Statements (50)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:artificial_intelligence
|
| gptkbp:advantage |
long-term dependency learning
|
| gptkbp:citation |
Hochreiter & Schmidhuber, 1997
|
| gptkbp:component |
recurrent neural networks
|
| gptkbp:developedBy |
gptkb:Sepp_Hochreiter
gptkb:Jürgen_Schmidhuber |
| gptkbp:fullName |
Long Short-Term Memory
|
| gptkbp:hasApplication |
autonomous vehicles
robotics bioinformatics financial forecasting question answering sentiment analysis chatbots anomaly detection language modeling music generation video analysis machine comprehension |
| gptkbp:hasComponent |
cell state
forget gate input gate output gate |
| gptkbp:hasVariant |
bidirectional LSTM
convolutional LSTM peephole LSTM stacked LSTM |
| https://www.w3.org/2000/01/rdf-schema#label |
LSTMs
|
| gptkbp:implementedIn |
gptkb:TensorFlow
gptkb:Keras gptkb:PyTorch |
| gptkbp:introducedIn |
1997
|
| gptkbp:limitation |
computationally expensive
difficult to parallelize can still forget long-term information |
| gptkbp:relatedTo |
gptkb:GRU
gptkb:RNN transformers |
| gptkbp:solvedBy |
vanishing gradient problem
|
| gptkbp:usedFor |
machine translation
natural language processing speech recognition time series prediction sequence modeling |
| gptkbp:usedIn |
gptkb:Google_Translate
handwriting recognition text generation speech-to-text systems |
| gptkbp:bfsParent |
gptkb:Transformer_models
|
| gptkbp:bfsLayer |
7
|