dimension op• Data kinds: conn_graph × labels × matrix → matrix
• Call: import fullseye as fs; fs.ledger.graph_layer_propagate(W: 'Any', labels: 'Any', U: 'Any', activation: 'str' = 'kwta', active_frac: 'float' = 0.1, gain: 'float' = 1.0) -> 'np.ndarray' (to call the implementation directly, import conngraph; conngraph.graph_layer_propagate(W: 'Any', labels: 'Any', U: 'Any', activation: 'str' = 'kwta', active_frac: 'float' = 0.1, gain: 'float' = 1.0) -> 'np.ndarray'; from the registry, opsconngraph.get("graph_layer_propagate"))
> This operator's description has not been translated yet. The original text follows as it is.
層 0 の状態 `U (N, n_0) を、ブロック W[layer a → layer a+1]` で前向きに一段ずつ通した全層の状態
`(N, n)(matrix、列はノード順で層 0 の列は U` そのもの)。
受け手ごとに入力重みの和を 1 に正規化してから重みつき和を取り(層の大きさに依らない)、`activation` で
活性化する: `"linear" はそのまま(平均絶対値を gain に)、"tanh" は飽和、"kwta"` は各刺激で
入力が上位 `active_frac` の受け手だけが(閾値上の余剰で)発火し、平均が 1 になるよう正規化する。
層をまたぐ結線(層 a → a+2)と層内の再帰は使わない(前向きの一段ごとの写像だけを見る道具)。
• Sample-data catalog (download URLs / licences) — 2-D uses skimage.data (BSD/public domain) plus synthetic images; 3-D lists download URLs for real data sources (Stanford, PDS, …).
• Operator provenance and references — the sources of the research/methods this op family came from.
• The canonical algorithm (author, year) and its uses are named in the family usage guide above.
• poc_connectome_motor_bottleneck — py -3.11 examples/poc_connectome_motor_bottleneck.py
matrix as input)reservoir_states · reservoir_encode · ridge_readout · ridge_predict · graph_activation_latency · graph_activity_spread · points_activity_video · states_participation_ratio
dimension)graph_block_shuffle · states_participation_ratio · states_layer_dimension
*Provenance: conngraph.py — CONNGRAPH operator registry. This per-op note is generated by tools/opdocs.py md (do not hand-edit).*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.