To represent features of relation candidates, Fonduer uses a bidirectional Long Short-term Memory (LSTM) with attention. Relying on LSTM, along with weak supervision, obviates the need to create large sets of training data by hand, an important consideration since it is difficult to build proper training data at a large scale.
If you want to change selection, open document below and click on "Move attachment"
pdfcannot see any pdfs
|status||not read|| ||reprioritisations|
|last reprioritisation on|| ||suggested re-reading day|
|started reading on|| ||finished reading on|