Loading timeline…
19821980s
Recurrent Neural Networks
Neural networks with feedback loops, allowing memory over time.
Why It Was Important
Unlike feedforward networks (like standard CNNs), RNNs contained cyclical connections, allowing the network to maintain a 'hidden state' of previous inputs. This made them uniquely suited for analyzing sequential data like time series, speech, and text.
Who Invented It
John Hopfield (1982) / Elman (1990)
Physicists and cognitive scientists studying time-dependent dynamics.
Applications
- Natural Language Processing
- Speech Recognition
- Time Series Prediction
Key Papers
- Finding Structure in Time
Jeffrey L. Elman · Cognitive Science · 1990
Videos
The Power of Recurrent Neural Networks (RNN)
IBM Technology
Recurrent Neural Networks (RNNs), Clearly Explained!!!
StatQuest with Josh Starmer