Search for contacts, projects,
courses and publications

LSTM - Long Short Term Memory

People

 

Schmidhuber J.

(Responsible)

External participants

Beringer Nicole

(Third-party beneficiary)

Eck Douglas

(Third-party beneficiary)

Graves Alex

(Third-party beneficiary)

Abstract

Most work in machine learning focuses on machines with reactive behavior. Recurrent neural networks or RNNs, however, are more general sequence processors inspired by human brains. They have adaptive feedback connections and are in principle as powerful as any computer. Until recently, however, RNNs could not learn to look far back into the past. But our novel RNN called "Long Short-Term Memory" (LSTM) overcomes the fundamental problems of traditional RNNs, and efficiently learns to solve many previously unlearnable tasks, including: Recognition of certain context sensitive languages; Reinforcement learning in partially observable environments; Metalearning of fast online learning algorithms; Music composition.

Additional information

Acronym
LSTM
Start date
01.10.1997
End date
01.09.2000
Duration
35 Months
Funding sources
SNSF
Status
Ended
Category
Swiss National Science Foundation / Project Funding / Division II - Mathematics, Natural and Engineering Sciences