Node Memory
A per-node indexed vector in a temporal graph neural network, used to summarize a node's historical interaction information up to the current moment and serving as a compressed summary for model prediction.
Timeline (last 90 days)
In TGNNs, each event updates node memories, and subsequent events depend on the memories updated by preceding events, making training difficult to parallelize.
RAFT uses the magnitude of node memory change to judge the importance of an event: a small memory change means the event has limited impact on subsequent predictions and can serve as a basis for redundancy filtering.
The study found that prediction targets involving stable node memories tend to exhibit lower prediction loss.
Node memory is one of the core components of TGNNs, usually a vector indexed by node, used to summarize the historical interaction information of that node up to the current moment.
All Facts (4)
In TGNNs, each event updates node memories, and subsequent events depend on the memories updated by preceding events, making training difficult to parallelize.
50%UnverifiedRAFT uses the magnitude of node memory change to judge the importance of an event: a small memory change means the event has limited impact on subsequent predictions and can serve as a basis for redundancy filtering.
50%UnverifiedThe study found that prediction targets involving stable node memories tend to exhibit lower prediction loss.
50%UnverifiedNode memory is one of the core components of TGNNs, usually a vector indexed by node, used to summarize the historical interaction information of that node up to the current moment.
50%