Continual Learning of Recurrent Neural Networks by Locally Aligning Distributed Representations

Computer Science

• Recurrent Neural Network 100%
• Network Coding 66%
• Activation Function 66%
• Learning Algorithm 33%
• Echo State Network 33%
• Parallelization 33%
• Training Process 33%
• Language Modeling 33%
• Backtracking 33%
• Synaptic Weight 33%
• Recurrent Network 33%
• Speech Processing 33%

Keyphrases

• Sequence Modeling 50%
• Zero-shot Adaptation 50%
• Task Sequencing 25%
• Representation Alignment 25%
• Generative Knowledge 25%
• Real-time Recurrent Learning 25%