Il prossimoProssima guida
Come trasformare video lunghi in cortometraggi con l'intelligenza artificiale
IA visiva
GUIDA AI FONDAMENTALI
Long Short-Term Memory (LSTM) cells are recurrent units designed to help preserve information and error flow across long sequences.
Ascolta questa guidaNarrato dall'IA · ~3 min
Their gated cell state can mitigate difficulties that vanilla RNNs face with long-range dependencies, but LSTMs do not eliminate every vanishing-gradient condition or guarantee perfect long-term retention.
Introduced by Sepp Hochreiter and Jürgen Schmidhuber in 1997, LSTM was designed to help recurrent networks learn long time dependencies despite gradient-flow difficulties. The original 1997 architecture used memory cells with input and output gates and a constant error carousel; it did not include the now familiar forget gate. Gers, Schmidhuber, and Cummins introduced an adaptive forget gate in 2000 for continual prediction, allowing a cell to reset stored state when appropriate. In common modern three-gate variants, forget, input, and output gates regulate retention, writing, and exposure. The gates commonly use sigmoid activations, which produce values between zero and one. A typical cell update is c_t = f_t * c_(t-1) + i_t * g_t. The additive cell-state path can make gradient flow more favorable than repeatedly multiplying through ordinary recurrent transformations, but it does not guarantee gradients never shrink or that every dependency hundreds of steps away will be learned. Performance depends on gate behavior, model configuration, data, and optimization. At each time step, the recurrent unit also computes a hidden state from the exposed cell content; downstream layers use this representation for prediction. Gate activations are learned from task data, so a model may choose to retain, update, or expose different information at different positions. This controlled path supports memory, but actual long-range behavior must be measured on the sequence lengths and task of interest. LSTMs have been used in sequence tasks such as speech recognition, translation, and time-series prediction, though the details of each deployed system vary.
Alimentare la traduzione automatica all'inizio del sistema neurale di Google Translate prima che Transformers prendesse il sopravvento.
Riconoscimento vocale in assistenti vocali e software di dettatura.
Previsione dei valori futuri in serie temporali come la domanda di energia, le letture dei sensori o i prezzi delle azioni.
Generazione di testo o musica un token alla volta e completamento automatico delle sequenze.
Transformers have largely overtaken LSTMs for large-scale language tasks because they parallelize across a sequence and capture long-range context via attention, whereas LSTMs process tokens one step at a time. Still, LSTMs remain valuable for streaming, low-latency, and resource-constrained settings, and on modest time-series data. Recent work like xLSTM (2024) revisits and modernizes the architecture with new gating and memory to compete at scale, showing the idea is not finished. Comparisons should report actual latency, memory use, task and sequence lengths rather than assuming one architecture wins everywhere.
Newsletter gratuita
Tre storie verificate sull'IA ogni mattina nei giorni feriali, scritte in un inglese semplice. Gratuito per sempre, senza pubblicità.
Un'email ogni giorno feriale. Disiscriviti con un clic. Non vendiamo né condividiamo mai il tuo indirizzo.
Mettiti alla prova
Feedback immediato su ogni risposta e un certificato condivisibile con un ID verificabile una volta superato un corso.
Sostieni l'istruzione gratuita sull'IA. AI Understanding è un'organizzazione non profit 501(c)(3) — nessun annuncio, nessun paywall, mai. Fai una donazione
Long Short-Term Memory (LSTM) cells are recurrent units designed to help preserve information and error flow across long sequences. Their gated cell state can mitigate difficulties that vanilla RNNs face with long-range dependencies, but LSTMs do not eliminate every vanishing-gradient condition or guarantee perfect long-term retention.
Common modern three-gate variants use forget, input, and output gates; the original 1997 formulation preceded the forget gate, which Gers and colleagues introduced in 2000.
The original LSTM was designed to improve gradient flow and address difficulty learning long time dependencies; it does not eliminate every vanishing-gradient condition.
Sepp Hochreiter e Jurgen Schmidhuber hanno pubblicato l'LSTM nel 1997.
The additive route can help preserve error signals when gate values allow it, but does not guarantee that gradients never shrink.
I gate utilizzano un sigmoide, il cui output da 0 a 1 scala la quantità di informazioni che passano, agendo come un interruttore morbido.
Continua a imparare
Altre guide selezionate per questo argomento
Il prossimoProssima guida
Come trasformare video lunghi in cortometraggi con l'intelligenza artificiale
IA visiva