A position-wise feedforward network processes each zone’s enriched representation independently. In a Transformer encoder layer, it follows multi-head self-attention; both sub-layers use residual connections and layer normalization. section 9.6
Position-wise feedforward network
A feedforward neural subnetwork applied independently to each position or entity representation.
How Applied Reinforcement Learning covers it
- How deep it goes
- Defines its position-wise role but does not detail the network’s internal computation.
- Examples
- Uses cooling-floor zones as entities whose representations are processed independently.
- What you can do after
- Understand how the encoder processes each zone’s representation after adding relational context.
- How it connects
- Connects with multi-head self-attention, residual connections, and layer normalization within the encoder layer.
Related concepts
- fully connected neural networkuses · A fully connected neural network is a neural network whose units in each layer connect to every unit in the adjacent layer.
- feedforward networkkind of · A feedforward network is a neural network whose computations proceed from inputs toward outputs through directed connections without recurrent cycles or feedback.
This explainer was generated by Manning's knowledge graph from the text of Applied Reinforcement Learning. The section links point to the parts of the book it draws on.
